Language - PDF Free Download (2024)

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 1 Learning to read a dialect ErfMR evidence from Südtirol Jubin Abutalebi 1,3 , Roland Keim 4 , Simona M Brambati 1 , Marco Tettamanti 1 , Massimo Danna 1 , Stefano F. Cappa 1,2 , Ria De Bleser 3 , Daniela Perani 1,2 1 Vita-Salute San Raffaele University and Scientific Institute, Milan, Italy, 2 IBFM-CNR, Milan, Italy, 3 University of Potsdam, Potsdam, Germany, 4 Psychologischer Dienst, Sanitätsbetrieb Brixen, Südtirol Reading is an ability acquired through explicit schooling, and many factors, such as phonological processing, contribute to its development, while dialects are acquired implicitly and informally, without any orthographic counterpart. Dual route models of reading postulate a direct lexical and an indirect orthography-to-phonology pathway. The latter is believed to be responsible for reading pseudo-words, which lack of an orthographic lexicon. The aim of this study was to investigate word-reading using a language that has no orthographic lexicon, but, contrary to pseudo-words, has an existing semantic system. We performed two studies in right-handed subjects speakers of Südtirol dialect: behavioral , in a group of 18 adults, and fMRI in 12 adults. In the behavioral study, voice onset times (VOTs) were recorded during reading of German words, dialect words and pseudo-words. In the imaging study subjects where scanned with the event-related technique fMRI during identical conditions. Functional imaging data were analyzed with SPM2 and direct comparisons between the conditions were performed on a second level analysis (random effects). In both studies subjects were exposed to the stimuli several times (six trials in the behavioral study and four in fMRI) to investigate learning and practice effects. The behavioral results showed that VOTs in reading dialect words were in-between German words and pseudo-words (German 540 msec; dialect 612 msec; pseudo-words 681 msec). VOTs significantly decreased with repeated exposure, with an effect maximal for pseudowords (558 msec), followed by dialect(532 msec), and German words (517 msec). Reading performance is thus most influenced by repeated exposure to the stimuli when no orthographic lexicon is available. In the case of South Tyrol dialect, lacking an orthographic lexicon, the existing phonological lexicon and semantic system may have similar influences. All three reading conditions elicited brain activation patterns classically associated with reading. However, both dialect and pseudo-word when compared to German engaged more extensively areas associated with phonological processing (i.e., Brocas area, left frontal operculum, left supramarginal gyrus). Only reading of the dialect words activated the left hippocampal system and the left caudate nucleus (see figure1). The involvement of the left hippocampal system was present only during the initial exposure to dialect words (see figure 2). These findings may reflect the role of learning and consolidation mechanism for new written lexical entries. The lack of a comparable activation in the case of pseudowords suggests that the availability of corresponding semantic knowledge is a crucial factor for the engagement of learning-related brain areas. The present findings provide for the first time evidence that written material lacking orthographic lexical representations (pseudowords and dialect words) differ in the pattern of brain activation according to the availability of corresponding semantic representations. This neural correlate reflects the learnability effect evidenced by behavioral testing.

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Brain activity patterns elicited by reading dialect words as compared to German words (top row) and pseudowords as compared to German words (bottom row).

Brain activity associated resulting by comparing the first to the last session of the dialect word reading task.

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 2 Neural correlates of orthographic processing Markus Aichhorn 1 , Martin Kronbichler 1,2 , Heinz Wimmer 1 , Florian Hutzler 3 , Alois Mair 2 , Wolfgang Staffen 2 , Gunther Ladurner 2 1 Department of Psychology, University of Salzburg, 2 Christian Doppler Clinic, PMU, Salzburg, 3 Bereich Allgemeine Psychologie, Freie Universität Berlin Introduction The present study explored which brain regions are involved in accessing and activating visual-orthographic representations in response to spoken words. Information on brain regions involved in accessing and activating orthographic information is of importance for an understanding of spelling problems. With only some exceptions (e.g. 1) the question of neural correlates of accessing orthography from phonology has received little attention. Method 18 normal volunteers participated in this study. T2*-weighted functional images were obtained during two runs while performing 6 blocks for each of three tasks. (1) In the orthographic task subjects had to decide for each of 24 auditory presented words whether it is written with three or four letters. All the items were one-syllable words consisting of three phonemes so that orthographic judgment could not rely on phonology. (2) In the semantic task they had to determine for each of 24 auditory words whether it referred to living or non - living object. (3) During a similar auditory control task, participants had to decide whether a single tone was high or low. Altogether 48 auditory items - one half spelled with three, the other with four letters - were used. Within each length category half of the items denoted living entities. Whether an item was presented for orthographic or semantic judgment was counterbalanced over subjects. All images were realigned, normalised, smoothed (FWHM = 9mm) and analysed using SPM2 (www.fil.ion.ucl.ac.uk/spm). With a random effects analysis we first computed the following contrasts: semantic decisions > control task and orthographic decisions > control task. Regions showing higher activity in at least one of these contrasts, were then used for evaluation of the orthographic > semantic contrast. All comparisons were thresholded at p < .001, minimum extent = 5 voxels. Results As can be seen from Figure 1, higher BOLD activity for orthographic than semantic decisions was limited to regions of the left hemisphere including supramarginal gyrus, occipito-temporal cortex, the superior parietal cortex and putamen. Discussion The present finding of orthographic specific activation in left occipito-temporal cortex and supramarginal gyrus corresponds to a recent study (1) contrasting orthographic and phonological decisions with auditory words, in contrast to these study we did not find orthographic activation in left and right frontal regions. This finding suggests that, in response to our orthographic task, auditory information had to be used to access and activate visual-orthographic information located in regions of the left occipito-temporal cortex, near to the so called visual word form area (2). The supramarginal gyrus may, in this context, serve a area for cross-modal conversion (1), through which auditory and visual word information, located in different brain areas, are integrated. 1.Booth et al., NeuroImage, 16, 7-22. 2.McCandliss et al., Trends Cogn Sci, 2003, 7, 293-99

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Higher BOLD activity for orthographic versus semantic decisions

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 3 Brain activities associated with repetition priming effects: An event-related fMRI study Yuko Akitsuki 1,2 , Yuko Sassa 2,3 , Masaki Nakamura 1 , Shuichi Awata 1 , Hiroo Matsuoka 1 , Ryuta Kawashima 2 1 Department of Psychiatry, Tohoku University Graduate School of Medicine, Sendai, Japan, 2 NICHe, Tohoku University, Sendai, Japan, 3 LBC Research Center, Tohoku Uniaversity 21st Century Center of Excellence Program in Humanities, Sendai, Japan Introduction The amplitude of N400 component of event-related brain potential (ERP) has been used as an index of semantic and repetition priming. According to a recent report (Matsumoto et al., Clinical Neurophysiology 112: 662-673, 2001), compared to normal groups, patients with schizophrenia showed attenuation of the N400 peak occurred for immediate word repetition. And the authors suggest that this deficit might be related to an abnormal N400 priming effect in schizophrenia. However, brain areas associated with this repetition priming effect itself are studied less extensively. In this experiment, an event-related fMRI technique was used to identify regions of the brain whose activity is associated with the repetition priming effects. Methods Fourteen healthy right-handed volunteers (12 men and 2 women, mean age = 21.9 years, SD = 3.42) participated in this study. All of the subjects were native speakers of Japanese. There were 242 trials including two dummy trials. The stimuli used here were 132 meaningful words composed of two kana letters (Japanese phonographs); 24 were targets belonging to a group of animal names and the other 108 were non-targets. 54 non-targets were repeated immediately after their first presentation. Each word was presented for 200 ms on a screen on the head coil with random interstimulus intervals of 2.4 - 2.8 s. In 54 trials, only a fixation point was presented as a baseline condition. The subjects were asked to press a button with the right index finger as quickly and accurately as possible when they recognized animal names. Whole-brain gradient-echo echo-planar fMRI scans were obtained using a Siemens 1.5T Simphony scanner (TR = 2000 ms, FOV = 192 mm, matrix 64x64, 20 axial slices, slice thickness = 6 mm, 312 scans). All image data preprocessing and statistical analyses were performed using SPM2. The model function was tailored for each of the following four conditions for each subject: (1)initial presentation, (2) immediately repeated presentation (3) target presentation, and (4) baseline. Results Compared to the baseline, significant brain activations were found in the left prefrontal cortex including Brodmann’s area 45 and 47 in the initial presentation, but not in the immediately repeated presentation. Significant brain activations in the left posterior inferior prefrontal cortex (including Broca’s area), the bilateral fusiform cortex, and the bilateral lateral occipital cortex, and primary visual cortex were also revealed by the subtraction analyses of "initial presentation - immediately repeated presentation". (P < 0.05, corrected). However, no significant activation was found by the subtraction analysis of "immediately repeated presentation - initial presentation". Discussion In our data, relative to initial presentation, repetition priming is associated with reductions in neural activity in multiple regions. This finding is consistent with the results of other imaging studies (Simons et al., 2003: Wiggs et al., 1998). Our data also indicate that the left inferior prefrontal cortex (including Brodmann’s area 47) plays an important role in the processing of immediate repetition effects on visual word presentation. This finding is consistent with the recent studies which reported the strong involvement of the left inferior prefrontal area in lexico-semantic process (Vigliocco, Current Biology 200, 10: R78-80).

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 4 Covert speech arrest induced by rTMS Lisa Aziz-Zadeh , Luigi Cattaneo , Magali Rochat Dipartimento di Neuroscienze, Università di Parma Introduction Stewart et al (2001) reported two different types of overt speech arrests (SA) induced by repetitive transcranial magnetic stimulation (rTMS). The first one, associated with EMG activity in lower facial muscles through activation of the corticobulbar pathway (motor SA) and a second one, not associated with EMG activity in facial muscles (non-motor SA), which is thought to be mediated by stimulation of Brocas area, over a frontotemporal region of the scalp anterior to the spot of the motor SA. Will stimulation of one or both of these areas induce a covert (internal) speech arrest as it does for overt speech? The current study explored this question. Methods SA was evoked with short trains of magnetic stimuli at 5 Hz. Polysyllabic words were presented briefly on a computer screen. Participants were required to report the number of syllables in each word as quickly and as accurately as possible via a manual response box. In half of the trials the subject performed the task internally covertly, while in the other half they were asked to perform it speaking aloud. Words were chosen in order to avoid visual recognition of the number of syllables. Right opponens pollicis (OP) and mentalis (M) muscles hot spots were localised on the scalp over the left hemisphere and their respective resting motor thresholds (rMTh) were assessed. Then the optimal stimulation intensity capable of evoking a motor SA over the M muscle was measured. Subjects who displayed EMG responses with short latencies, compatible with a peripheral activation of the facial nerve were discarded. Subsequently a non-motor SA was looked for, applying rTMS anteriorly to the M muscle hot spot. Stimulation intensity was increased until a SA was evoked. For both conditions (overt and covert speech) subjects completed the task in four blocks. The first block was a baseline condition, in the other counterbalanced blocks TMS was applied coincident with word presentation, to the motor SA spot, to the non-motor SA spot and to the hom*ologous point of the non-motor SA on the right side. Data were analysed using a two-way ANOVA with repeated measures [Stimulation Site (motor SA, non-motor SA, right hemisphere), Language Type (covert, overt)] using response latency as the dependent measure. Results All participants reported a counting strategy that involved mental rehearsal of the word. None of the subjects reported rTMS as being more than mildly painful. All subjects showed significantly longer latencies when stimulation was either over the motor SA hot spot or the non-motor SA hot spot, as compared with the right hemisphere hot spot. This result was observed in all participants for the overt speech task and for the covert speech task alike. Discussion Our data indicate that overt and covert speech arrests induced by TMS follow similar patterns. Thus it appears that the posterior/motor site is intrinsically involved in covert speech as the anterior/Brocas site is. This finding may have implications for language evolution and mentalese/the language of thought.

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 5 Functional MRI assessment of modality-related and age-related effects during passive presentation of written and spoken words Monica V. Baciu 1 , Rebecca S. Coalson 2,3 , Steven E. Petersen 2,3,4,5 , Bradley L. Schlaggar 2,3,4,6 Mendes-France University, Grenoble, France; Department of Psychology, 2 Washington University School of Medicine, St. Louis, MO, USA; Department of Neurology, 3 Washington University School of Medicine, St. Louis, MO, USA; Department of Radiology, 4 Washington University School of Medicine, St. Louis, MO, USA; Department of Anatomy and Neurobiology, 5 Washington University School of Medicine, St. Louis, MO, USA; Department of Psychology, 6 Washington University School of Medicine, St. Louis, MO, USA; Department of Pediatrics 1 Pierre

Introduction. Past studies have investigated the developmental functional neuroanatomy of simple and controlled lexical processing tasks requiring active responses. Here, our goal was to investigate simple passive lexical tasks, including silent reading of, and passive listening to, single words. We evaluated the effect on cerebral activity of 1) modality (visual or auditory) and 2) development (adults vs 7-10 year old children). Methods. Subjects. Thirty-one children (mean age 9.1 y, 21 female) and seventeen adults (mean age 27.0 y, 10 female) were examined. Tasks. Subjects performed one fMRI run each of passive reading of printed words (without reading aloud) and passive listening to spoken words (without repeating). Each 210 sec. run presented 21 words in an eventrelated design with the stimulus onsets jittered by 6, 9 or 12 sec. MR acquisition. A Siemens MAGNETOM Vision 1.5 Tesla scanner (Erlangen, Germany) using a standard EPI sequence gathered 73 whole brain acquisitions per run. Data processing. In-house software (1) did pre-processing and statistical analysis. A voxel-wise ANOVA used correlations to stimuli (GLM) from all subjects to generate Z images, a region-finding algorithm found coordinates of the peak-voxel from these images, and a random effects ANOVA generated time courses and statistics for each set of regions. Results Most importantly, we identified regions activated by passive language, regardless of modality (Table), many of which (e.g. inferior frontal and superior temporal regions) are commonly found in more active lexical processing tasks. Interestingly, no age-related effect met our statistical threshold for these simple lexical tasks (Monte Carlo corrected Z>3.5, at least 24 contiguous voxels). Also, as expected, we found visual regions selectively activated during passive reading and auditory regions for passive listening to spoken words. Conclusions. Results suggest that passive reading of printed words and passive listening to spoken words (1) activate an expected constellation of regions involved in word recognition, (2) involve both modality-specific and modality-independent regions, and (3) are supported by a common functional anatomy in adults and children, suggesting that visual input automatically activates a lexical processing network even in early readers. Reference. (1) Miezin et al. (2000), Neuroimage, 11, 735-59.

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MODALITY INDEPENDENT REGIONS x

anterior and posterior insula

Left hemisphere inferior frontal (BA 44)

-34

-47

-46

-52

-36

-56

-37

-20

-12

-53

-48

-49

-30

-44

superior temporal (BA 22)

-43

-39

anterior cingulate (BA 24)

supramarginal (BA 40)

-49

-41

inferior frontal (BA 45, 46, 47)

anterior cingulate (BA 24)

-1

-21

-4

posterior cingulate (BA 31)

-33

superior parietal lobule (BA 5)

-20

-41

middle frontal (BA 46)

parahippocampal

-20

-21

-10

precuneus (BA 31)

-53

superior temporal (BA 22)

middle frontal (BA 10, 46)

lentiform nucleus

Right hemisphere

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 6 Oscillatory Brain Dynamics during language processing: theta band responses in a lexical decision task. Marcel Bastiaansen , Peter Hagoort FC Donders Centre for Cognitive Neuroimaging, Nijmegen, the Netherlands Introduction Oscillatory brain dynamics are thought to play a role in transiently binding functionally related brain areas [1]. In line with this notion, previous work has shown that theta and gamma oscillations show reliable power changes during online sentence processing. Theta power increases during the course of a sentence over bilateral temporal and midfrontal areas [2], upon the occurrence of syntactic and semantic violations in sentences over midfrontal areas [3], and after presentation of open-class as compared to closed-class words over left temporal and left occipital areas [5]. Gamma power increases over right frontal areas during the processing of correct sentences, but this effect is abolished when a semantic violation is encountered. Here we investigate the reactivity of theta and gamma band oscillations in a lexical decision task, in order to compare the oscillatory brain dynamics during language processing at the word level with those previously observed at the sentence level. Methods 16 subjects performed a lexical decision task while their EEG was recorded. Stimuli were legal Dutch words, pronounceable non-words, consonant strings, pictures of objects or nonsense (scrambled) pictures. For the picture stimuli, subjects had to indicate whether or not the picture represented a real object. All stimuli were matched on the relevant dimensions. The EEG data (61 electrodes, bandpass 0.3-70 Hz, sampling rate 500 Hz) were subjected to a time-frequency analysis using a multitaper technique [5]. Power changes from 1 to 60 Hz were expressed as a percentage in- or decrease relative to a 150 ms prestimulus baseline. Results All stimuli induced power increases in the 4-7 Hz band over left and right occipital electrodes, between 100 and 500 ms post-onset. Words and pseudowords elicited comparable power increases, while consonant strings induced a larger response over the right hemisphere (Figure 1). Overall, picture stimuli elicited larger theta responses than the letter stimuli over the entire scalp (Figure 2). A direct comparison between real pictures and the concrete words corresponding to these pictures (Figure 2) shows that differences are most prominent over right occipital scalp. None of the stimuli elicited significant power changes in the gamma frequency range. Discussion Theta responses increased in the following order: from words to consonant strings, to real pictures, to nonsense pictures. This pattern seems to indicate that less automatized processing corresponds to more theta power. This fits our previous sentence processing results [3,4], where syntactic and semantic violations elicit larger theta responses. The theta increases reported here are therefore hypothesized to be parametrically related to retrieval effort [6]. The lack of right frontal gamma increases, and of theta effects over left temporal and midfrontal areas contrasts with the previous sentence-level results[4], and suggests that these effects are more specifically related to sentence-level language processing. References [1] Varela et al. Nat Rev Neurosci 2, 229-39. (2001). [2] Bastiaansen et al. Neurosci Lett 323, 13-6 (2002). [3] Bastiaansen et al., Neuroimage 17, 1479-92 (2002). [4] Bastiaansen et al., submitted (2003). [5] Mitra et al. J. Biophys 76:691-708(1999). [6] Klimesch, Brain Research Reviews 29, 169-195 (1999).

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e10

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 7 Neural network of affective sentences processing: effect of affective content, gender, prosody. Virginie Beaucousin 1 , Anne Lacheret-Dujour 4 , Marie-Renée Turbelin 2 , Michel Morel 4 , Marc Joliot 1 , Guy Perchey 2 , Bernard Mazoyer 1,2,3 , Nathalie Tzourio-Mazoyer 1 1 GIN, UMR 6194, CNRS/CEA/Univ. Caen & Paris 5, France, 2 IRM CHU Caen, France, 3 Institut Universitaire de France, 4 CRISCO, UMR 6170, CNRS/Univ. Caen, France. Introduction While grammatical prosody tightly linked with discourse syntactico-semantic component defines sentences syntactic structure, affective prosody relies on emotional processing, a right hemisphere function. However the picture is more complex: although neuropsychological studies evidenced troubles in affective prosody comprehension after right temporal lesions 1 , recent fMRI studies showed the implication of a bilateral fronto-temporal network 2 during emotional prosody processing. The aim of the present study was to design an approach allowing the dissociation of these different components. Materials and methods 28 right-handed subjects (14 men) were submitted to BOLD acquisitions with a block design alternating 34 sec duration periods of language tasks or cross fixation. The language tasks consisted in listening to series of 45 sentences, and classify them in three categories. First task was to determine the sentences subject of neutral sentences (first, second or third person). In the second, subjects had to classify the emotion promoted by each sentence (anger, joy, sadness). In the third, subjects had to judge if the sentences contained irony, doubt or obviousness. These tasks were performed twice varying the mode of sentences enunciation: 1- synthesized by software including grammatical prosody but devoid of affective prosody (Kali®) 3 ; 2- enounced by actors. Statistical analysis was performed with SPM99 and investigated the networks showing a main effect of gender, voice (actor, Kali®), affective content (affective or neutral prosody) and their interactions. Results Main effect of gender A larger signal increase was present in men bilaterally in the superior part of the temporal lobe, the parietal lobe, and in the left inferior frontal gyrus (IFG). The anterior cingulate was recruited by men only. The left IFG and the superior parietal gyrus showed significantly larger activations in women. Main effect of affective discourse Listening to affective discourse recruited a large network of language regions in the left hemisphere and their right hom*ologous: superior temporal sulcus (STS), IFG, median F1 and cerebellar cortex. Interaction between voice and affective discourse: areas dedicated to the processing of human voice during affective prosodic speech listening. Larger activations were present in the right STS (coordinates 54, -2, -12; z = 4.81) when actors enounced the affective discourse rather than when it was synthesized by Kali®. Discussion Processing of prosody can be modelled in the following way: The right STS is in charge of the processing of the human voice when it contains affective prosody, an evidence in line with the neuropsychological findings 1 , suggesting that this area could be an epicenter for affective prosody perception; The left hemisphere semantic network and its right hom*ologous together with the medial frontal wall involved in theory of mind processing 4 integrates both syntactico-semantic and emotional components to lead to the interpretation of affective sentences whatever the presence of emotional prosody. The effect of gender was independent of the type of task or voice, related to different classification strategies: men that activated the anterior cingulate dedicated to error detection 5 , performed better. 1. Ross ED, 1981. 2. Kotz SA, 2003. 3. Morel M, 2001. 4. Ferstl EC, 2002. 5. Badgaiyan RD, 1998.

e11

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e12

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 8 Cerebral hemodynamics during listening to emotional prosody of an unfamiliar language studied by functional transcranial Doppler ultrasonography. 1 Laboratory

Celine Berckmoes 1 , Elio Troisi 2 , Mariella Mattheis 2 , Guy Vingerhoets 1 for Neuropsychology, B-9000 Ghent, Belgium, 2 Santa Lucia Foundation ICRRS 00179 Roma, Italy

Simultaneous measurement of blood flow velocity (BFV) in the middle cerebral arteries was achieved by fTCD in 27 right-handed Flemish and an equal number of Italian volunteers who were unfamiliar with the Dutch language. They were instructed to categorize the emotion conveyed by the prosody in a number of Dutch sentences earlier described in Vingerhoets (2003). A between subjects MANOVA showed a significant effect for lateralization, reflecting the common prediction of the right-hemispheric increase of BFV when categorizing emotional prosody. But compared to the Italians, the Flemish volunteers exhibit a significant left hemispheric involvement, presenting the undeniable influence of the verbal content of the sentences. References: Vingerhoets, G. Berckmoes, C., Stroobant, N. Cerebral Hemodynamics during Discrimination of Prosodic and Semantic Emotion in Speech Studied by Transcranial Doppler Ultrasonography. Neuropsychology, 2003, 17 (1), 93-99

e13

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 9 Modality effects in Verbal Working Memory: and fMRI study Bradley R Buchsbaum , Rosanna K Olsen , Paul Koch , Karen Berman Unit for Integrative Neuroimaging, Clinical Brain Disorders Branch, National Institutes of Health Bradley R. Buchsbaum, Rosanna Olsen, Paul Koch, Karen Berman Introduction It has long been known that performance on short-term memory tasks is better for auditory presentation than for visual presentation. It has been hypothesized that this auditory advantage is due to the relative persistence of acoustic information in the sensory store of the auditory system. Utilization of this lingering auditory trace to replay sensory events has been termed echoic memory. Much work over the last decade has used brain imaging to probe the neurobiological basis of the phonological loop and its subcomponents. Much of this work has focused on the dorsolateral prefrontal (BA 46) and posterior parietal (BA 40) cortices areas that are not thought to play a role in sensory memory storage. This study uses event-related fMRI to investigate the activation dynamics in a verbal working memory task with simulataneous presentation of visual and auditory stimuli. Methods Thirteen normal right-handed subjects (eight men, five women) volunteered to participate in the study. Their ages ranged from 23 to 38, and they all had normal vision and normal hearing. There were eight experimental runs, each 340s in duration. Each run was organized as a sequence of 12 trials where each trial consisted of several phases: (1) stimulus presentation followed by (2) instruction cue and (3) a 12-s delay period and then, in two of three conditions, (4) overt recall. During stimulus presentation subjects were shown sets of four or six common English language words, half presented through headphones in the auditory modality, and half presented in Times Roman typeface on a large screen placed at the foot of the MRI bed. For every word presented in the auditory modality a different word was presented in the visual modality. Immediately after the last word pair was presented, an instruction cue was shown on the screen for 750ms. If the cue was a picture of an EAR, subjects were to silently rehearse those words that they heard. If the cue was an EYE, subjects were instructed to silently rehearse those words that they saw. After a 12-s rehearsal period a picture of a mouth appeared on the screen, instructing the subject to overtly recall the set of words that he or she had heard or seen. Results A multiple regression analysis was carried out using FEAT (FMRI Expert Analysis Tool) Version 5.00, and a contrast of parameter estimates of regressors generated for the 3rd phase (rehearsal period) of the EAR and EYE condition, revealed greater activity during auditory retrieval in three major locales: dorsolateral prefrontal cortex (BA 46) bilaterally, inferior frontal cortex (BA 47) on the left, and the posterior superior temporal sulcus (BA 22/21) on the left. The reverse contrast (EYE > EAR) showed effects in the anterior cingulate and hippocampus bilaterally. Discussion We have found evidence for modality specific enhancements of delay period activity during a task that required the selective retrieval of items of the same modality from short-term memory. Specifically, evidence of heightened delay period activation in the superior temporal sulcus points to this areas central role in persistent auditory, or echoic, memory.

e14

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 10 Neural Processes Underlying the Effects of Rewards on Learning of Semantic Declarative Knowledge Daniel Callan 1 , Nicolas Schweighofer 2 1 ATR Human Information Science Laboratories, 2 ATR Computational Neuroscience Laboratories Little is known about the neural processes underlying the effects of rewards on learning of semantic declarative knowledge. Encoding of semantic declarative knowledge is thought to involve establishment of associations between the stimulus to be learned and existing knowledge via attention dependent verbal short-term working memory; the extent of processing is essential for establishment of knowledge in long-term memory. From a neurobiological perspective rewards may serve as a modulatory signal to facilitate encoding. Results of psychological experiments are unclear, some suggest encoding declarative knowledge is reward-independent, but attention is reward-dependent, and others suggest external rewards are detrimental to learning because they serve as a stressor shifting the focus of attention from the learning material to the reward. In this experiment the effects of rewards on learning of semantic declarative knowledge was investigated by a task involving acquisition of second-language vocabulary. From the results of a pretest, ten correct words were selected for the ’known’ condition, and thirty incorrect words were divided into three separate monetary reward conditions: 300; 100; 0 yen. The event-related fMRI experiment entailed encoding of various English-Japanese word pairs presented visually (ISI=7 sec), and enclosed in a color frame denoting the potential reward given correct post-test response (incorrect responses=-100 yen). Imaging results for the unknown-known contrast denoted activity in brain regions involved with verbal working memory and encoding into LTM (Fig. 1a). The known-unknown contrast showed activity in regions thought to be involved with semantic processing (Fig. 1b). Behavioral results revealed no effect on performance depending on reward condition (300=63%; 100=59%; 0=66%). One might conclude from these behavioral results that rewards do not have an effect on encoding. However, the regression of brain activity related to the 300-0 contrast with behavioral performance for the 300 reward condition revealed brain regions involved with LTM encoding and reward modulation (Fig. 2). Lack of activity in PFC and presence of activity in neuromodulatory sites, suggests that rewards are acting to modulate learning in hippocampus directly, rather than through modulation of attention in PFC. The regression of brain activity for the 300-0 contrast with qualitative ratings of stress (1-6 point scale) for the 300 condition revealed brain regions involved with anxiety and right hemisphere PFC regions involved with attention to the reward cue (Fig. 3). The regression of brain activity for the 0-300 contrast with behavioral performance for the 0 condition revealed brain regions involved with semantic associative processing (Fig. 4). The results suggest that rewards have a modulatory effect on encoding but that this is counterbalanced by anxiety which divides attention to the reward resulting in less efficient associative semantic processing and thus no difference in performance for rewarded and un-rewarded conditions. These finding are important because they provide insight into underlying processes involved with effects of rewards on learning declarative knowledge that cannot be gleaned by psychological experiments that do not take aspects of the neurobiology into account.

e15

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1a. Unknown-known: Brain activity present in regions involved with verbal working memory (Broca’s area, DLPFC, SMA, SMG) areas involved with LTM encoding (parahippocampal gyrus, brainstem) as well as posterior parietal, occipital, and fusiform areas. b. known-unknown: Brain activity present in regions involved with semantic association memory (SMG, AG, STG, MTG, PFC, temporal pole) as well as MFG, insula, ocipital areas, and ACC).

Figure 2. Regression: 300-0 with 300 Behavioral Performance: Brain activity present in regions involved with reward modulation (VTA/SN, ACC, amygdala), LTM encoding (hippocampus), as well as MTG, ITG, fusiform gyrus, and cerebellum.

e16

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 3. Regression: 300-0 with qualitative stress rating: Brain activity present in regions involved with anxiety (ACC, medial thalamus), as well as PCC, left SMG, right PFC and temporal pole.

Figure 4. Regression: 0-300 with 0 behavioral performance : Brain activity present in regions involved with semantic associative processing (SMG), as well as right Broca’s and fusiform gyrus.

e17

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 11 Neural Correlates of Nouns and Verbs in Early Bilinguals Alice H.D. Chan 1 , Geng Li 2 , Ping Li 3 , Li Hai Tan 1 1 Joint Laboratories for Language and Cognitive Neuroscience, University of Hong Kong, Hong Kong, 2 The Jockey Club MRI Engineering Centre, University of Hong Kong, Hong Kong, 3 University of Richmond, Virginia, USA Introduction Research on neural correlates of nouns and verbs has been a focus of recent neuroimaging studies of language. A general result in English shows that verbs are represented in the frontal area, such as the left prefrontal cortex, while nouns in the posterior regions (the temporal-occipital regions). Recent fMRI studies with Chinese, however, have indicated that Chinese nouns and verbs activate a wide range of overlapping brain areas, without a significant difference (1). These findings lead to a question on whether nouns and verbs of Chinese and English are processed differently in Chinese-English bilinguals. In this fMRI study, we used a lexical decision paradigm to examine the neural representation of nouns and verbs in Chinese-English bilinguals. Methods Eleven early Chinese-English bilinguals participated. There were four types of stimuli, English verbs, English nouns, Chinese verbs, and Chinese nouns. Subjects performed a lexical decision task in which they judge whether or not a visually presented stimulus was a real word. Direct comparisons of brain activation were made between nouns and verbs within the same language. Each experimental task was also compared with the fixation baseline. The study was performed on a 1.5 T MRI scanner. A T2*-weighted gradient-echo EPI sequence was used, with the slice thickness = 5 mm, in-plane resolution = 4.3mm x 4.3mm, and TR/TE/FA= 3000 ms/60 ms/90 degree. Twenty-four contiguous axial slices were acquired to cover the whole brain. SPM99 was used for image analysis. Results and Conclusions In early bilinguals, Chinese nouns and verbs provoked a wide range of overlapping brain areas, whereas English nouns and verbs activated different neuroanatomical circuits weighted by word category. In particular, the left insula, left inferior frontal areas (BAs 45 and 47), bilateral precuneus (BA 7), and culmen in right cerebellum played a crucial role in the processing of English verbs, compared with English nouns. In both languages, however, nouns provoke less brain activation than verbs. Our brain mapping study agrees with the prominent theory that functional plasticity of the human brain is driven by language (2). (This research was supported by a grant HKU 3/02C from Research Grants Council of the Hong Kong Government awarded to L.H. Tan.) References (1) Li, P., et al (in press) NeuroImage. (2) Tan, L.H., et al (2003) Hum. Brain Mapp. 18: 158-166.

e18

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 12 Lexico-Semantic Ambiguity is Represented by the Left Dorsal Lateral Frontal Cortex Alice H.D. Chan 1 , Ho-Ling Liu 2 , Peter T. Fox 3 , Jia-Hong Gao 3 , Li Hai Tan 1 1 Joint Laboratories for Language and Cognitive Neuroscience, University of Hong Kong, Hong Kong, 2 School of Medical Technology, Chang Gung University, Taoyuan, Taiwan, 3 Research Imaging Center, University of Texas Health Science Center at San Antonio Introduction One important issue in neuroimaging research on language is how the brain processes and represents lexical semantics. Past studies with various paradigms reveal that the left inferior prefrontal and mid-superior temporal regions play a crucial role in semantic processing (1-4). Those studies, however, typically utilize words having a precise and dominant meaning as stimuli and have not manipulated lexico-semantic ambiguity, a key feature of human language, as an experimental variable. Here we hypothesize that distinct brain areas are recruited in processing semantically precise words and semantically ambiguous words. A word generation paradigm was used to examine whether neuroanatomical networks for meaning are modulated by lexical ambiguity. Methods Eight native Chinese volunteers participated in this study. There were two types of stimuli, words with high semantic ambiguity and words without semantic ambiguity. The subject was asked to silently generate a word that was semantically related to the word they just viewed. Functional images were grouped into semantically ambiguous word and semantically unambiguous (precise) word groups. Activation maps were calculated and compared between the two groups, using a students group t test. The study was performed on a 2 T GE/Elscint Prestige whole-body MRI scanner. A T2*-weighted gradient-echo echo planar imaging (EPI) sequence was used for fMRI scans, with the slice thickness = 6 mm, in-plane resolution = 2.9 × 2.9 mm, and TR/TE/θ = 2000 ms / 45 ms / 90 ο . Twenty contiguous axial slices were acquired to cover the whole brain. Matlab (The Math Works, Inc., Natick, MA, USA) and in-house software were used for image data processing. Results and Conclusions Our fMRI results show that compared with semantically precise words, semantically ambiguous words were mediated by strong brain activations in the left dorsal-lateral frontal areas (BAs 9 and 46), the anterior cingulate (BA 32), and the right inferior parietal lobe (BA 40). Semantically precise words, instead, were associated with the left inferior prefrontal (BA 47) and mid-superior temporal sites. These findings indicate that semantic analysis of written words is a dynamic process involving coordination of widely distributed neural subsystems which are weighted by semantic ambiguity. (This research was supported by a grant from Research Grants Council of the Hong Kong Government (HKU 3/02C) awarded to L.H. Tan.) References (1) Fiez, J.A. (1997) Hum. Brain Mapp. 5: 79-83. (2) Gabrieli, J.D., et al (1998) P. Natl. Acad. Sci. USA 95: 906-913. (3) Petersen, S.E, et al (1989) J. Cognitive Neurosci. 1: 153170. (4) Roskies, A.L., et al (2001) J. Cognitive Neurosci. 136: 829-843.

e19

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e20

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 13 Differences in Cortical Activation Between Balanced and Non-Balanced Bilinguals Performing a Phonological n-back Task 1 Cognitive

Michael WL Chee 1 , Chun-Siong Soon 1 , Hwee-Ling Lee 1 , Christophe Pallier 2 Neuroscience Lab. , Singapore General Hospital, 2 Cognitive Imaging Unit, INSERM U562

Introduction Several lines of evidence suggest that phonological working memory (PWM) plays an important role in determining language acquisition ability (1, 2). We investigated the neural correlates of phonological working memory in young healthy adults who grew up in a English-Chinese bilingual environment (Singapore). Half were balanced bilinguals (BB) highly proficient in both English and Chinese; the other half were non-balanced bilinguals (NB) highly proficient in English but much less proficient in Chinese. Methods Thirty right-handed, English-Chinese bilingual volunteers, 15 in each group, were studied in a block-design fMRI experiment. Language proficiency was assessed by standardized language test-scores. All volunteers were exposed to both languages from 5 years of age or earlier. BB and NB were matched for age [BB: Mean=26.07, SD=5.05; NB: Mean=22.40, SD=2.53] and performance in Raven’s progressive matrices [BB: Mean=28.53, SD=3.80; NB: Mean=29.33, SD=4.48]. Each volunteer was evaluated while performing an auditory n-back task involving unfamiliar French words spoken by two native French speakers: one male, one female. 1-back task blocks contained lures that that appeared in the 2-back or 3-back position, while 2-back task blocks contained lures that appeared in the 1-back or 3-back position. Imaging data was analyzed using random-effects analysis implemented within Brain Voyager 4.9 (threshold of t(28)>2.8, p<.01, uncorrected). Results In both groups, response time increased [F(2,56)=35.076, p<.001] and accuracy declined [F(2,56)=62.235, p<.001] with increasing PWM load. The two groups did not differ in target identification accuracy [F(1,28)<1] but lures elicited more errors in NB than in BB [t(28)=2.24, p<.05] for the 2 back condition. Activations in bilateral prefrontal (BA9/44/45/46), superior temporal (BA41/42/22), superior parietal (BA40/7), left anterior frontal (BA10) and anterior cingulate regions (BA8/32) showed an effect of PWM load in both NB and BB. In the 2-back condition, BB showed significantly greater left anterior insula (BA13) and inferior frontal (BA45) activation relative to NB (Fig. 1). The parametric increase of left insula activation with PWM load is consistent with the hypothesis that it is involved in PWM, possibly contributing to sub-vocal rehearsal. Comparisons of left insula activation between BB and NB further suggest that differences in sub-vocal rehearsal exist between BB and NB. NB showed greater task related activation in the anterior cingulate and greater deactivation in the anterior medial frontal region (BA24/32) relative to BB (Fig. 2), suggesting that PWM exerts greater demands on cognitive resources relating to task performance in NB (3). Conclusions Our results reflect functional anatomical differences between BB and NB, suggesting that these two groups recruit different processing strategies when performing a phonological task. Given the similar educational environment encountered by BB and NB, these differences in phonological processing might be genetically determined. Alternatively, the group differences could reflect use-dependent neural plasticity. Finally, the group differences may be accounted for by some combination of the aforementioned reasons. The differentiation between these possibilities will be the subject of future studies. References 1. Baddeley, A. (2003) J Commun Disord 36, 189-208. 2. Papagno, C. & Vallar, G. (1995) Q J Exp Psychol A 48, 98-107. 3. Gusnard, D. A. & Raichle, M. E. (2001) Nat Rev Neurosci 2, 685-94.

e21

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig. 1 Contrast map showing group differences between BB and NB. Activation is shown in red, deactivation in blue.

Fig. 2 Graphs showing parameter estimates in the left insula and anterior medial frontal regions

e22

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 14 An fMRI study comparing reading and repetition in children and adults Jessica A. Church 1,2 , Heather M. Lugar 1,2,3 , Rebecca S. Coalson 1,2,3 , Steven E. Petersen 1,2,3,4,6 , Bradley L. Schlaggar 1,2,3,4,5 1 Washington University School of Medicine, St. Louis, MO USA, 2 Department of Neurology, 3 Department of Radiology, 4 Department of Anatomy and Neurobiology, 5 Department of Pediatrics, 6 Department of Psychology Introduction An important issue in cognitive neuroscience is how the brain matures to accommodate specific language skills including more recent skills (in an evolutionary sense) such as reading. This study attempts to increase our understanding of the reading process as it develops from children to adults by comparing simple single-word reading and repetition. Methods The brain activation differences between reading aloud printed words and repeating aloud aurally presented words were compared in 50 normal right-handed subjects: 25 children (7-10 yrs, 16 female) and 25 adults (18-35 yrs, 14 female). Task vocabulary was appropriate for 6-7 year-old readers. Each child was tested with Wechsler Abbreviated Scale of Intelligence to screen for unsuspected learning disabilities. Children were also rigorously screened for the absence of neurological or psychological problems. Behavioral data were collected through a voice recorder. Subjects were performance-matched for accuracy and reaction time on both tasks. Event-related fMRI was performed on a Siemens MAGNETOM Vision 1.5T scanner using a standard EPI-BOLD sequence. Correct trials were analyzed according to standardized protocols. Imaging data were analyzed using repeated-measure ANOVAs for the within subject factors of time and task, and the between subject factor of age. Post-hoc ANOVAs used region as a factor. Results For this study, analysis focused on 4 left hemisphere locations deemed important for the functional anatomy of reading: the angular and supramarginal gyri, and anterior and posterior extrastriate cortex. A region in the left angular gyrus was significantly activated only by children for both tasks (A in Fig); adults did not engage this region. A region in the left supramarginal gyrus was activated significantly more by children than adults, though both groups showed significant activations for the region (B in Fig). As expected, a left anterior extrastriate region showed a task effect with greater activation for the read task than for the repeat task, and also showed a significant age effect with children having greater activation than adults (C in Fig). Surprisingly, a left posterior extrastriate region was significantly activated by both children and adults for both tasks; no significant differences were found for task or group (D in Fig). Discussion These findings suggest a systematic functional differentiation among these brain regions during simple word-processing tasks. The age-related decreasing use of regions in the angular and supramarginal gyri may reflect a more general increase in linguistic skill during development, while the decrease in the anterior extrastriate region may reflect increasing specialization in the visual system as reading becomes more skilled. The surprising pattern of activation in the posterior extrastriate region may suggest that this region has close ties to auditory processing. Acknowledgments NIH: NSADA, R21 NS41255

e23

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e24

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 15 Time Course of Use of Semantic and Syntactic Context Information during Spoken-Word Processsing Van den Brink Dannie 1, 2 , Hagoort Peter 1, 2 1 F.C. Donders Centre for Cognitive Neuroimaging, 2 Max Planck Institute for Psycholinguistics Introduction An event-related brain potential experiment was carried out to investigate the time course of the use of different sources of higher-level context information in relation to lexical information during sentence processing. Previous ERP language research has demonstrated the sensitivity of several ERP components to different types of information processing. Whereas the N400 reflects semantic integration processing, the Left Anterior Negativity (LAN) and the P600 are related to syntactic processing. The sensitivity of these ERP components was used to investigate whether specific syntactic information about the word category has priority over semantic information, as suggested by syntax-first models, or whether these two types of information are processed in parallel as proposed by interactive models. Methods Subjects (n = 21) listened to Dutch constraining sentences that contained a critical word that was either (a) a semantically and syntactically congruent noun, (b) a semantically and syntactically incongruent verb, but beginning with the same initial phonemes as the congruent critical word, or (c) a semantically and syntactically incongruent verb, beginning with phonemes that differed from those of the congruent critical word. The moment at which the verb and a possible noun continuation started to diverge was labeled the category violation point (CVP), and was on average 330 ms after word onset for both the incongruent conditions. Results and Discussion The results revealed that relative to the congruent critical words, both incongruent conditions elicited a large N400 followed by a Left Anterior Negativity (LAN) time-locked to CVP and a P600 (see Figure). Important to the issue at hand, the N400 in both semantically and syntactically incongruent conditions set in before information about the word category had become available, as reflected by the LAN. These results provide clear evidence against the claim that building of syntactic-phrase structure, based on word category information, is autonomous and precedes semantic integration processes, as claimed by syntax-first models. Instead, the present study testifies to the incremental and cascaded nature of on-line language processing. It shows that the language system works with the information it receives and uses it immediately. In sum, our data point to an optimal use of contextual information during spoken-word identification by allowing for semantic and syntactic processing to take place in parallel, and integration to proceed with a limited number of candidates.

e25

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1: Frontal LAN effect and posterior N400 and P600 effects in relation to semantic and syntactic violations; CVP: Category Violation Point

e26

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 16 Is morphology a fundamental component of language? Joseph T. Devlin 1 , Helen Jamison 1 , Paul M. Matthews 1 , Laura M. Gonnerman 2 1 Centre for Functional Magnetic Resonance Imaging of the Brain, University of Oxford, 2 Department of Psychology, Lehigh University Morphology is the aspect of language concerned with the internal structure of words. For instance, teacher is composed of a stem (teach) and affix (-er) while corner is not. Although all languages exhibit morphological structure, it is not clear whether morphology is a basic element of language (similar to phonology or semantics), or whether it arises from systematic correspondences in the form and meaning of words. Here we looked for neural evidence of morphological structure. We used fMRI to detect reductions in BOLD signal in a masked priming paradigm where prime-target pairs came from four conditions: Condition

Relations

Examples

1. Unrelated

[form, meaning]

award-MUNCH

2. Orthographically related

[+form, meaning]

corner-CORN

3. Semantically related

[form, +meaning]

sofa-COUCH

4. Morphologically related

[+form, +meaning]

hunter-HUNT

This design allowed us to identify priming effects for morphologically related word pairs and to determine whether these were specific to morphology or were also present for pairs sharing either form or meaning. Methods: Twelve healthy right-handed, native British English speakers performed a visual lexical decision task during scanning (gradient-echo EPI, B0=3T, TR=3s, TE=30ms, 4x4x5mm). Trials began with a fixation point (*) for 1000msec, followed by a 500msec forward mask (##%#$#%##), a 33msec prime word (e.g., conceal) and a 200msec target (e.g., HIDE). A button press response indicated whether the target was a word or not, with equal numbers of word and non-word trials. Images were preprocessed using standard techniques and an RFX analysis identified significant (Z>3.09, p<0.001) activations in areas engaged by word reading. Results and discussion: Participants responded significantly faster to morphologically (605msec) and orthographically (607msec) related word pairs than to unrelated word pairs (630msec), consistent with previous behavioural studies. Semantically related pairs (618 msec) showed a small facilitation effect, although this did not reach significance (p<0.1). Relative to unrelated word pairs, morphologically related pairs led to a reduction in BOLD signal in the angular gyrus (ANG), bilaterally (see Table 2). Similar reductions were observed for both orthographically and semantically related word pairs suggesting that overlap in either form or meaning sufficed to produce this effect. No other regions showed significant morphological priming effects at Z>3.09. In order to reduce false negatives, we lowered the Z-threshold to 2.3 (p<0.01) and identified two additional effects. There was a reduction in a region of left posterior occipito-temporal (pOT) cortex when the visual form of the word pairs overlapped (i.e., in the orthographic and morphological conditions), consistent with a role in orthographic processing. The opposite pattern was seen in the left middle temporal gyrus (mTG): word pairs that shared meaning (i.e., the semantic and morphological conditions) were associated with a reduced BOLD signal consistent with the mTGs role in semantic processing. None of the priming conditions increased BOLD signal relative to unrelated word pairs. While one cannot rule out the possibility that regions not detected in this study also may contribute to morphological processing, the current findings are strikingly consistent with the claim that morphology emerges from the convergence of form and meaning. e27

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Regional BOLD signal reductions in the three priming conditions. For each, the coordinate of the peak voxel in standard space and the Z-value of the reduction are shown. Morphological pairs

Orthographic pairs

Semantic pairs

Region

L. ANG

3.1

4.1

R. ANG

3.2

3.1

3.4

L. pOT

2.7

3.8

L. mTG

2.6

3.3

e28

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 17 Word related mismatch negativity enhancement after binaural stimulation Tanja Endrass 1,2 , Bettina Mohr 3,4 , Friedemann Pulvermüller 3 of Psychology, University of Konstanz, Germany, 2 Department of Psychology, Humboldt University of Berlin, Germany, 3 Medical Research Council, Cognition and Brain Sciences Unit, Cambridge, UK, 4 APU, Cambridge, UK 1 Department

Introduction Earlier studies on word processing demonstrated improved performance in lexical decision tasks after bilateral redundant compared to unilateral visual word stimulation. In contrast, pseudowords did not elicit a bilateral redundancy gain [1,2]. It has been stated that the bilateral redundancy gain is caused by the summation of activity in neuronal cell assemblies distributed over both hemispheres, existing only for known stimuli (words) but not for unknown stimuli (pseudowords). Event-related potential (ERP) studies revealed larger mismatch negativity (MMN) brain responses for words compared with pseudowords [3,4], which has been explained with the activation of stored long-term memory traces for words. The present study was designed to replicate the MMN enhancement to words and to find a neurophysiological equivalent of the bilateral redundancy gain for spoken words. Methods An oddball paradigm was assigned with the German word [ap] (written as "ab") and a pseudoword [ak] which were played monaurally, to the left or right ear, or binaurally simultaneously to both ears. ERPs were recorded from 17 right-handed native speakers of German using a 64-channel electroencephalogram. The MMN mean amplitude was determined over fronto-central electrode locations, separately for the word and pseudoword stimuli and for the three auditory presentation modes (left ear, both ears and right ear). Results The MMN to words was larger than that to pseudowords, possibly reflecting the existence of memory traces for spoken words (see Fig. 1). In the time range from 280-360 ms after stimulus onset, bilateral redundant stimulus presentation led to a further increase of the word related MMN relative to both monaural word conditions. This bilateral redundancy gain was absent for pseudowords (see Fig. 2). Source distributions (minimum norm estimate) of the MMN response indicated higher overall cortical activity in the binaural word condition compared to the monaural word conditions and the pseudoword conditions (Fig. 3). Discussion and Conclusion The enhancement of the MMN after binaurally presented words represents a neuropyhsiological manifestation of a word-specific bilateral redundancy gain. These findings provide evidence for interhemispheric cooperation in the automatic access to memory traces for spoken words. Accordingly, word-related cortical networks distributed over both hemispheres may allow summation of neural activity between and within hemispheres, thereby potentiating the word-related MMN [5]. References [1] Mohr B, Pulvermüller F, Zaidel E (1994). Neuropsychologia 32:105-124. [2] Mohr B, Pulvermüller F, Mittelstädt K, Rayman J (1996). Neuropsychologia 34:1003-1013. [3] Pulvermüller F, Kujala T, Shtyrov Y, Simola J, Tiitinen H, Alku P, Alho K, Martinkauppi S, Ilmoniemi RJ, Näätänen R (2001). NeuroImage 14(3):607-616. [4] Shtyrov Y, Pulvermüller F (2002). NeuroReport 13(4):521-525. [5] Endrass T, Mohr B, Pulvermüller F (2004). Eur J Neurosci (in press).

e29

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1. MMN response elicited by words and pseudowords in the three presentation conditions. Figure 2. The significant Wordness x Condition interaction in the time range from 280-360 ms after stimulus onset. Figure 3. Minimum norm source estimate for binaurally presented words.

e30

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 18 Processing word meaning in the auditory domain: a 3 Tesla Silent Event-Related fMRI study of passive listening to concrete single words and scrambled control stimuli Almut I. Engelien 1, 2 , Oliver Tuescher 1 , Hong Pan 1 , Jarred Finkell 1 , Dan Zimmermann 1 , Yihong Yang 1 , Anna Dickermann 1 , Wolfgang Hermans 1 , Bethany Hodges 1 , David Silbersweig 1 , Emily Stern 1 1 Functional Neuroimaging Laboratory, Weill Medical College of Cornell University, New York, NY, USA, 2 IZKF Research Group 4, Dept. Psychiatry, University of Muenster, Germany Introduction We previously developed a silent event-related imaging technique for auditory fMRI studies (Yang et al., 2000; Engelien et al., 2002). This protocol allows one to measure the BOLD response to individual acoustic events without interference with scanner noise. Here we used it to probe spoken word processing at 3 Tesla. The organization of processing in the brain for words in particular and meaningful events in general is being actively debated (see e.g. Price et al., 2003; Fiebach and Friederici, 2004; Sharp et al., 2004). This study was designed to examine the automatic extraction of meaning during passive listening to spoken linguistic material vs. listening to scrambled control stimuli matched for each individual word. Our scrambling procedure, which is also applicable to other complex sounds (Tuescher et al., submitted), allowed us to maintain the overall frequency spectrum and the envelope curve, while rendering the words unintelligible. This is complementary to studies in the literature relying on white noise or reversed types of acoustic control conditions. Methods The words were recorded by a single female speaker, digitalized, and manipulated in Matlab (Mathworks, Inc.). Words and scrambled words were reliably recognizable and unrecognizable, respectively, as confirmed by behavioral testing. The duration of individual words varied between 490 and 1100 ms. Words were carefully selected to represent both animated semantic categories (human actions, animals) and non-animated object ones (household machines, transportation). Concrete nouns and actions verbs of 2 to 3 syllable length were included. Behavioral pre-testing of the words in 30 healthy volunteers established familiarity, valence, and imageability ratings. During scanning (3T GE scanner), sounds were presented through Koss headphones / IFIS (MRI devices, Inc.). 7 healthy right-handed adults (3F/4M, Age 20-40) participated in the study to date. Words, scrambled words, or baseline epochs (silence) were presented every 20s in 4 runs with 36 stimuli each. The order of epochs within a run was counterbalanced and pseudo-randomized. Data were analyzed with a customized SPM 99 at p < 0.05. Results Both words and scrambled words compared to baseline show highly significant bilateral activation of primary and secondary auditory cortices. Processing of words, when directly compared to scrambled words, was specifically associated with activation in left >> right frontal cortices (dorsolateral prefrontal region) and lateral temporal areas bilaterally (Fig.1). The exact locations of temporal activation were bilaterally in the anterior-lateral part of the planum temporale extending into the superior temporal sulcus. An additional activated area in the left hemisphere only was located in the posterior planum temporale / angular gyrus region (Fig.2). Conclusion In sum, processing of linguistic, meaningful acoustic material vs. matched control events is specifically associated with temporal and frontal areas known to be involved in semantic/linguistic processing, even in the absence of active task demand. References Engelien A et al.(2002) Neuroimage 16: 944. Fiebach CJ and Friederici AD. (2004) Neuropsychologia. 42(1):62-70. Price CJ et al. (2003) Brain Lang 86(2):272-86. Sharp DJ et al. (2004) Cereb Cortex 14(1):1-10. Tuescher O et al.(submitted) Yang Y et al. (2000) MRM 43: 185-190. e31

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e32

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e33

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 19 Artificial grammar, syntactic processing and the left inferior frontal region Christian Forkstam 1,2 , Karl Magnus Petersson 1,2 , Martin Ingvar 1 , Peter Hagoort 2 1 Cognitive neurophysiology research group, Department of clinical neuroscience, Karolinska institutet, Stockholm, Sweden, 2 F.C. Donders centre for cognitive neuroimaging, University of Nijmegen, The netherlands Introduction In an event-related FMRI-study of an artificial grammar (AG) learning task we explored the involvement of the left inferior frontal cortex (Brodmann’s areas [BA] 44, 45 and 47) during a grammaticality classification task using the Reber grammar (Reber 1967). Materials and Methods During acquisition of the AG, 12 subjects were once each day for 8 consecutive days presented a set of consonant strings generated from an AG on a letter-by-letter basis in a sequential order on a computer screen (300ms duration followed by 300ms ISI for each of the 5-12 letter strings), and reconstructed it from memory immediately after the string disappeared. Grammatical classification was performed on day1 and day8 of the study, during which the subjects were asked to classify (’guess’) a new set of consonant-strings (50% grammatical, 50% non-grammatical, controlled for associative chunk strength, Meulemans et al 1997) as accurate as possible based on their immediate intuitive ’gut-feeling’. We also included a sensorimotor decision control task. Whole brain event-related BOLD-FMRI data (Siemens Trio 3T, TR=2.8s) was acquired during classification. The data was analyzed with SPM99 in a region of interest procedure (local maxima, P<0.05, corrected) using a priori masks derived from a recent study (Petersson et al 2003). Results Performance improved with repeated acquisition sessions ([mean,range] = test1[61,49-71], test2[84,53-96] %correct). Relative baseline, several brain regions were activated by grammaticality classification (cluster extent P<0.05 corrected), including the middle-inferior prefrontal (left BA44 and BA45, bilateral BA6&47, right BA46), the anterior cingulate (BA32), the medial frontal (BA8), the bilateral posterior parietal (BA7 and 40), and the left fusiform (BA19) regions, as well as the cerebellum, left thalamus (day1) and the left caudate (day1, P=0.087). Regions within both ROIs (figure) were equally activated relative the sensorimotor baseline on day1 and day8. Grammatical violations (NG vs. G) yielded relative increase on the day8 but not on the day1 testing. Also, the processing of strings with high ACS, showed increased activity in the left BA47. Discussions These data replicate our previous findings (Petersson et al 2003) in that grammaticality classification vs. baseline include several brain regions related to language processing. In addition, ACS had an effect on the anterior-ventral parts of the left inferior frontal region. Following a suggested phonologic versus syntactic/semantic differentiation of the computational infrastructure in the left inferior frontal for language processing (Bookheimer 2002), our results indicate an additional taxation of syntactically related processing regions as proficiency increased, in line with ERP observations (Friederici et al. 2002). We conclude that increased proficiency in an artificial grammar correlate with increased activation in the left mid-anterior inferior frontal region. S Bookheimer, Annual review of neuroscience 25, 151 (2002). AD Friederici, K Steinhauer, E Pfeifer, Proc Natl Acad Sci 99, 529 (2002). T Meulemans, M Van der Linden, Journal of experimental psychology: learning, memory, and cognition 23, 1007 (1997). KM Petersson, C Forkstam, M Ingvar, Cognitive science in press (2003). AS Reber, Journal of verbal learning and verbal behavior 5, 855 (1967).

e34

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

table ROI analysis of left inferior frontal cortex ROI[-48/16/18]

p(corr)

ROI[-44/30/-4]

p(corr)

grammatical vs

day1

-50/8/22

4,41

0,001

-40/20/-4

5,06

<.0005

sensori-motor

day8

-50/8/20

4,92

<.0005

-40/20/-6

3,72

0,007

decision

day1 > 8

nongrammatical

day1

-50/6/26

2,76

0,144

vs. grammatical

day8

-50/8/26

3,22

0,040

-44/20/-4

3,53

0,017

strings

day1 > 8

-48/12/22

2,69

0,155

-40/24/-4

2,90

0,100

low vs high

day1

-40/18/-4

2,87

0,111

associative

day8

-42/24/-4

3,16

0,035

chunk strength

day1 > 8

Main local maxima from each region is reported, ns = no surviving voxels at p>0.05 (uncorrected).

e35

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 20 Left dominant activation of the sensorimotor cortex during lip protrusion Atsushi f*ckunaga 1 , Takayuki Ohira 1 , Masayuki Kamba 2 , Takenori Akiyama 1 , Kenji Hiraga 1 , Seiji Ogawa 2 , Takeshi Kawase 1 1 Department of Neurosurgery, Keio University School of Medicine, 2 Ogawa Laboratories For Brain Function Research, Hamano Life Science Research Foundation Lip protrusion requires bilateral coordinated movements of the facial muscles, but the function of the sensorimotor cortex is still unclear. In this functional study, blood-oxygenation-level-dependent (BOLD) responses in the sensorimotor cortex during lip protrusion were evaluated in a 3-Tesla magnetic field. A block-design paradigm including four 10-sec tasks was carried out in seven healthy right-handed men {ages 22-37, mean 26 years, the Edinburgh Handedness Inventory (mean laterality index = 84.5, standard deviation = 17.1)}. The task consisted of performance of a 2-Hz voluntary oral movement without speech. Single-shot echo planar images were obtained using the following parameters: echo time = 20 ms, repetition time = 1 second, flip angle = 90°, 64×64 matrix, field of view = 200 mm, 21 axial slices, 5 mm thickness. The fMRI software package, Brain Voyager™ 4.9.6 , was employed for all the data analyses. In the group data analysis, the activated sensorimotor area on the left side was larger than that on the right side (Fig. 1), and the peak value of the %BOLD signal responses on the left side (Fig. 2) was about twice as high as that on the right side (Fig. 3). Number of activated voxels in the left sensorimotor cortex was significantly larger than that in the right sensorimotor cortex (data not shown). Thus, the left sensorimotor cortex appears to be more closely involved in lip protrusion than the right sensorimotor cortex. This finding might be related to the fact that Brocas area is in the left hemisphere in most right-handed subjects.

Group data analysis by Brain Voyager™ software showing coronal and sagittal images

e36

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Detailed analysis of the time-course of changes of the BOLD responses in the regions-of-interest in the left sensorimotor cortex

Detailed analysis of the time-course of changes of the BOLD responses in the regions-of-interest in the right sensorimotor cortex

e37

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 21 General networks and performance-related effects in visual speech perception Clayton Fussell , Quentin Summerfield , Deborah Hall MRC Institute of Hearing Research, Nottingham, UK. Speechreading is an important visual language ability requiring the analysis of faces in action and their mapping to stored representations of heard speech. Neuroimaging studies reveal that visual speech processing engages regions in the temporal, frontal and parietal lobes in which activation can vary as a function of speechreading performance. In our view, these studies have several shortcomings. First, simple word lists are presented for speechreading and yet they do not tap the same operations required by the perception of natural speech (natural speech has more complex articulatory boundaries and contains both prosodic cues and semantic context) and are too simple to elicit the normal wide range of individual differences in visual speech analysis. Second, small numbers of participants obviate rigorous group analysis to account for non-specific variability in the data. The present fMRI study recruited 33 healthy subjects to identify regions that are reliably engaged by visual speech processing, regions that overlap with auditory speech processing and regions in which activation varies systematically according to individual speechreading performance. Methods We present the results of 3 experimental conditions: viewing a talker speaking short sentences such as Four ships sailed up the river (talking face), hearing the talker speak the same sentences (heard speech) and non-linguistic closed mouth gestures (gurning face). Subjects identified the number (1-10) in the sentences to provide an on-line measure of speechreading accuracy. FMRI data were analysed using a random-effects analysis in SPM99 and thresholded at P<0.05 (corrected). The network The talking-face versus the gurning-face contrast identified widespread bilateral activation, greater in the left hemisphere. Regions included superior temporal sulcus (STS), middle temporal gyrus (MTG) and inferior frontal gyrus (IFG) that could all play a role in the modality-free processing of linguistic information. See Table for details. Heard speech also generated activation in STS and MTG, but no IFG - pointing toward a certain automaticity of auditory speech perception. Whilst heard speech engaged the superior temporal gyrus (STG), its absence in visual speech suggests that the representation of pre-lexical sound-based information is not a necessary component of the speechreading process. Performance effects Scores on number identification ranged 22-89% across the group. Positive correlations between performance and activation occurred in fusiform and lingual gyri and posterior cingulate (see Table) suggesting that good and poor speechreaders can be distinguished by functional differences in visual analysis and integrating stored speech knowledge with the incoming visual speech stimulus. Although the talking-face contrast for the group reveal no significant STG activation, it was present in some individuals at P<0.001 (uncorrected). Moreover, growth in the size of STG activation was significantly related to speechreading performance. The surprising absence of performance-related activation in IFG could be due to the high cognitive load imposed by the visual analysis of natural speech. Summary Visual speech perception reliably engages a widespread network of regions that are implicated in both visual and linguistic processing and that vary as a function of performance. However, we suggest that sound-based processing in STG might be a consequence of good speechreading rather than a critical stage.

e38

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e39

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 22 Lexical competition and grammatical categories: a PET study of verb and noun generation E. Goutines , P. Péran , G. Raboyeau , D. Cardebat , J-F. Démonet Inserm U455, Toulouse, France Introduction Several neuroimaging studies showed the role of the anterior cingulate cortex l (ACC) or the left inferior prefrontal cortex 2 in verb generation tasks in case of high degree of competition between possible lexical responses. Our study aimed at assessing the neural correlates of this competition effect in word generation tasks involving two lexical categories: nouns and verbs. Methods Eleven right-handed healthy subjects (mean age 56,45 ± 7,49) performed a noun/noun and noun/verb generation task during a PET scan. The protocol included for each task (80 items per task, auditory presentation, one item per 6 seconds) two conditions, Dominant vs Competitor: on the basis of prior normative data, we selected 80 nouns that elicited a dominant response, either in a noun/noun (40) or in a noun/verb (40) generation task and 80 nouns (40 for noun/noun and 40 for noun/verb) that elicited multiple responses. Scanning was performed using the O15-water method and a HR+ Siemens PET camera. Neuroimaging data were analysed using SPM 2 (threshold: P < .01). Results Behavioral results A two-factor ANOVA (grammatical, competition) with repeated measures showed a significant effect of competition (p<.005) with more errors for Competitor items for both grammatical categories. Nevertheless, the interaction between grammatical and competition factors tended to be significant (p<.09) with a quasi significant difference in the competitor condition between the two grammatical categories, the noun/noun generation task eliciting more errors than the noun/verb one (p = .056). PET results The (Competitor Dominant) contrast revealed a pattern involving the right dorsal lateral prefrontal cortex (BA 9), the ACC (BA 24) and the left temporal parietal cortex (BA 22, BA 40). The (noun competitornoun dominant) contrast showed few activated clusters involving BA6 bilaterally and the left putamen. The (verb competitor-verb dominant) contrast showed mainly a left-sided network with temporal parietal areas (BAs 22, 39, 40), inferior frontal area (BA 47) and ACC (BA 24). Finally, the (noun competitor verb competitor) showed mainly a right-sided network involving dorsal lateral frontal areas (BA 9), SMA and premotor areas (BA 6), inferior frontal areas (BA 44, BA 46), temporal-parietal cortex (BAs 21, 40) and ACC (BA 24). Discussion Although a global behavioral competition effect was observed for both grammatical categories, it appeared more marked for nouns. In terms of brain activation, this effect is mainly associated with a right sided network including frontal temporal regions and the ACC. The present results concur with those of Barch et al. (2000) who underlined the role of the ACC in monitoring response conflict, and those of Thompson-Schill et al. (1997) when considering the verb category. For nouns, the results are less clear because of the discrepancy between a marked difference for response accuracy according to the competition factor, and lack of difference of activation between these conditions. A detailed analysis of the nature of responses and response latency might help clarify this problem. 1.Barch D.M et al (2000). Journal of Cognitive Neuroscience, 12, 298-309 2.Thompson-Schill SL et al (1997) Proc. Natl. Acad.Sci. USA, 94, 14792-14797.

e40

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 23 Processing of pitch and lexical tone in speakers of tonal and non-tonal languages. Thomas C Gunter 1 , Sonja Lattner 1 , Trevor B Penney 2 , Hsuan C Chen 2 1 Max-Planck-Institute for Human Cognitive and Brain Sciences, 2 The Chinese University of Hong Kong, Department of Psychology Differences in processing of lexical tone in native speakers of a tonal language (Cantonese) and speakers of a non-tonal language (German) were explored in both a behavioral and an ERP experiment. An easy contrast (tone 1 vs. tone 2) and a difficult contrast (tone 5 vs. tone 6) were presented as blocks of speech stimuli and blocks of sinusoidal sounds, which had the same pitch contour as the speech stimuli. The behavioral experiment showed that the Cantonese participants had an advantage for a same-different judgment of the stimuli. The ERP experiment showed two time windows of MMN-activity, which can be related to pitch level and contour processing. The ERPs also showed that the Cantonese participants were more sensitive to tone information than the Germans, but group performance was similar for sinusoidal sounds. Speakers of a tonal language probably use very early and subtle parameters when processing speech information in their language.

e41

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 24 Handedness effects on cerebral anatomical asymmetry Pierre-Yves Hervé 1 , Fabrice Crivello 1 , Guy Perchey 1 , Bernard Mazoyer 1,2 , Nathalie Tzourio-Mazoyer 1 1 GIN, UMR6194, CNRS/CEA/Univ. Caen & Paris 5, France, 2 IRM CHU Caen, Institut Universitaire de France Introduction Anatomical asymmetries are considered as a correlate and possible substrate of cerebral functional asymmetry, notably that of language. Due to the reduced proportion of left hemispheric specialization for speech in left-handers, reduced anatomical asymmetries in language related cortices of left-handers are expected and were found in several region of interest studies 1;2 . Voxel Based Morphometry (VBM) offers the opportunity to study the anatomical asymmetries of large samples on the whole brain. However, in two studies that used this method to investigate the effects of handedness on brain asymmetry 3;4 , no differences were found, possibly due to a small proportion of left-handers. In this context, we present here the results of a VBM study performed on two equally sized groups of 56 left- and right-handed young male subjects. Materials and methods The 56 right-handed subjects (age 22.7 ± 2.8 yrs) scored between +42.8 to +100 on the Edinburgh Handedness Inventory scale, while the 56 left-handed subjects (age 21.7 ± 2.5 yrs) ranged from 100 to +22.2. The 112 T1 weighted 3D MRI volumes were processed according to the SPM99 optimised VBM protocol 4 , using a symmetrical grey matter template on both segmentation and normalization steps. Resulting individual grey matter probability maps were smoothed with a 12mm FWHM kernel. The difference between these images and their left-right flipped versions were then computed, resulting for each subject in asymmetry index maps. A group comparison was then performed, in which clusters detected in the left hemisphere corresponded to more leftward asymmetry (and those detected in the right hemisphere to more rightward asymmetry) in right-handers compared to left-handers. Results and discussion Focal areas of differing asymmetries between left and right-handers were evidenced (see table and figure). Strikingly, all arose from right-handers being more leftward asymmetric, at the exception of a cluster found in the cerebellum. Most significant results were found in pre-motor and motor areas, namely in the inferior part of the precentral sulcus and in the middle part of the central sulcus, below the hand motor area. Paralleling previous findings 1;2 , right-handers had on average a leftward asymmetry in a part of the PT where left-handers exhibited symmetry (figure). Conclusion VBM allowed to confirm the previously established anatomical difference in PT asymmetry between left and right-handers. It also allowed for the first time to uncover local frontal sites subject to variation of asymmetry with handedness, outer the inferior frontal gyrus. Among these, the inferior precentral region was iteratively documented in phonological language tasks 5 . Whether the observed differences relate to grey matter amounts or shape and position of the considered sulci remains to be determined. References 1. Steinmetz, H., Annals of Neurology, 1991. 2. Foundas, A., Arch. Neurol., 1995. 3. Watkins, K. E., Cereb.Cortex, 2001. 4. Good, C. D., Neuroimage, 2001. 5. Zattore, R.J., Cereb. Cortex, 1996.

e42

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure: Regions of more leftward asymmetry in right-handers relatively to left-handers are presented on the left. The graph presents mean AI and SD of highlighted regions.

e43

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 25 The effect of form and colour integration on the object naming system Julia Hocking 1 , Erin Carroll 2 , Peter Garrard 2 , Cathy Price 1 1 Wellcome Deparment of Imaging Neuroscience, 2 Institute of Cognitive Neuroscience Introduction: Behavioral studies of object recognition have demonstrated a facilitation effect on naming latencies when objects are appropriately colored relative to their black and white counterparts (Price & Humphreys, 1989). Moreover, this facilitation effect is observed more strongly for objects such as fruit and vegetables which tend to have more distinct colors than manmade objects. The study reported here was designed to investigate the neuronal basis of these effects using positron emission tomography (PET). It was predicted that activation in perceptual and possibly semantic processing areas would be less for colored than black and white pictures due to the facilitation provided by the additional perceptual cues, and that this effect would be greater for natural compared with manmade kinds of objects. Methods: 12 subjects were scanned 12 times with PET. They were instructed to name pictures of objects, environmental sounds or color patches. The objects in the pictures were either vegetables, fruits or manmade items which were presented in color or black and white. This resulted in six conditions which enabled us to look for (1) the main effect of form (pictures > environmental sounds & color patches); (2) the interaction of color and form that occurred irrespective of category; and (3) the interaction of color and category. Data were analysed with SPM99 using standardized procedures. Results: (1) Naming objects from pictures increased activation in bilateral posterior and mid-fusiform areas relative to naming objects from sounds or color naming. (2) The integration of color and form (a) increased activation in V4 (-32, -66, -14, Z=5.4) relative to both form only and color only; and (b) decreased activation in the right parahippocampal gyrus (38, -22, -28, Z=5.0) with a corresponding but less significant effect in the left parahippocampal gyrus (-22, -30, -26, Z=4.1). (3) The interaction of color and category did not surpass a level of significance that was corrected for the entire brain. However, a left medial anterior temporal area (-28, 6, -32, Z=4.0) was activated by black and white pictures of natural objects relative to all other conditions, as previously reported (Moore & Price, 1999). Conclusions: The results of the present study reveal an interesting pattern of interactions between the processing of visual objects presented in black and white and in color. Activation in V4 (the perceptual color processing area) was greater when both form and color were present relative to form only or color only. In contrast, activation in the parahippocampal gyri was decreased in the presence of color suggesting that access to higher level semantic processing was facilitated. Likewise, decreased activation in the left medial anterior temporal area that has been reported for natural relative to manmade objects, irrespective of stimulus modality (Devlin et al., 2002) suggests that the presence of colour facilitates access to semantic representations of fruits and vegetables.

e44

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 26 Crosslinguistic fMRI Study of Chinese Tone and Music Perception Li Hsieh , Yow-Ren Chiang Dept. of Audiology & Speech-Language Pathology, Wayne State University An important question for speech perception is - how human brains process speech different from music? Previous neuropsychological evidence indicated existence of specialized neural circuitry for speech and music. Chinese tones provide an ideal proving ground to address this issue. Because Chinese uses pitch variations to express changes in meaning at the word level (tone) in contrary to pitch changes in melody (musical note). The goals of this fMRI study are to investigate the neural substrates of Chinese tone and music perception and to examine the differences between the Chinese speakers and the English speakers according to their language and musical experiences. Previous tonal studies show that Chinese tonal speakers seem to process Chinese tones as linguistic units; whereas, non-tonal English speakers treat the Chinese tones more like music due to lack of tonal experiences in their native language. Some studies on music perception indicate that musicians appear to shift the brain activations from the right hemisphere to the left side while processing music because of their sophisticated musical experiences. Hence, in this study we incorporate English-speaking musicians in the Chinese tonal perception study in order to see if English-speaking musicians treat Chinese tones more like Chinese speakers do, rather than the musically-untrained English speakers. Three subject groups (3 Chinese speakers, 3 English speakers, and 3 English speakers with musical training) performed pitch-comparison tasks on Chinese tones and on musical notes, and a passive-listening task. Subjects heard a sequence of sounds and were to provide a proper response. For the pitch-comparison tasks, subjects were to decide if the pitch patterns of the first and last sounds are the same or not. For the passive-listening task, subjects pressed a response-button in response to each sequence. Each subject had five fMRI scans. Results comparing Chinese tone tasks to passive-listening tasks indicate that Chinese speakers show tonal processing with lateralization to the left hemisphere, specifically in the left frontal area (including left inferior frontal gyrus, left premotor cortex, left orbitofrontal area) and bilateral auditory area, supplementary motor area, and insula. On the other hand, when listening to musical notes, Chinese speakers seem to process musical information in the right frontal region. In the comparison across different language groups, musically-untrained English speakers seem to process Chinese tones more like music in the right hemisphere. In addition, the English-speaking musicians seem to process tones and music bilaterally. Their musical skill training may foster a different strategy in processing Chinese tones than the musically-untrained English speakers do. Our findings indicate that the hemisphere lateralization for speech and music is mediated by individual linguistic background and musical experiences in despite of the fact that both Chinese tones and musical notes are mediated by pitch variations. The organization of neural network for processing pitch patterns seems to be modulated according to the speakers’ language and musical experiences.

e45

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 27 Changes in the activation pattern during the course of sentence comprehension Naho Ikuta 1,2 , Kazuki Iwata 2,3 , Yuko Sassa 2,3 , Jobu Watanabe 2,3 , Yuko Akitsuki 2,4 , Naoki Miura 2,5 , Hideyuki Okamoto 2,6 , Jorge Riera 2,3 , Shigeru Sato 1 , Yoshihiko Matsue 7 , Ryuta Kawashima 2 1 Graduate School of International Cultural Studies, Tohoku Univ., Sendai, Japan, 2 NICHe, Tohoku Univ., Sendai, Japan, 3 LBC Research Center, Tohoku University 21st Century of Excellence Program in Humanities, Sendai, Japan, 4 Dept. Psychiat., Tohoku Univ. School of Med., Sendai, Japan, 5 Graduate School of Eng., Tohoku Univ., Sendai, Japan, 6 Graduate School of Med., Tohoku Univ., Sendai, Japan, 7 Kansei f*ckushi Research Center, Tohoku f*ckushi Univ., Sendai, Japan Introduction In Japanese sentence, both of the sequences of words presented below are grammatically correct. subject, object, transitive verb (SOV) object, subject, transitive verb (OSV) For the SOV sentence comprehension, Ikuta et al. (2002) reported that the left lingual gyrus, the left inferior frontal gyrus, and the left inferior temporal gyrus were specifically activated when the subject, object and transitive verb were presented, respectively. In this study, we examined changes in the activation pattern during the course of OSV sentence comprehension by fMRI. Methods The stimuli consisted of 36 Japanese sentences (OSV) and 108 words composing the sentences. Sentences and words were presented alternately as visual stimuli. Each sentence was divided into the object (OS), the subject (SS), and the transitive verb (VS). In Japanese, a noun can be interpreted as subject or object depending on the case particle. Therefore, OS and SS consisted of a noun and a case particle and VS consisted of a transitive verb. Words consisted of nouns that were the same nouns as those in objects (OW) and subjects (SW) and transitive verbs that were the same as transitive verbs in sentences (VW). Therefore OW and SW consisted of a noun and VW consisted of a transitive verb. OW, SW and VW functioned as the control tasks of OS, SS and VS respectively. Each stimulus was presented one by one for 1000msec and an interstimulus interval was 4600msec. 44 right-handed healthy Japanese volunteers participated in this study. We instructed them to read the presented sentences and words silently. EPI were acquired using a 1.5 T Siemens Vision plus scanner (TR= 4000 msec, slice thickness = 3 mm, 34 slices). We performed data analysis using SPM99. Words used for stimuli and conditions mentioned above were identical with those of Ikuta et al. (2002). Results For the OS versus OW analysis, a statistically significant activation was found in the precuneus and left lingual gyrus. For the SS versus SW analysis, a statistically significant activation was found in the left inferior frontal gyrus. For the VS versus VW analysis, a statistically significant activation was found in the left putamen and right middle frontal gyrus. Discussion As compared with the result of Ikuta et al.(2002), object of OSV sentence activated the same region as subject of SOV sentence (left lingual gyrus). Subject of OSV sentence activated the same region as object of SOV sentence (left inferior frontal gyrus). SOV sentence and OSV sentence were different in regions activated when transitive verb was presented. Therefore, brain activation was defined depending on the order of recognition and grammatical functions like subject did not affect brain activation very much until the transitive verb was presented.

e46

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Therefore, brain activation did not reflect the difference of syntactic structure between SOV and OSV sentences until the transitive verb was presented. We suggest that decisions of syntactic structure in the brain are made at the point that a transitive verb is recognized. Reference Ikuta, N, et al. Bulletin of Tohoku University International Student Center, 2002;6:1-9.

e47

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 28 A meta-analysis on passive auditory language processing Peter Indefrey 1,2 , Anne Cutler 1 1 Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands, 2 F. C. Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands Neurophysiological and neuroimaging techniques attempt to achieve an unmediated reflection of processing. The problem-solving and decision aspects of tasks, such as phoneme detection, lexical decision or semantic categorization, may recruit cognitive subsystems beyond those involved in language processing, and observed contrasts may reflect differences in secondary tasks rather than in relevant underlying processes. In the present meta-analysis, therefore, we investigated what can be learned about auditory language processing from neuroimaging studies in the absence of task confounds. Procedures We analyzed the hemodynamic response maxima reported in 36 publications presenting 55 experiments in which there was no task other than listening to sentences, words, pseudowords or non-linguistic tone stimuli. The reported activation foci were entered in a Talairach and Tournoux (1988) coordinate system. Activation foci reported in MNI-coordinates were converted to the using the nonlinear algorithm of Brett (1999, available at www.mrc-cbu.cam.ac.uk/Imaging/mnispace.html). Reliability of overlap across studies was assessed using the procedure of Indefrey and Levelt (2000, 2004): Based on the average number of activated 1-mm 2 pixels we estimated the chance probability of every pixel to be activated in a single experiment. Assuming this probability, the chance level for a region to be reported as activated in a number of experiments is given by a binomial distribution. If this level was below 5%, we rejected the possibility that the agreement of reports about a given pixel was due to chance. The procedure controlled for the varying number of activation foci and field-of-view sizes across studies. Results and Conclusions Auditory language input is processed in a left posterior frontal and bilateral temporal cortical network. Within this network, no processing level is related to a single cortical area. Temporal areas of reliable focal activation overlap across studies are shown in the figures. Although auditory language processing activates both temporal lobes, they are not equipotential. In the right temporal lobe, activation foci for different auditory stimuli are found in largely overlapping areas. In the left temporal lobe, the reported activation foci for different kinds of auditory stimuli show clearly distinguishable patterns with areas specialized for phonological (posterior superior temporal gyrus), lexical (anterior and posterior superior temporal sulcus/middle temporal gyrus, posterior inferior frontal gyrus), and sentence-level (temporal pole, posterior inferior frontal gyrus) processing.

e48

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e49

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 29 Processing normal speech and hummed sentences Anja K Ischebeck , Kai Alter Max Planck Institute for Human Cognitive and Brain Sciences Natural speech contains segmental information, that is, information about the consonants and vowels making up individual syllables, words and phrases. It also contains prosodic cues, which indicate, for example, phrase boundaries or highlight parts of the utterance dependent on the focus of the speaker. When a naturally spoken sentence is hummed most of its segmental information is lost. Its global intonational contour, however, is preserved. Investigating the communalities and differences between the activation patterns observed for normally spoken and hummed sentence materials during an fMRI measurement can therefore shed some light on the processing of segmental and global prosodic information contained in natural speech. Hummed sentences as well as normally spoken sentences were presented to 16 participants during an fMRI measurement using an event-related design. Hummed and normal sentences strongly activated the superior temporal gyrus including the primary and secondary auditory cortex bilaterally, compared to baseline. Normal speech additionally activated regions in the posterior region of the left superior temporal gyrus, the left inferior frontal gyrus, the thalamus and the cerebellum, compared to hummed sentences. It has been proposed that both hemispheres are involved in the processing of auditory information and that they divide their processing load according to the nature of the auditory signal. A left hemisphere specialization is therefore assumed for the high frequency components of the speech signal (e.g., segmental information) whereas low frequency information (e.g., prosodic information) is processed more in the right hemisphere (Poeppel, 2001; Zatorre & Bellin, 2001). It has been reported that the right hemisphere is more strongly involved in the case of global prosody, that is intonational or sentence prosody (Gandour et al., 2003). Since hummed sentences do not provide segmental information they might involve the right hemisphere more than the left in listening. However, this hypothesis is not supported by the data. A voxel-wise test of laterality revealed that normal speech as well as hummed sentence materials are predominantly processed in the left hemisphere (see Figure). It is possible that speech and speech-like materials involve the left hemisphere more strongly than the right, whereas the right hemisphere might be specialized with regard to the processing of music or environmental sounds.

e50

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 30 Adaptation of auditory N1m responses to phonetic stimuli by visual speech Iiro P Jääskeläinen 1,2 , Ville Ojanen 1 , Jyrki Ahveninen 2 , Toni Auranen 1 , Sari Levänen 2,3 , Riikka Möttönen 1 , Iina Tarnanen 1 , Mikko Sams 1 1 Laboratory of Computational Engineering, Helsinki University of Technology, Espoo, Finland, 2 Massachusetts General Hospital - Massachusetts Institute of Technology - Harvard Medical School A. Martinos Center for Biomedical Imaging, Charlestown, MA, USA, 3 Low Temperature Laboratory, Helsinki University of Technology, Espoo, Finland Introduction Speech perception is known to be influenced by both acoustic and visual information. Behavioral studies have suggested that visual information influences auditory analysis at a relatively early stage [1], and fMRI studies have suggested involvement of superior temporal sulcus [2], purportedly part of a speech processing pathway [3]. Magnetoencephalography (MEG) recordings have indicated that audio-visual speech integration could occur as early as ~100150 ms from stimulus onset [4]. Most studies concentrated on finding cortical equivalents of audio-visual integration neurons observed in superior colliculi (that exhibit > 10-fold increases in firing rates upon spatial and temporal synchrony of auditory and visual stimuli). However, mere deviation from linear summation were observed [2], suggesting that the neural mechanisms underlying audiovisual integration might be different between cortical and subcortical structures. Indeed, recent ferret recordings suggested that the tuning properties of primary auditory cortex neurons are modulated by attentional/motivational state of the animal [5]. Here, we utilized 306-channel magnetoencaphalogaphy (MEG) in 8 healthy volunteers to test whether seeing speech modulates the responsiveness of auditory-cortex neurons tuned on phonetic stimuli. Specifically, we hypothesized that seeing a visual articulation causes adaptation of auditory cortex MEG responses to a subsequently presented phonetic sound. Methods Eight healthy volunteers participated in the study in accordance with the principles of the Helsinki declaration. Auditory test stimuli (Finnish vowels /æ/ and //) were preceded (500-ms lag) by auditory (/æ/, //, and the F2-midpoint between /æ/ and //) or visual articulatory (/æ/ and //) adaptor stimuli. As a separate control, the auditory /æ/ and // stimuli were presented without the adaptors. The subjects task was to behaviorally discriminate between the /æ/ and // test stimuli. At least 60 artifact free responses were collected per stimulus category. Response amplitudes were quantified using equivalent current dipoles fits (ECD) [6]. Results and Discussion The amplitude of the left-hemisphere N1m response to test stimuli was significantly suppressed with auditory (P<0.001) and visual (P<0.05) adaptors, this effect being significantly greater with the auditory adaptors (P<0.01) (see Figure 1). These findings suggest that seeing the articulatory gestures of a speaker influences auditory speech perception via modulation of the responsiveness of auditory cortex feature-detector neurons tuned on phonetic sounds features. This may relate to recent animal studies suggesting that tuning properties of auditory cortex neurons are modulated by the attentional/motivational state of the organism [5]. The fact that adaptation was significantly greater when auditory as compared to visual adaptors preceded the test stimuli can be explained by additional adaptation to acoustic stimulus features. The locations of the ECD fits agreed well the localization of audio-visual integration sites in previous fMRI and MEG studies [2, 4]. References 1. McGurk H, MacDonald J. Nature 264:746748, 1976. 2. Calvert GA et al. Curr Biol 10:649-657, 2000. 3. Binder JR et al. Cereb Cortex 10:512528, 2000. 4. Möttönen R et al. Brain Res Cogn Brain Res 13:417425, 2002. 5. Fritz J et al. Nat Neurosci 6:12161223, 2003. 6. Hämäläinen et al. Rev Mod Phys 65:413497, 1993. e51

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1: Single-subject ECD fits at N1m peak latency with the arrow depicting estimated source strengths and orientations. (a) Responses to the auditory phonemes when presented without preceding adaptor stimuli. (b) Responses to the auditory phonemes when preceded by auditory adaptor stimuli. (c) Responses to the auditory phonemes when preceded by visual articulations as adaptor stimuli.

e52

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 31 Involvement of the right hemisphere in linguistic processing - An event-related functional MRI study Andreas Jansen 1 , Bianca Dräger 1 , Jutta van Randenborgh 1 , Agnes Flöel 1 , Ann-Freya Förster 2 , Michael Deppe 1 , Stefan Knecht 1 1 Dept. of Neurology, University of Münster, Germany, 2 Dept. of Radiology, University of Bochum, Germany Extracting meaning from speech requires the use of semantic and syntactic information. It is still a matter of debate to what extent processing of these different types of linguistic information have common neuroanatomical substrates. In the present study, we used event-related functional magnetic resonance imaging (fMRI) to identify brain regions involved in semantic and syntactic processing. Twelve healthy subjects read well-formed sentences randomly intermixed with sentences with either semantic or morphosyntactic violations. Accuracy and reaction times of responses were measured on-line. As stimulus material, we used simple verb phrases adapted from speech therapy material. For all conditions (semantic violations, syntactic violations, correct sentences) the same lexical material was used. Half of the nouns were in singular, the other half in plural. The semantic condition was created by changing the correct combinations of noun and verb. For the morphosyntactic condition the inflection of the verbs was incorrect (Table 1). An extended bilateral network of brain areas, the epicenters of which are lateralized to the left hemisphere, was activated in all three conditions. Syntactic violations elicited significantly greater activation than semantic violations in right-sided frontal (area 1, BA 46/47) and parietal (area 2 and 3, BA 7/40) regions as well as in left-hemispheric frontal brain areas near classical Brocas area (area 4 and 5) (Fig. 1). No significant activation was found in the semantic condition compared to the morphosyntactic condition. The reaction times between both anomalous conditions were similar, thus indicating a comparable difficulty. Although a similar neural network is recruited for syntactic and semantic processing, distinct right-hemispheric areas are additionally recruited during detection of grammatical errors.

e53

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Table 1: Examples of noun-verb phrases Morphosyntactic

condition

Die Vögel fliegt. (The birds flies.) Der Löwe brüllen. (The lion roar.)

Semantic condition

Die Vögel brüllen. (The birds roar.) Der Löwe fliegt. (The lion flies.)

Correct sentence

Die Vögel fliegen. (The birds fly.) Der Löwe brüllt. (The lion roars.)

Syntactic violations vs semantic violations (random effects analysis, p<0.05 corrected for multiple comparisons)

e54

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 32 Modulation of Right Hemisphere Beta Oscillations (13-35 Hz) in an Auditory Language Perception Task Ole Jensen 1 , Nienke Weder 1 , Marcel Bastiaansen 1 , Ton Dijkstra 2 , Peter Hagoort 1 1 F.C. Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands, 2 Nijmegen Institute for Cognition and Information, Nijmegen, The Netherlands Introduction Event-related potentials (ERPs) have been intensively explored in language research. ERPs have provided important insight into language processing. However, relevant information might also be contained in induced oscillatory responses. Even though many theories predict oscillatory brain activity to play an important role in neuronal processing, limited work has been done on brain oscillations and their relation to language. The aim of this study was to explore oscillatory responses elicited by semantically incongruent words in a classical sentence comprehension paradigm. Methods We recorded the magnetoencephalogram (MEG) from 10 subjects using a 151-sensor CTF system. Auditorily presented sentences in which the last word was inappropriate to the sentence context (semantically incongruent) were compared to congruent sentences. The MEG signals from the axial gradiometers were transformed to the planar gradient. In the planar gradient, the strongest signals are typically observed directly above the source. Time-frequency representations were calculated for the individual trials and then averaged. The sources of oscillatory responses were identified using the beamformer technique SAM and mapped to the individual subjects’ structural MRIs. Results As reported in several previous studies on semantic violations we observed a left lateralized event related field (ERF) over temporal regions peaking ~300-500 ms after the onset of the last word in the sentence. This N400m component was stronger for incongruent compared to congruent words (the N400m effect). The wavelet analysis revealed strong modulation of oscillatory signals in the 18-25 Hz beta band. Over the left hemisphere the beta activity was clearly depressed ~500 ms after the onset of the last word. Over the right hemisphere, the beta activity increased ~1 s after the onset. Both the left hemisphere decrease and the right hemisphere increase were significantly stronger for the incongruent compared to the congruent sentence endings. The SAM analysis identified the oscillatory source of the beta increase in the right hemisphere around the posterior part of the Sylvian fissure. Discussion We suggest that the early increased suppression of beta oscillations in the left hemisphere in response to incongruent sentence endings is explained by increased processing demands to language specific areas. We hypothesize that the subsequent increase in beta oscillations in the right hemisphere is related to suppression. The suppression prevents the right hemisphere from interfering with the increased processing demands in the left hemisphere after incongruent sentence endings.

e55

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Time-frequency representations of congruent subtracted from incongruent sentence endings. Note the left decrease and right increase in the beta band.

The topography of the beta band response (18 - 25 Hz) for congruent (C), incongruent (IC) and incongruent minus congruent (IC-C) sentence endings. The reponse was calculated 0.8-1.5 s after the onset of the last word.

e56

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The beta response following incongruent sentence endings localized by SAM in a representative subject.

e57

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 33 FMRI Evidence for the Neural Correlates of the Typological Differences among L1, L2, and L3 Hyeonjeong Jeong 1,2 , Kazuki Iwata 6,2 , Jobu Watanabe 6,2 , Yuko Sassa 6,2 , Yuko Akitsuki 5,2 , Naho Ikuta 1,2 , Naoki Miura 4,2 , Hideyuki Okamoto 3,2 , Satoru Yokoyama 1,2 , Jorge Riera 6,2 , Tomoki Haji 7 , Nobuo Usui 8 , Masato Taira 7 , Kaoru Horie 1 , Shigeru Sato 1 , Ryuta Kawashima 2 1 Graduate School of International Cultural Studies, Tohoku University, 2 NICHe, Tohoku University, 3 IDAC, Tohoku University, 4 Graduate School of Engineering, Tohoku University, 5 Department of Psychiatry, Tohoku Univ. Graduate School of Medicine, 6 LBC Research Center, Tohoku University, 7 Nihon University Graduate school of medical science, 8 Japan Science and Technology Agency Introduction The purpose of this study is to investigate how typological differences as reflected in the basic word order are related to the activations in multilingual speakers brains and to examine the functional organization of the human brain involved in the comprehension of Korean, Japanese, and English by native speakers of Korean who learned English and Japanese in this order after puberty. Korean and Japanese stand in contrast with English in terms of the basic word order: Korean and Japanese belong to the S (subject) O (object) V (verb) group, whereas English belongs to the SVO group. Methods Thirty right-handed Korean subjects participated in this experiment. All the subjects were late multilinguals who learned English and Japanese after puberty. The average age of the subjects when they started learning Japanese was 20.6(SD 2.8), while their average age was 12.3(SD 1.5) when they started learning English. A block-subtraction paradigm experiment was conducted in which subjects were instructed to listen to simple sentences(each of which consists of four phrases including a transitive verb)of one language per block. White noise was used as the baseline. The data were acquired using an EPI sequence in a 1.5 Tesla Siemens Symphony MRI scanner(TR=5.0 s, TE=50 ms, flip angle 90°, voxel size 4 x 4 x 3 mm, matrix 48×64). All the data processing and group analyses were performed using statistical parametric mapping(SPM 99). The statistical threshold was set at P<0.05 after the correction for multiple comparison. Results During the Korean and Japanese tasks, the bilateral superior temporal cortex (STC) was activated compared with the baseline. During the English task, the bilateral STC, the left inferior frontal cortex (IFC), the supplementary motor area (SMA), and the right cerebellum showed activation. Direct comparisons among the three languages, only English vs. Korean and English vs. Japanese revealed significant differences in the left IFC, the bilateral posterior STC, the SMA, and the right cerebellum. No significant difference was observed when the Korean task was compared with the Japanese task. Discussion We found that there was a significant correlation between the typological contrast with respect to the basic word order and the significantly activated areas of the brain and that languages learned after puberty as L2 or L3 can activate different areas of the brain in accordance with their typological characteristics. In the present experiment, it was found that the bilateral STC was activated in Korean, Japanese, or English. It is worthy of notice that the left IFC including the Brocas area was significantly activated only when the subjects were listening to English despite the fact that all the Korean subjects had learned English(L2) much longer than Japanese(L3). These results suggest that the left inferior frontal cortex is associated with the word order contrast between Korean/Japanese and English when native speakers of Korean are listening to their L2(English)and L3(Japanese).

e58

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 34 Interaction between modality and complexity during reading and listening task Gaël Jobard 1 , Mathieu Vigneau 1 , Marc Joliot 1 , Guy Perchey 1 , Bernard Mazoyer 1,2 , Nathalie Tzourio-Mazoyer 1 1 GIN, UMR 6194, CNRS/CEA/Univ. Caen & Paris 5, France, 2 IRM CHU Caen, Institut Universitaire de France Introduction A previous meta-analysis 1 of word reading allowed revealing a neural network including many similarities with those described during speech 2 and sentences processing 3 . In order to evaluate the specificity of neural networks involved during word reading, we have conducted an fMRI study of listening and reading of linguistic stimuli in the same subjects. Materials and methods 10 right-handed male were submitted to 6 runs of fMRI (BOLD) acquisitions (1.5-T GE Signa) using a block design. Runs alternated 30 sec duration periods of either reading or listening (words, sentences and texts) with a cross fixation task. The different runs were balanced in terms of number and type of information to process (same numbers of words, syllables, level of imageability). Statistical analysis was performed with SPM99. We investigated the networks common to all tasks with a conjunction approach. Then we searched for a main effect of modality (listening, reading), of linguistic complexity (words, sentences, texts), and for their interactions. Results and discussion The common network was mainly constituted of left hemisphere language regions including inferior frontal gyrus (IFG) and superior temporal sulcus (STS), bilateral precentral areas and the so-called visual word form area (VWFA) at the temporo-occipital junction (table 1). A modality effect was detected within bilateral inferior occipital gyri, the left precentral and the left superior parietal gyri where signal increase was significantly higher during reading than listening. Larger signal was observed within bilateral primary auditory area during the speech listening tasks (table 2). A complexity effect was observed within bilateral lingual gyri, and in the anterior part of left superior temporal sulcus (higher signal during text and sentence processing compared to words, without any difference between these two conditions). Activation specific to text was observed in the left middle temporal gyrus (table 3). Finally, among these areas, only the low order visual and auditory areas revealed an interaction between modality and complexity. Within the bilateral lingual gyri, BOLD signal proved to be higher during text and sentences than during words reading, while in these regions a weaker deactivation was observed during words listening than during texts and sentences listening. Likewise, the right Heschl gyrus showed a signal increase during text and a sentence listening while the deactivation observed during reading tasks was larger for text and sentences (table 4). Conclusion The left hemisphere network constituted of the IFG, STS, and VWFA proved to be independent of the modality. Within this network the left temporal regions were sensitive to the complexity as well as, interestingly, the left precentral area. These regions can thus be considered as part of an integrative semantic language network. On the opposite, the right precentral, although activated in all conditions, was not modulated by the linguistic effect attesting for its probable role in a supramodal attentional component of the different tasks. In addition, this approach allowed evidencing the existence of interaction between complexity and modality seated in unimodal areas that is possibly related to an increase in attentional recruitment linked with the structural complexity of the material. 1. Jobard, G. et al. Neuroimage. 2003. 2. Carpentier, A. et al. Epilepsia.. 3. Michael, E. B. Human Brain Mapping. 2001.

e59

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e60

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e61

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 35 Lateralization of syntactic speech in right- and left-handers by fMRI Silke Joergens 1 , Raimund Kleiser 1 , Peter Indefrey 2 , Rüdiger J. Seitz 1 1 Department of Neurology, University Hospital Duesseldorf, Germany, 2 Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands Introduction: Word production and semantic encoding have been reported to show a bilateral hemispheric activation in left-handers in contrast to a left-sided activation in right-handers (Pujol et al. 1999, Hund-Georgiadis et al., 2002). It is unclear whether activation in syntactic speech production shows a similar bihemispheric representation in left-handers. Methods: Twelve right-handed subjects (six female, six male; mean age 29.0 +/- 1.5 years) and twelve left-handed subjects (six female, six male; mean age 27.6 +/- 3.8 years) according to the Edinburgh Handedness Score (Oldfield, 1971) participated in our study. During the experiment verb forms were presented visually for 1250 ms and had to be transformed in past or past perfect and inserted into a neutral sentence frame, which was presented for 1000 ms before the verb. Subjects were instructed to repeat silent the complete sentence including the inflected verb (Sach et al., 2004). As control condition without syntactic information processing we used matched, nonsense letter strings, which had to be looked at. In a block design the conditions were arranged in a pseudo-randomized order. Functional imaging was performed on a Siemens Vision 1.5 T MRI scanner (Erlangen, Germany) using standard echo-planar imaging (EPI, TR 4 s, TE 66 ms, flip angle 90 deg, voxel size 3 x 3 x 4,4 mm3). Image analysis was performed using Brain Voyager 4.9 (Brain Innovation, Maastricht, The Netherlands). We contrasted epochs for the sentence production vs. letter string condition. For contrasting both groups a random effect analysis was performed and the regions of interest were thresholded at p < 0,001 (uncorrected for multiple comparisons). Areas > 50 voxels were reported. Additionally a lateralization index (-1 for strong right, 1 for strong left hemispheric activation) was calculated for each subject. Results: The right-hander showed a significant left lateralized activation in Broca´s area (BA 44), Wernicke´s area (BA 21) and inferior frontal Cortex (BA 9). Additionally a bihemispheric activation difference was found in the parietal cortex (BA 40). In contrast we found in the left-handed group a strong bilaterally activation in frontal speech areas (BA 44/47, BA9). Between group comparison showed a significant greater activation in right frontal cortex (BA 47) in left-handers (Fig. 1). Lateralization indices determined in each subject confirmed a strong left hemispheric lateralization in most right-handed subjects and a strong bilateral lateralization in the left-handed subjects (mean lateralization indices left-handed: 0.37 +/- 0.59 ; mean lateralization indices right-handed 0.89 +/0.19 ; p<0.01) (Fig. 2). Discussion: Syntactic speech is related to a left lateralized activation in right-handers and a bilateral organization in left-handers, particularly in BA 47. Since a similar right hemispheric dominance was not observed for word production and semantic encoding, we suggest a right-hemispheric predominance of syntactic speech production in left-handers. References: (1) Pujol, J. et al., Neurology, 1999, 52(2), p.1038-43 (2) Hund-Georgiadis, M. et al., Experimental Brain Research, 2002, 145:166-176 (3) Oldfield, R.C., Neuropsychologia, 1971, 9(1): 97-113 (4) Sach et al., Neuroreport, 2004, (in press)

e62

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e63

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 36 The left planum temporale, handedness, brain volume, and the neural bases for language comprehension and production. Goulven Josse 1 , Pierre-Yves Hervé 1 , Fabrice Crivello 1 , Bernard Mazoyer 1,2 , Nathalie Tzourio-Mazoyer 1 1 GIN, UMR 6194 CNRS, CEA, Univ. Caen & Paris 5, 2 Institut Universitaire de France Introduction In previous studies we have shown that the inter-individual variability of the neural bases for language comprehension was partly explained by the surface area of the left planum temporale (PT), but not by handedness (HAND) 1,2 . Here, in the same subjects 2 , we first focused on language production based on the hypothesis that HAND explains functional variability of Brocas area 3 which is strongly activated during production. In addition to PT size and HAND, we used brain volume (BV) as a predictor of variability based on Ringos hypothesis that BV explains hemispheric specialization 4 . Also, we re-analysed data on comprehension 2 including BV in the analysis. Methods 20 males (12 left-handers) were submitted to T1 MRI and PET-H215O acquisitions 2 . Left and right PT surface areas were measured 2 and BV was estimated after non-linear normalization to the SPM99 T1 template 5 . Three conditions were replicated: listening to stories (STORY, comprehension), generating verbs semantically associated to heard nouns (VERB, production) and REST. Normalized regional Cerebral Blood Flow (NrCBF) values were computed in anatomical regions of interest (AROI) defined on the MNI single subject template 6 . Selected AROIS were left and right precentral gyri, inferior (F3) and middle (F2) frontal gyri. STORY and VERB were contrasted to REST. Right Left functional lateralization indices (FLIs) were also computed. Left and right PT, HAND and BV were used as predictors in a multiple regression analysis of these functional data. Results Right-handers, subjects with a larger left PT and subjects with a larger brain were more likely to show left activations during VERB (Table and Figure 1). While the previous analysis of comprehension-related data did not evidence any HAND effect 2 , a re-analysis including BV showed that right-handers had leftward asymmetry during comprehension (F2 pars orbitaris). Also, subjects with a larger brain showed leftward asymmetry of the pars triangularis of F3 (Figure 2). In short, not taking into account BV in the previous analysis masked a HAND effect. Indeed, left-handers, who were more likely to have leftward functional asymmetry, also had a larger brain (Figure 3), which made them less likely to have leftward asymmetry. Discussion The relationship between handedness and left prefrontal activations supports that Brocas area was involved during language evolution in gestural communication 3 . The left PT is composed of auditory cortex thought to be specialized for fast processing 7 . In addition, larger brains would require faster processing 4 . Thus, the fact that a larger left PT and a larger brain predicts leftward asymmetry supports that hemispheric specialization for language originated from the need to process language at a certain speed 4 . References 1. Tzourio, Neuroreport,1998. 2. Josse, Cog Brain Res, 2003, 3. Corballis, Behav Brain Sci, 2003. 4. Ringo, Cereb Cortex 1994. 5. Ashburner, Neuroimage, 2000. 6. Tzourio-Mazoyer, Neuroimage, 2002. 7. Jäncke, Neuroimage, 2002.

e64

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Results of the multiple regression analysis of the VERB-REST NrCBF differences. AROI

R2 (p)

Left PT

Right PT

HAND

Coef. (p) Precentral gyrus

F3 pars opercularis

F3 pars triangularis

F3 pars orbitaris

F2 pars orbitaris

Left

0,23 (0,40)

-0,15 (0,55)

0,24 (0,39)

0,19 (0,48)

0,45 (0,09)

Right

0,33 (0,17)

0,42 (0,09)

0,13 (0,60)

0,05 (0,85)

0,42 (0,09)

Left

0,20 (0,50)

0,10 (0,71)

-0,06 (0,84)

0,47 (0,10)

0,26 (0,32)

Right

0,13 (0,69)

0,27 (0,32)

0,13 (0,65)

-0,05 (0,85)

-0,08 (0,77)

Left

0,39 (0,10)

0,10 (0,65)

0,44 (0,09)

0,13 (0,59)

0,52 (0,03)

Right

0,09 (0,82)

0,15 (0,60)

0,23 (0,44)

-0,09 (0,75)

0,03 (0,92)

Left

0,33 (0,17)

0,32 (0,20)

0,17 (0,50)

0,28 (0,26)

0,46 (0,06)

Right

0,20 (0,49)

0,23 (0,39)

0,001 (0,997)

0,27 (0,33)

-0,15 (0,55)

Left

0,40 (0,11)

0,38 (0,12)

0,36 (0,17)

0,09 (0,69)

0,40 (0,11)

Right

0,24 (0,35)

0,23 (0,36)

0,34 (0,22)

0,07 (0,80)

0,20 (0,44)

Left

0,71 (0,0007)

0,53 (0,004)

0,11 (0,53)

0,38 (0,03)

0,72 (0,0003)

Right

0,29 (0,25)

0,31 (0,22)

0,10 (0,71)

0,38 (0,15)

0,07 (0,77)

e65

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e66

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 37 brain activation in picture naming task including picture-word matching, semantic categorization and word reading tasks based on neuropsychological language process model : An fMRI study JaeBum Jung 1 , SungBom Pyun 2 , HyoJung Son 1 , KiChun Nam 1 , Hokyu Lee 3 of Psychology, Korea university, South Korea., 2 Department of Rehabilitation Medicine, Asan medical center, University of Ulsan,College of medicine, 3 Department of Radiology, Asan medical center, University of Ulsan,College of medicine

1 Department

Background Previous neuropsychological studies supposed that there are several stages in picture naming. The first stage is visual perception of pictures, and in order of the next stages are visual input lexicon, semantic system, output lexicon, and finally through the articulation stages who can speak name of the picture. And the reading pathway differs from that of the picture naming. This study was designed to investigate the brain activation pattern according to each stage of neuropsychological model of picture naming (Ellis-Young 1988) using fMRI. Methods We studied 9 healthy right handed adults using four tasks, (1) picture naming task for whole stage from visual input to articulation, (2) picture-word matching task, to exclude articulation stage, (3) word reading task for phonological lexicon, (4) semantic categorization task for semantic system. Study paradigm was block design (12 sec dummy scan, 4 activation block with 5 control blocks per session, total 4 sessions). Pictures and words during activation blocks were selected on the basis of word frequency, word length and semantic categories. Together with this, each activation task has corresponding control task to exclude simple visual search, motor control of bulbar muscles or finger movements. The data acquisition was performed using an ISOL 3.0 Tesla forte MR scanner with EPI sequence (TR/TE = 300/35ms, 4 mm no gap, 20 slices, 64x64 matrix, FOV=220mm, Flip angle 80°). All data processing and statistical analysis were performed using SPM 99 and statical significance level was p<0.01. Results In picture naming task, the middle temporal gyrus, supramarginal gyrus, fusiform gyrus, and anterior inferior frontal lobe (Broca’s area) of left brain was predominantly activated. After subtracting control task (murmuring with meaningless drawing) the left anterior temporal lobe and angular gyrus was activated. In picture-word matching task predominate control task at Broca’s area and superior parietal lobule activation. During semantic categorization left Broca, pars triangularis and superior parietal lobule activation was typical and reading task supramarginal and angular gyrus activation is more remarkable. Difference of activation between picture naming and reading was left fusiform gyrus activation and the post-lexical and articulation process was left fusiform gyrus and Broca’ area and adjacent prefrontal cortex. The left superior parietal lobule, Broca’s area, fusiform gyrus activation was noted after subtracting semantic categorization from picture-word matching task. conclusion Left inferior frontal lobe including Broca’s area was final common pathway of naming and reading tasks. The supramarginal and angular gyrus were more activated compared to predominent activation of fusifrom gyrus in picture-naming task. Semantic and picture-word matching task broadly activates bilateral temporal and parietal area and Broca’s area which demand a decision. In sight of neuropsyhological language processing model, a elaborated analysis is possible.

e67

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 38 fMRI evidence for the ACV98 connectionist model in reading and lexical decision Alexandra Juphard 1 , Monica V. Baciu 1 , Sylviane Valdois 1 , Serge Carbonnel 1 , Bernard Ans 1 , Laurent Lamalle 2 , Mathilde Pachot-Clouard 2 , Christoph Segebarth 2 1 Laboratoire de Psychologie et NeuroCognition, UMR 5105 CNRS - Université Pierre Mendès-France/Université de Savoie, BP 47, 38040 Grenoble Cedex 09, France, 2 Laboratoire de Neuroimagerie Fonctionnelle et Métabolique, U594, UMR INSERM - Université Joseph Fourier, Centre Hospitalier Universitaire, Pavillon B, BP 217, 38043 Grenoble Cedex 09, France Introduction. Following the ACV98 connectionist model of reading [1], behavioral experiments have shown that syllabic length affects naming latencies for pseudo-words but not for words. No effect was observed during lexical decision [2]. This event-related fMRI study aimed at assessing length effect on cerebral activity during silent reading and lexical decision tasks, in order to obtain additional anatomofunctional support to the ACV98 model. Material and Methods. Subjects. Fourteen right-handed volunteers (6 males, mean age 24.4 y) were examined. Tasks. Seven subjects performed a lexical decision task and seven performed a silent reading task on the same French printed words and pseudo-words. Paradigm. A pseudo-randomized event-related paradigm was used. 144 stimuli of 6 types (words and pseudo-words of 2, 3 and 4 syllables) were presented for 500 msec each. MR acquisition was performed on a 3 T MR imager (Brucker), with echo-planar acquisition. 336 volumes of 25 adjacent, axial slices, 5 mm thickness each were imaged (TR = 2000 ms). Data processing was performed using SPM99 [3].Two separate random effect group analyses were performed, one for each task. Contrasts between conditions were determined voxelwise using the GLM. Statistical significance threshold for individual voxels was p<.01 and only clusters of activated voxels were retained (p<.05 uncorrected, 35 voxels). Results. (1) relative to short pseudo-words, reading long pseudo-words elicited increased cerebral activity in regions related to visual, semantic, phonological and motor processing; (2) relative to short words, lexical decision on long words significantly activated greater visual, attentional, phonological, memory and motor processing; (3) relative to short pseudo-words, lexical decision on long pseudo-words significantly activated greater attentional and phonological processing. The table summarizes activated regions. Conclusion. Our results are partially in agreement with ACV98 and dual-route [4] and in disagreement with PDP [5] models of reading. 1. Ans, B. (1998), Psychological Review, 105, 678-723. 2. Juphard, A. (in press), Brain and Cognition. 3. Friston, K. J. (1995), Human Brain Mapping, 3, 189-210. 4. Coltheart, M. (2001), Psychological Review, 108, 121-143. 5. Plaut, D. C. (1996), Psychological Review, 103, 56-115.

e68

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Activated anatomical regions

Talairach coordinates (x, y, z)

Reading Long PW/Short PW Left cerebellum

-18, -50, -10

Right lingual (BA 19)

15, -55, -1

Bilateral inferior frontal (BA 46, 45, 44)

-53, 4, 27; 33, 21, 17; -30, 38, 7

Bilateral lingual (BA 18)

-3, -58, 3; 15, -76, -0

Left middle temporal (BA 39)

-27, -66, 22

Bilateral pre-motor (BA 6)

-24, 1, 28; 30, -4, 32

Lex. Decision Long W/Short W Bilateral cerebellum

-30, -50, -14; 3, -50, -14

Bilateral hippocampus (BA 28, 35)

18, -27, -7; -9, -10, -16; -21, -15, -8

Right cuneus (BA 18, 31)

24, -69, 17

Right fusiform (BA 19, 37)

33, -47, -10

Bilateral cingular (BA 32, 24, 31, 23)

15, 44, 7; -24, -61, 12

Left superior temporal (BA 22)

-68, -40, 11

Lex. Decision Long PW/Short PW Right superior frontal (BA 10)

9, 64, 1

Bilateral superior temporal (BA 22)

-59, 6, 4; 59, -40, 16

Right posterior cingular (BA 23, 31)

6, -61, 8

e69

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 39 Does sex / gender influence language processing? Anelis Kaiser , Esther Kuenzli , Cordula Nitsch Institute of Anatomy, Department of Functional Neuroanatomy, University of Basel In neurosciences sex/gender is a definitive variable like age, that is not prone to measurement error or misclassification. In fact, it is probably considered the variable with the strongest discriminative power. Differences between women and men, if found in language processing, for example, are regarded as an unquestionable matter of fact. Unlike this point of view, poststructuralists see sex/gender as a mere discursive construction. The aim of this study is to combine these two epistemological approaches. Multilingual men and women were interviewed with respect to their individual language biography. Three languages were selected in which they were further examined with functional imaging. The fmri-experiment consisted of thirty-second language production contrasted with an attentional task. During the language condition multilinguals were requested to narrate the events of the previous day. Each language was examined in a separate block in order to guarantee a monolingual mode of language production and to avoid inadvertent mixing of languages. Classical data-processing of a group of N = 30 was carried out concerning the following variables: sex/gender, age of second language acquisition, language learning strategy and proficiency. Particular attention was payed in regard to lateralized or bilateral activation in the classical language processing brain areas. In a second step this primarily neuroscientific experiment dealing with the postulate of gender differences was subjected to a poststructuralistic perspective. In a step-by-step approach the methodological procedures resulting in this postulate were analyzed. So both, a deconstructivistic and a differential approach are synthesized in this study.

e70

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 40 Functional neuroanatomy underlying story processing: a PET study Eunjoo Kang 1 , Misun Yoon 2 , Wonsik Kim 2 , Sunyoung Lee 2 , Hyejin Kang 1 , Dong Soo Lee 1 , Jungmo Lee 3 , Sung-il Kim 2 1 Department of Nuclear Medicine, Seoul National University, Korea, 2 Department of Education, Korea University, Korea, 3 Department of Psychology, Sungkyungkwan University, Korea Introduction Few research has been conducted on linguistic stimulus beyond sentence or word level. In this study, we investigated neural substrates underlying written narratives using three-sentence story with H 2 15 O PET. As a control condition, the sentence-like word sequence was used for word class judgment task. Methods Eight right-handed healthy female volunteers were presented with either stories (Story condition) or syntactic prose (Word condition) while undergoing twelve (4 scans per condition) H 2 15 O PET scans (ECAT EXACT 47,Siemens-CTI, 3d acquisition, 10 mCi per scan). In story condition, participants read eight three sentences, presented visually one at time (3s duration, 3 to 5 word lengths, 1s ISI) and asked to determine whether each story was plausible or not at the end of the third sentence of each triples by pressing mouse buttons. In Word condition, participants read syntactic prose (grammatically correct but nonsensical sentences) while they counted total number of adjectives in each triples and decided if they were above two or not. PET images were preprocessed with motion correction, spatial normalization, and smoothing (FWHM 16 mm) and subjected to individual analysis using general linear model using SPM 99. Group analysis was performed with a random effects model (p < .001 uncorrected, k = 50). Results The rCBF (regional cerebral blood flow) was increased at anterior pole of superior temporal gyrus (BA 38) in the right hemisphere and the dorso-medial frontal gryus (BA 9) in the left hemisphere during the Story condition in comparison to the Word condition. During the Word condition, greater activation was found in the right cerebellum and the left fusiform (BA 20/BA 37) region relative to the Story condition. Conclusion The activation in the left fusiform gyrus in word condition relative to story condition indicated that the task of adjective counting required more detailed search for the visual form of each word on order to check a particular word class (adjective). The present result suggest that the fusiform area is a cortical region for word form processing, which is quite consistent with visual word form area (VWFA) in previous research. The activation of cerebellum in word condition might be due to the mental counting by using their fingers to count the number of adjectives. The right anterior superior temporal and the left dorso-medial frontal regions involved in story processing which requires integration among sentences and thematic processing to maintain coherence. This research was supported by Korean Ministry of Science and Technology (R01-2003-000-11700-0).

e71

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 41 The initial brain activity of imaginary articulation in the right insular cortex. Yutaka Kato 1,2,4 , Motoichiro Kato 1,2 , Masuro Shintani 2 , Fumihiro Yoshino 2,3 of Neuropsychiatry, School of Medicine, Keio University, 2 Laboratory of Brain Research, Oral Health Science Center, Tokyo Dental College, 3 Department of Neuropsychiatry, Tokyo Dental College Ichikawa General Hospital, 4 Inokashira Hospital, Tokyo 1 Department

Introduction Brain activities related to the articulation and word generation are still unclear and controversial. There are several reports on the neural correlates of retrieving mechanical and natural sound in human brain. However, few studies have examined the neural basis on the brain activity of actual articulation or vocalization, in which noise deflection with muscle movements is problematic confounder, specifically, upon MEG measurement. In this study, we demonstrated the event-related-magnetic-fields (ERFs) for the imaginary articulation without actual utterance of letters through the unique procedure, in which the timing of silent articulation was controlled by the sequential stimuli and the visual and auditory stimuli are synchronized. Methods Six subjects were participated in this study. We devised the new tasks that were designed to allow high temporal resolution and reduce the noise from muscle movement to a minimum. In this task, a letter is fixed at the middle of the screen, and the subjects are instructed to concentrate on the letter in order to avoid the effects of extra-ocular muscle movement. A spotlight comes down toward the letter from the top of the screen, and when the spotlight hit the letter, the subjects are requested to pronounce it. At the same time, auditory sound of the letter is provided. After the subjects mastered this procedure, they were instructed to imagine the articulation without actual pronunciation. Moreover, auditory sound was randomly not given once in ten stimuli, which is the target condition for the MEG measures. Using these procedures, we extracted the ERFs of imaginary articulation, which are free from the effects of auditory stimuli and actual vocalization. The magnetic response of "imaginary articulation" was averaged and we evaluated the ERFs by excluding the responses related to visual perception and identified the brain regions which were related to the initial brain activities of articulation. Results From the full view window, the brain activities were detected dominantly at the right temporal to bilateral occipital regions with the latency around 160 ms in all subjects. We analyzed these responses using time-varying multi-dipole estimation and identified three to five equivalent current dipoles (ECDs) in bilateral V1 or MT/V5 regions and the right insula in each subject. Since V1 and MT/V5 were known to be activated by the visuospatial processing of stimuli, we concluded that the right insular cortex was involved in the initial process of articulation. Discussion Functional neuroimaging studies reported that the right insular region and frontal lobes were activated by the sound retrieval task. Neuropsychological studies demonstrated that the patients with apraxia of speech had lesions that encompassed the insula. Therefore, we suggested that the right insular cortex could have a key role in the initiation of the verbal articulation.

e72

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 42 Stereological Volume Estimation of Brocas Area on MR Images Simon S Keller 1,2 , Vanessa Sluming 2,1 , Jack Taylor 2 , Robin Highley 1 , Neil Roberts 1 1 MARIARC, University of Liverpool, UK, 2 Department of Medical Imaging, University of Liverpool, UK There have been few attempts to apply volumetric techniques to investigate the morphology of Brocas area typically defined as the pars opercularis (PO) and the pars triangularis (PT) of the inferior frontal gyrus (IFG) (Figure 1) - using MR imaging. The present study aimed to validate a new mathematically unbiased stereological technique to measure the volume of the PO and PT, using the sulcal contours of the IFG. Methods T1-weighted MR images were acquired for ten healthy individuals (5 males, age range 22-50 years). All images were realigned orthogonal to the bicommissural plane, the grey matter surface rendered, and ROIs demarcated using BrainVoyager, (www.BrainVoyager.com). The PO was demarcated caudally by the inferior precentral sulcus (iPCS), dorsally by the inferior frontal sulcus (IFS) and rostrally by the ascending ramus of the Sylvian fissure (AR). The PT was demarcated caudally by the AR, dorsally by the IFS, and rostro-ventrally by the horizontal ramus of the Sylvian fissure (HR). The medial-most aspect of the IFG was the deepest point of the IFS dorsally, and the tip of the circular insular sulcus ventrally. Point counting and volume estimation was performed using Easymeasure (www.easymeasure.co.uk). Sectioning and point counting intensities were optimised to achieve a coefficient of error of less than 5%. Separation between test points used for point counting was 0.234 cm (i.e., 3 pixels). The slice interval was 0.1 cm (i.e., every section) for the PO and 0.2 cm (i.e., every second section) for the PT. The first section for point counting was selected randomly from the initial section on which the PO or PT could be visualised. Approximately 200-300 points were recorded per structure. Two raters (SK and JT) performed blinded measurements of all cases for an inter-rater study. An intra-rater study was also administered, where one rater (SK) performed measurements of each case twice, four weeks apart. Intra-class correlation coefficients (ICC) were used for inter-and intra-rater analyses. Results Of the ten subjects, the sulcal contours of the IFG prevented measurements of the PO in one individual and of the PT in two individuals. In one hemisphere, the absence of an AR prevented measurement of the right PO and PT. The absence of an HR in the left and right IFG prevented PT measurements in another subject. ICCs for intra- and inter-rater studies of the left PO (n=9) were 0.98 and 0.86, of the right PO (n=9) were 0.97 and 0.86, of the left PT (n=8) were 0.93 and 0.94, and of the right PT (n=8) were 0.92 and 0.89 respectively. Discussion Macroscopic volumetric analysis of Brocas area on MR images cannot be performed for all subjects due to inter-individual differences in the sulcal contours surrounding the IFG. When the sulcal contours permitted volumetric analysis, the method was found to be repeatable and reproducible. We are currently applying this method and morphological analysis to investigate whether left-right asymmetries of Brocas area are related to the side of hemispheric language dominance.

e73

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The PO (purple) and PT (royal blue) is shown in conjunction with the pars orbitalis (light blue).

e74

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 43 An fMRI study of scrambling effects on sentence comprehension Jungho Kim 1 , Masatoshi Koizumi 1 , Naho Ikuta 2,3 , Yuichiro f*ckumitsu 1 , Yuko Akitsuki 3,4 , Kazuki Iwata 3,5 , Hyeonjeong Jeong 2,3 , Naoki Miura 3,6 , Hideyuki Okamoto 3,4 , Yuko Sassa 2,5 , Jobu Watanabe 3,5 , Satoru Yokoyama 2,3 , Noriaki Yusa 7 , Shigeru Sato 2 , Kaoru Horie 2 , Ryuta Kawashima 3 1 Department of Linguistics, Graduate School of Arts and Letters, Tohoku Univ., Sendai, Japan, 2 Graduate School of International Cultural Studies, Tohoku Univ., Sendai, Japan, 3 New Industry Creation Hatchery Center, Tohoku Univ., Sendai, Japan, 4 Graduate School of Medicine, Tohoku Univ., Sendai, Japan, 5 LBC Research Center, Tohoku University 21st Century Center of Excellence Program in Humanities, Sendai, Japan, 6 Graduate School of Engineering, Tohoku Univ., Sendai, Japan, 7 Department of English, Miyagi Gakuin Women’s Univ., Sendai, Japan Introduction Scrambling (or word order alternation) in "free word order languages" such as Japanese and German has generated a considerable literature in theoretical linguistics and psycholinguistics (Karimi ed., 2003). However, few imaging studies have been performed on the processing of non-canonical (or scrambled) word order (Friederici et al., 2003; Inui et al., 1998). Using functional magnetic resonance imaging (fMRI), we investigated the effects of scrambling on cortical activation by directly comparing the brain regions involved in the processing of Japanese transitive sentences with canonical (Subject-Object-Verb) and scrambled (OSV) word order. Since a scrambled sentence is considered to have a more complex syntactic structure than its canonical counterpart in linguistic literature (Saito, 1985), and since Broca’s area has been shown to be sensitive to syntactic complexity (Caplan et al., 2000), we hypothesized that subjects would demonstrate more activation in Broca’s area during the reading of scrambled sentences. Methods Thirty-two right-handed native speakers of Japanese participated in the experiment. In Canonical and Scrambled conditions, grammatical sentences with canonical and scrambled order were visually presented at the center of a screen phrase by phrase. In either condition, half of the sentences were semantically plausible, and the others were semantically anomalous. Each sentence consisted of three phrases, i.e. two noun phrases followed by a verb. Subjects were instructed to judge whether or not the sentences they just read made sense by pressing one of the two buttons (Yes and No) with their right hands. In the Working Memory condition, sequences of three syntactically unrelated phrases were presented. Subjects were asked to judge whether or not the sequences just presented included two identical phrases. In these conditions, the duration time between the onset of the third phrase and the button press for each test item was measured as reaction time. In the Rest condition, only a cross for fixation appeared on the screen. The data were acquired with a 1.5 Tesla Siemens system using an EPI sequence (TR=4000 ms, TE=50 ms, flip angle 90 deg., voxel size 3x3x5 mm). All data processing and group analyses were performed using SPM2. Results and Discussion In the Canonical and Scrambled conditions (each compared with the Rest condition), similar regions were significantly activated including Broca’s, Wernicke’s, premotor and visual areas (figure 1 and 2). This suggests that most cognitive processes involved in the comprehension of scrambled sentences are also responsible for the comprehension of canonical sentences. When these conditions were directly compared (Scrambled - Canonical), we found activated areas in the left dorsal prefrontal cortex and the left inferior frontal gyrus (figure 3); these areas have been claimed to be selectively involved in syntactic processing (Hashimoto and Sakai, 2002). Furthermore, scrambled sentences resulted in significantly greater reaction times than canonical sentences. These results support the linguistic hypothesis that a scrambled sentence is syntactically more complex than its canonical counterpart in that the former contains a filler-gap dependency absent in the latter. The findings are also consistent with the observation that Broca’s aphasics have difficulty comprehending scrambled sentences (Hagiwara and Caplan, 1990). e75

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1: Canonical-Rest (corrected)

Figure 2: Scrambled-Rest (corrected)

Figure 3: Scrambled-Canonical (uncorrected 0.1%)

e76

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 44 THE PROCESSING OF ICONICITY IN GERMAN SIGN LANGUAGE: AN FMRI-STUDY Juliane Klann 1,2,4 , Frank Kastrau 1,3 , Walter Huber 2 1 Interdisciplinary Centre of Clinical Research, 2 Neurolinguistics at the Dept. of Neurology, 3 Dept. of Neurology, University Hospital Aachen, 4 Dept. of Linguistics, University of Cologne, Germany Introduction: Sign language differs from sound language respectively written language in its linguistic use of space and in iconicity (transparency). This leads to the assumption that the cerebral representation of sound/written and sign language does not completely overlap. This view is at variance with recent lesion data and imaging studies which show left localization of sign language in much the same way as for sound language. Therefore both language types might share the same neural substrate, thereby implying a supra-modal representation of language functions in the brain [1-3]. In order to pursue these issues further, we investigated the functional neuroanatomy of processing transparent and non-transparent signs conducting fMRI. Subjects: The participants were 11 prelingually deaf, age ranging from 22 to 43 years, with primary competence in German Sign Language. The control-group consists of 11 hearing, matched for sex, age and education, with no competence in sign language. All subjects were right-handed. Methods: In order to prevent the application of conscious strategies the participants were not informed about the purpose of the study and their focus was directed to a lexical decision task. The detection of possible differences in the processing of transparent versus non-transparent signs was guaranteed through a systematic variation of the sign stimuli in respect to the degree of transparency. The stimuli were presented pseudo-randomized in an event related design. Imaging is done with a Philips 1.5 Tesla Gyroscan NT with standard bird-cage head-coil using a multi-shot T2* weighted gradient echo EPI sequence. Image analysis and statistical evaluation will be carried out using SPM2 (Welcome Department of Cognitive Neurology). Results: In accordance with former empirical data and in disagreement with the expectations derived from linguistic differences, the activation patterns in the deaf participants show recruitment of the left hemisphere’s temporal network, including Wernickes area for processing non-transparent as well as transparent signs. The network activated involves perisylvian areas within the superior temporal gyrus. Elsewhere this was interpreted as an recruitment of primary auditory cortex [4]. Contrary to expectations we found no activation of Broca’s are in deaf participants. This area is consequently documented to be involved in sign language processing in deaf signers as well as in sound and written language processing in hearing non-signers [3]. Even in the present study our hearing participants activated Broca’s area. Therefore the non-activation of this region in deaf subjects seems to mirror an modality- and task-specific effect. In the sign language task hearing non-signers show the expected bilateral activation within occipito-parietal regions. References: [1] Poizner G, Klima ES, Bellugi U: What the Hands Reveal about the Brain. Cambridge: MIT Press (1987) [2] Rönnberg J, Söderfeldt B, Risberg J: The cognitive neuroscience of signed language. Acta Psychologica 2000 (105): 237-254 [3] Klann J, Kastrau F, Kemény S, Huber W (2002): The Neuropsychology of signed and written language: an fMRI-study. Cortex 38: 874-877 [4] Finney EM, Fine I, Dobkins KR (2001): Visual stimuli activate auditory cortex in the deaf. Nature Neuroscience 4 (12): 1171-1173. Acknowledgment: This research project was supported by the "Interdisciplinary Center for Clinical Research on Biomaterials and Tissue-Material-Interaction in Implants (BMBF grant No. 01KS9503/9)".

e77

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 45 PERCEPTION OF VERBAL VERSUS NON-VERBAL MOVEMENTS OF HANDS AND ARMS IN DEAF SIGNERS AND HEARING NON-SIGNERS: AN FMRI-STUDY Juliane Klann 1,2,4 , Frank Kastrau 1,3 , Walter Huber 2 Centre of Clinical Research, 2 Neurolinguitics at the Dept. of Neurology, 3 Dept. of Neurology, University Hospital Aachen, 4 Dept. of Linguistics, University of Cologne, Germany

1 Interdisciplinary

Introduction: Sign languages make extensive use of hand- and arm-movements in three-dimensional space. This implies, that production and perception of sign language requires systems, whose functions are represented in different neural networks, namely the planning respectively the analysis of hand- and arm-movements, that is known to be represented in parietal cortex and the processing of language, that is represented in a perisylvian network mainly concerning frontal and temporal cortical regions. This well established dissociation of neuro-anatomical correlates leads to the assumption, that deaf signers may recruit both systems when they process sign language, whereas hearing non-signers should activate parietal regions only, since they are not familiar with sign language as a natural language system. Lesion data as well as imaging studies document that deaf signers do not recruit parietal cortex for the processing of sign lexemes. In order to pursue these issues further, we compared the processing of verbal and non-verbal hand- and arm- movements in deaf signers and hearing non-signers. To gain an insight in anatomical localization, functional Magnetic Resonance Imaging (fMRI) was conducted. Subjects: The participants were 11 prelingually deaf, age ranging from 22 to 43 years, with primary competence in German Sign Language (DGS = Deutsche Gebärden Sprache). The control-group consists of 11 hearing, matched for sex, age and education, with no competence in DGS. All subjects were right-handed. Methods: The same task was offered to all participants, hearing as well as deaf, in DGS and written German. Both conditions were presented separately, each condition in one block. The first block included written nouns, pseudo- and non-words. The second block contained videotaped single signs, pseudo-signs and non-verbal hand-/arm-movements and non-signs. The participants were instructed to hit a button when they identified a known sign or word. The stimuli were linguistically parallel in both tasks. We chose 60 sign-lexemes, representing objects, 30 pseudo-signs and 30 non-signs. For the written material condition we presented 60 nouns, 30 pseudo-words and 30 non-words. All signs and words consisted of a linguistically simple structure (no compounds) and represented high frequent nouns. The length of written words, pseudo- and non-words varied between four and nine letters. The stimuli were presented pseudorandomized in an event related design. Imaging is done with a Philips 1.5 Tesla Gyruscan NT with standard bird-cage headcoil using a multishot T2* weighted gradient echo EPI sequence. Image analysis and statistical evaluation will be carried out using SPM2 (Welcome Department of Cognitive Neurology). Preliminary Results: Data analysis is not yet finished. Inspection of nine deaf individuals demonstrates indeed activation of temporal regions with no activation of parietal cortex. Acknowledgment: This research project was supported by the "Interdisciplinary Center for Clinical Research on Biomaterials and Tissue-Material-Interaction in Implants (BMBF grant No. 01KS9503/9)".

e78

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 46 The Perception of Phrase Structure in Music Thomas R. Knösche 1 , Christiane Neuhaus 2 , Jens Haueisen 2 , Kai Alter 1 , Angela D. Friederici 1 , Otto W. Witte 2 1 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, 2 Friedrich-SchillerUniversity, Dept. of Neurology, Jena, Germany

For the perception of phrase structures in music or language, the processing of phrase boundaries is crucial. In a previous study, an electrophysiological correlate for this process in speech had been discovered (Closure Positive Shift, CPS) (Steinhauer et al., 1999). We investigate if such a component does exist in music as well, how it relates to the CPS in language, and what are the underlying neural networks. Twelve right-handed musicians listened to 100 different melodies. Each melody was presented in two versions, which consisted of either one or two phrases and were otherwise identical. ERP (32 electrodes) and MEG (whole-head, 148 magnetometers) were recorded. For analysis, a one second period following the offset of the phrase boundary in the two-phrase version was compared to the appropriate period in the single-phrase version using ANOVA. The Multiple Signal Classification (MUSIC) method (Mosher et al., 1992) was employed to identify potentially active brain regions. EEG revealed a centro-parietal deflection peaking at about 550 ms after pause offset (Fig. 1) in response to the phrase boundary (F 1,12 =7.58; p<0.02). No significant lateralization effect was found. The analysis of the MEG waveforms resulted in a longer period of time with significant differences between both conditions (400-500 ms: F 1,14 =12.37, p=0.0043; 500-600 ms: F 1,14 =4.54, p=0.051; 600-700 ms: F 1,14 =11.63, p=0.0042). See Fig. 2. The results for the source analysis are depicted in Fig. 3 and 4. Active regions include posterior and anterior cingulate, retrosplenial cortex, and the posterior part of the hippocampi. The topology of the ERP effect between 500 and 600 ms suggests that it is related to the CPS reported for language (Steinhauer et al., 1999; Steinhauer & Friederici, 2001). Its likely origin from limbic structures such as cingulate cortex and parahippocampal region points towards an interplay of attentional and memory processes being involved in the processing of musical phrase structures. Hence, it seems that the CPS reflects processes that are necessary for shifting resources from the previous phrase to the next one. The different time course of the EEG and MEG effects within the investigated time window proves that several subprocesses involving different neural networks are activated sequentially. References Mosher JC, Lewis PS, Leahy RM. IEEE Transactions on Biomedical Engineering 39(6), 541-557 (1992) Steinhauer K, Alter K, Friederici AD. Nature Neuroscience 2(2), 191-196 (1999) Steinhauer K, Friederici AD. Journal of Psycholinguistic Research 30(3), 267-295 (2001)

e79

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Traces of selected EEG channels. Solid lines: original condition with two phrases. Dotted lines: modified condition with only one phrase.

Traces of selected MEG channels. Solid lines: original condition with two phrases. Dotted lines: modified condition with only one phrase.

e80

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Results of the source localization based on EEG data. The yellow areas denote brain regions that could have contributed to the measured data. Only cerebral areas were considered.

Results of the source localization based on MEG data. The yellow areas denote brain regions that could have contributed to the measured data. Only cerebral areas were considered.

e81

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 47 An fMRI study: Cerebral lateralization to human voice specific response Michihiko Koeda 1 , Hidehiko Takahashi 3 , Yumiko Ikeda 2 , Azusa Hayakawa 2 , Taeko Sasai 2 , Madoka Yamazaki 2 , Kenji Oda 2 , Noriaki Yahata 4 , Kunihiko Asai 3 , Yoshiro Okubo 2,4 , Hiroshi Tanaka 1 1 Bioinformatics, Medical Reseach Institute, Tokyo Medical and Dental University, 2 Biofunctinal Informatics, Tokyo Medical and Dental University, 3 Asai Hospital, 4 Department of psychiatry, Nippon Medical School Voice is recognized selectively along the superior temporal sulcus (STS), and supports conveyance of language information including personal identity or affective state. In contrast to cerebral response to auditory comprehension, the tendency of right lateralized activation of the human voice specific area has been found. Evidence from studies of atypical lateralization, right hemispheric dominance of semantic processing in right-handed people, may rely on human vocal sounds, but there is little data on the neural activity of the temporal lobe to comprehension excluding influence of the voice. In the present study using functional magnetic resonance imaging (fMRI), we asked how laterality of cerebral response to voice was influenced by the presence or not of linguistic information in vocal input. Sixteen subjects listened attentively to forward sentences (SEN), reverse sentences (REV), and recognizable sounds: i.e. bell sounds, etc., (SND) during fMRI scanning respectively. In Group analysis, the REV minus SND signal activity identified right-lateralized activation along the STS, and cigulate response that responded to the voice characteristics. In the same right temporal lobe, the SEN minus SND signal activity were also detected, while the SEN minus REV condition effects was not present. We conclude from this study that the influence of voice should be considered in the assessment of laterality of temporal activation to comprehension.

Cerebral activation; upper: reverse sentences minus recognizable sounds; middle: sentences minus recognisable sounds; lower: sentences minus reverse sentences

e82

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 48 Language processing in the bilingual brain: The role of task and stimulus dimension Sonja A. Kotz 1 , Arturo E. Hernandez 2 , Angela D. Friederici 1 1 Max-Planck-Institute for Human and Cognitive Brain Sciences, 2 Department of Psychology, University of Houston, Texas, USA The respective vulnerability of semantic and syntactic processes in L2 has found renewed interest. Wartenburger et al. (2003) argued that L2 syntax is influenced by AOA, while L2 semantics is modulated by proficiency. ERP evidence supports latter conclusion (Kotz, 2001; Kotz & Elston-Güttler, 2003). Next to proficiency, task demands and semantic stimulus properties may play a crucial role in L2 semantic processing. In a first fMRI experiment (Hernandez et al., 2004) highly proficient L2 English speakers (L1=German) performed a concreteness judgment task. L1 and L2 words were translation equivalents and varied in their orthographic (cognate status) and semantic properties (concreteness). Only a concreteness effect was found in L1, while there was a graded effect of concreteness and cognate status in L2. Particularly interesting in L2 was the activation of a right hemisphere network involved in memory retrieval (the precentral sulcus, the superior parietal lobe, and the hippocampus for words that shared the least orthographic and semantic overlap between L1 and L2 (e.g. abstract non-cognates). Did this activation pattern result from stimulus properties or from task demands? In a second experiment (same subject group) we manipulated task demands while keeping stimulus properties constant. The task was a lexical decision (LDT) go-no go task (20% targets = pseudowords). If the effects of Exp.1 were due to task demands, orthographic and semantic stimulus properties should mainly activate a left hemisphere lexical-semantic network as in other fMRI LDT experiments (i.e., the inferior frontal gyrus (IFG), middle temporal gyrus (MTG), the fusiform gyrus). 160 (80 abstract cognates/non-cognates and 80 concrete cognates/non-cognates) and 80 pseudowords were visually presented to 13 participants during fMRI measurement using an event-related design. A classical concreteness effect, but no cognate effect was found in L1 (left IFG and MTG, right STG). In L2 a graded effect of concreteness and cognate status occurred. Abstract and concrete non-cognates activated the left IFG (44/45)/insula and the left MTG. However, concrete non-cognates showed less IFG activation than abstract non-cognates, additional activation in the right MTG as well as activation in the left inferior parietal lobe. Abstract cognates revealed activation in the right posterior MTG and the left cuneus and concrete cognates in the STS and the ITG bilaterally. These results suggest that task demands clearly influence the processing of semantic stimulus properties in L2. The current data further support the role of proficiency in L2 semantic processing. References Hernandez, A.E., Kotz, S.A., & Friederici, A.D. (2004). Is a maus a mouse: the influence of cognate status on concreteness judgments. Journal of Cognitive Neuroscience, Suppl. Kotz, S.A. (2001). Neurolinguistic evidence for bilingual memory representation: A comparison of reaction times and event-related brain potentials. Bilingualism: Language and Cognition, 4, 143-154. Kotz, S.A. & Elston-Guettler, K. E. The role of proficiency on processing categorical and associative information in the L2: Reaction times and event-related brain potentials. Journal of Neurolinguistics, 17, 215-235. Wartenburger, I, Heekeren, H.R., Abutalebi, J., Cappa, S.F., Villringer, A., & Perani, D. (2003). Early setting of grammatical processing in the bilingual brain. Neuron, 37, 159-170.

e83

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 49 Cortical Dynamics of Chinese Character Processing: an MEG Study Wen-Jui Kuo 1,2 , Chou-Ming Cheng 2 , Daisy L. Hung 1,3 , Ovid J.L. Tzeng 1,3 , Jen-Chuen Hsieh 2,3,4,5 1 Laboratory for Cognitive Neuropsychology, National Yang-Ming University, Taiwan, 2 Laboratory for Integrated Brain Research, Department of Medical Research and Education, Taipei Veterans General Hospital, Taiwan, 3 Institute of Neuroscience, School of Life Science, National Yang-Ming University, Taiwan, 4 Institute of Health Information and Decision Making, National Yang-Ming University, Taiwan, 5 Faculty of Medicine, School of Medicine, National Yang-Ming University, Taiwan Background We used Magnetoencephalography (MEG) to study the cortical dynamics of Chinese character processing. Visual word form analysis serves word recognition as providing information for the subsequent processing, e.g., phonology transformation. This is a dynamical process to extract the invariant, abstract structural representation of written words. Understanding of how the brain proceeds with this course requires information about activation timing within and between different brain sites. Such data can be obtained by MEG that tracks cortical activation sequences with a millisecond temporal accuracy. Since each Chinese character consists of strokes or stroke patterns different from the linear arrangement of alphabetical words, we would expect that the temporal profiles of the neural correlates for Chinese character processing can help illustrate how the information interplay evolves in the orthographic processing network identified by our previous fMRI study (Kuo et al., 2004). Methods Subjects were seven right-handed native Chinese speakers, with no history of neurological disorders, and had normal or corrected-to-normal vision. Subjects were seated upright with a screen at 60 cm apart, and were instructed to read the displayed words covertly. Written words were presented visually at a varying rate (~ 4 sec) to evade subjects anticipation and were exposed for 250 ms. Subjects were asked to name words covertly when they saw the words. MEG measurements were performed in a magnetically shielded room using a whole-head 306-channel neuromagnetometer (Vectorview 4-D Neuroimaging, Helsinki, Finland). MEG signals were sampled at 1000Hz and band-pass filtered at 0.03 to 100Hz. Approximately 80 artifact-free trials were averaged. Before source analysis, the averaged responses were further low-pass filtered at 40 Hz. Equivalent current dipole (ECD) model was employed (Hamalainen et al., 1993), and the ECDs were identified by gradiometer signals one by one, at time points where each specific field pattern was clearest. The sources were then brought into a multi-dipole model where the source locations and orientations were kept fixed while their amplitudes were allow to vary as a function of time to best account for the signals. Results and discussion For each subject, the sensors above visual areas showed early magnetic field changes after stimulus onset. Later in time, signal changes of the sensors above bilateral temporal regions were evident.Although there existed individual differences in the exact spatio-temporal sequence of activation, the bilateral posterior-anterior progression of cortical processing was consistent across the participants, which was consistent with previous evidence (Salmelin et al., 2000). References Hamalainen, M., Hari, R., Ilmoniemi, R. J., Knuutila, J., & Lounasmaa, O. V. (1993). Magnetoencephalography Theory, instrumentation, and applications to non-invasive studies of the working human brain. Reviews of Modern Physics, 65, 413497. Kuo, W.J., Yeh, T.C., Lee, J.R., Chen, L.F., Lee, P.L., Chen, S.S., Ho, L.T. Hung, D.L., Tzeng, O.J.L., and Hsieh, J.C. (2004). Orthographic and phonological processing of Chinese characters: an fMRI study. NeuroImage (In press). Salmelin R., Schnitzler A., Schmitz F., Freund H.J. (2000). Single word reading in developmental stutterers and fluent speakers. Brain, 123 ( Pt 6): 1184-202.

e84

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 50 Language lateralization in young children and adults: Linguistic dimensions matter Jenny Sze-Wei Kwok 1 , Wai Ting Siok 2 , Zhen Jin 3 , Yawei Zeng 3 , Li Hai Tan 1 1 Joint Laboratories for Language and Cognitive Neuroscience, The University of Hong Kong, Hong Kong, 2 Department of Psychology, Stanford University, USA, 3 MRI Division, Beijing 306 Hospital, China A number of imaging studies have compared age-dependent activation and laterality differences between adults and children in order to elucidate the developmental maturation of brain networks for language (Gaillard et al., 2003; Schlaggar et al., 2002). Functional MRI studies with Chinese have found that neural systems for semantic processing as indexed by word generation are as lateralized and stabilized in children of age 7 as in adults (Kwok et al., in press). Here we investigated whether language lateralization in the two age groups is modulated by different linguistic dimensions such as semantics and phonology. In this fMRI study, we used a tone judgment paradigm to examine the neural basis of phonological processing in Chinese adults and children. Methods Twelve normal right-handed children (4 boys, 8 girls; mean age, 7 years and 5 months) and nine normal right-handed adults (5 men, 4 women; mean age, 21 years and 5 months) were scanned. All were native Mandarin speakers. Direct comparisons of brain activation were made in the two age groups performing the same phonological task, in which subjects were required to judge whether the two presented Chinese characters had the same tone. The study was performed on a 2 T GE/Elscint Prestige whole-body MRI scanner. T2*-weighted gradient-echo EPI sequence was used, with slice thickness = 6 mm, in-plane resolution = 2.9 mm x 2.9 mm, and TR/TE/flip angle = 3000 ms/ 45 ms/ 90 ο . Twenty contiguous axial slices were acquired to cover the whole brain. Results and Discussion Significant differences in location and laterality of activation were found between adults and children during Chinese phonological processing. For adults, activations were left lateralized in frontal, parietal and occipital regions; and right lateralized in temporal region. Peak activations occurred in the left lingual gyrus (BA 18) and left inferior parietal lobule (BA 40). The left inferior frontal gyri (BAs 6 and 44) were also involved, though with a relatively weaker activation. Phonological processing, however, provoked more bilateral activations in children, particularly in the frontal region. The left inferior frontal gyri (BAs 44 and 45/46) and right middle frontal gyrus (BA 10) were dominant for children’s tone judgment, while no activation was found in the temporal cortex. These results suggest that the neural systems for Chinese phonological processing are less consolidated in children than in adults, reflecting developmental brain plasticity. Language lateralization at the phonological level is still developing at age 7. In assessing language laterality, linguistic dimensions matter. (This research was supported by a grant from the Research Grants Council of the Hong Kong Government (HKU 3/02C) awarded to L. H. Tan.) References Gaillard, W. D., et al. (2003) Human Brain Mapping 18: 176-185. Kwok, J. S. W., et al. (in press) Human Brain Mapping. Schlaggar, B. L., et al. (2002) Science 296: 1476-1479.

e85

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e86

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 51 Lateralized calcarine activity in binocular symmetric and asymmetric blind subjects (fMRI) Rupert Lanzenberger 1,4 , Christian Windischberger 2 , Alexander Geissler 1 , Markus Barth 3 , Andreas Gartus 1 , Frank Uhl 1 , Daniela Prayer 3 , Lueder Deecke 1 , Ewald Moser 2,3 , Roland Beisteiner 1 1 Department of Neurology, 2 Department of Medical Physics, 3 Department of Radiodiagnostic, 4 Department of General Psychiatry. General Hospital and Medical University of Vienna, Austria INTRODUCTION Several fMRI studies have shown activation in the calcarine cortex during Braille reading in the blinds, but the exact function, representation and modality in the primary visual area are a matter of debate [e.g., 1,2]. Animal studies have clearly demonstrated that age of onset and binocular symmetry have a relevant impact on intra- and intermodal integration, lateralization and topological reorganization of subcortical and cortical visual areas [e.g., 3,4]. Comparing unilateral and bilateral blind subjects might be an important new paradigm to investigate cross-modal plasticity. Data presented here are focused on effects of binocular balance and competition in blind subjects. SUBJECTS 14 blind experienced Braille readers participated in this study (mean age 33.0 +/- 9.5 years; 12 right-handed). 8 of them had binocular symmetric and 6 asymmetric onset (3 earlier or stronger decline of visual acuity in the eye ipsilateral to the preferred reading hand) . METHODS FMRI Measurements: 3T BRUKER Medspec MR scanner. Anatomical images: MDEFT sequence, 32 slices, 256*256, 256*256 FOV, ST= 3mm. Functional data: single-shot EPI sequence, TE/TR=60/6000ms, 128*128, 256*256 FOV, 30 axial slices, ST=3mm. Paradigm: Braille reading was performed in 6 runs using a block design (4 ON-phases/run for Braille reading). Each phase lasted 30sec (5 volumes/phase). 24 different sentences consisting of 120 Braille letters each were presented under the tip of the reading finger using a moving paper tape (2.4cm/sec) with embossed Braille characters. All subjects had to close their eyes during measurments. Preprocessing and Data Analysis: All volumes were 3d-realigned to the first volume (A.I.R). For definition of subject-specific regions of interest (ROI) on high-resolution MR images a triplanar tracing technique was used [6]. ROIs were assigned manually to corresponding structures on EPI images (see Fig.). The ROIs were restricted to the medial calcarine sulcal area. Data analysis was performed using a cross-correlation based technique considering also voxel reliability (risk maps) [5]. Results were calculated for correlation thresholds r>0.2 - r>0.6. Univariate ANOVA with binocular symmetry (symmetric, asymmetric) and hemisphere (ipsilateral, contralateral) as factors and voxel numbers (r>0.4, reliability 50%) as dependent variables were calculated to assess hemispheric differences in ROI voxel counts. RESULTS All symmetric blinds showed lateralization of calcarine activity contralateral to the reading hand irrespective of handedness (50% preferred their left index finger for Braille reading) and age of total blindness. Asymmetric onset in binocular blindness shifted this lateralization pattern towards the ipsilateral calcarine area. This change of lateralization in asymmetric blinds reached from clearly reduced contralateral to strong ipsilateral lateralization. DISCUSSION In 4 of 5 subjects with asymmetric blindness, we found more calcarine activity in the hemisphere contralateral to the later or less affected eye (50% left eye). Therefore, lateralized visual representation (visual imagery, memory, etc.) in asymmetric blind subjects caused by asymmetric influence of the best eye on both hemispheres [7] could be the cause of the lateralization change. REFERENCES [1] Burton H. et al., J. Neurophysiol. 87(2002)589-607.

e87

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

[2] Sadato N. et al., Neuroimage 16(2002)389-400. [3] Kahn D.M. et al., PNAS 99(2002)11429-11434. [4] Toldi J. et al., Prog. Neurobiol.48(1996)191-218. [5] Beisteiner R. et al., Neuroimage 13(2001)1016-1026. [6] Lanzenberger R. et al., HBM-Conference-CD, Sendai 2002. [7] Miki A. et al., Ophthalmic.Res.33(2001)276-82.

ROIs of calcarine sulcal area on corresponding anatomical (left) and functional (right) slices. Cortex of upper bank in red, cortex of lower bank in purple, activated voxels in yellow (r>0.4, reliability: 50%)

e88

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 52 Separating phonology and morphology: new evidence on the functional anatomy of English past-tense Aureliu Lavric , Stela Lavric School of Psychology, University of Exeter, UK The status of regular morphological transformations (e.g. past-tense ed suffixation: call-called, talk-talked) is central to the rekindled controversy between dual- and single-route models of morphological processing. The two theoretical perspectives diverge in postulating (dual-route) or not (single-route) a default combinatorial process for regular items. There is a wealth of psycholinguistic and neuropsychological evidenc, with a considerable number of recent contributions. Moreover, there is a growing body of ERP waveform paradigms that have investigated specific ERP deflections (components) thought to be differentially sensitive to language processing. Unfortunately, the data on the functional anatomy of the morphological phenomena under scrutiny in healthy individuals still remain scarce, with only one detailed report of anatomical dissociation in normal subjects (Lavric et al., 2001). Yet, functional anatomical data could constrain theoretical accounts. For instance, single-route accounts of neuropsychological deficits with regular past-tense in English have centred on phonological factors (phonological complexity and phonological similarity). An interesting question in this context is whether there could be brain regions that are preferentially involved in processing regulars, but which are not intimately associated with phonological processing. The current study attempts to answer this question by using multi-electrode ERPs and sound temporal and spatial (source) analyses. The design is an improvement of an earlier past-tense production paradigm: 70 regular and 70 irregular English verb stems were randomly presented on the screen with the participant instructed to silently generate the corresponding past-tenses. The 64-channel ERPs, time-locked to the presentation of stems, diverged in a late positive component (syntactic positive shift, SPS), with larger amplitude in regulars (Fig.1). In order to avoid arbitrary decisions on time-windowing of ERP analysis, we employed a robust and assumption-free statistical technique (Topographic Analysis of Variance, TANOVA) that compares the ERP maps corresponding to each condition at each time point, establishes at which time-points the maps of the two conditions are reliably different and adjusts the significance threshold in a randomisation procedure. TANOVA found that the only time-window, in which the regular and irregular ERP maps were different, was a portion of the STS (505-560ms). Statistical comparisons of the intra-cortical current density in this time-window, as computed by means of Low-Resolution Tomography (LORETA), revealed higher activity for regulars in dorsolateral prefrontal regions and the anterior cingulate (bilateral, more left) and higher activity for irregulars in the left temporal and temporo-occipital cortices (Fig.2). We conclude the following. The results are generally consistent with the anterior (frontal) vs posterior (temporal) dissociations corresponding to the regular vs irregular types, respectively. More importantly, the areas of higher activity for regulars (e.g. dlPFC), have not been intimately associated with phonological processing. Furthermore, the ERP component that had larger amplitude for regulars (SPS) has been linked to combinatorial (syntactic) processes. Lavric, A., Pizzagalli, D., Forstmeier, S., & Rippon, G. (2001). A double-dissociation of regular and irregular English past-tense production revealed by Event-Related Potentials and Low Resolution ElectromagneticTomography. Clinical Neurophysiology, 112,1833-49.

e89

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1. The ERP difference at the parietal electrode P2. The arrows indicate the extent of the syntactic positive shift (SPS); the box indicates the amplitude difference found by TANOVA.

Figure 2. Voxel-by-voxel LORETA t-tests comparing regular and irregular past-tense production in the 505-560ms time-window; t-values thresholded at p< 0.05, uncorrected for multiple comparisons (df=20). LORETA solutions were first obtained for each time point, averaged within the 55 ms time-window and submitted to voxel-wise t-tests. Red corresponds to REG>IRR, blue corresponds to IRR>REG.

e90

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 53 Lateral Assymetry of the Superior Longitudinal Fasciculus: A White Matter Tractography Study Mariana Lazar , Aaron S Field , JongHoon Lee , Andrew L Alexander University of Wisconsin - Madison Introduction White Matter Tractography (WMT) uses directional information provided by DTI to infer white matter connectivity patterns in the human brain [1]. In this study, WMT was used to reconstruct and segment the superior longitudinal fasciculi (SLF) of both cerebral hemispheres. Left-right asymmetry of the fasciculi volume was investigated. SLF connects cortical regions of the frontal and temporal lobes that are involved in language processing. Along its course the fasciculi also connect with cortical regions of the parietal lobe. Methods The study included fifteen healthy, right-handed subjects (10 males and 5 females, median age =20 years). DTI images were obtained on a 3T MRI scanner using a cardiac-gated spin-echo EPI pulse-sequence. 3 mm thick axial slices were acquired with an in-plane resolution of 0.94x0.94 mm 2 . 2D image registration [2] and field map correction were applied to correct for motion and image distortions. WMT was performed using a streamline algorithm [1]. A complete set of brain fiber trajectories was obtained by placing seeds in voxels with FA greater than 0.4. Trajectories were terminated if they reached regions with FA lower than 0.2 or if the angle between two consecutive steps was greater than 45°. Those trajectories that intersected right and left SLF coronal cross-sections defined onto FA maps were assigned to the tract. SLF sub-segmentation: The trajectories that connected with the temporal lobe regions were segmented out from the complete set of SLF trajectories. Results SLF tractograms for six subjects are shown in Figure 1. The shape of the estimated tracts varied both among subjects, and between hemispheres. For 14 out of 15 subjects, the left SLF presented a larger volume of connections with the temporal lobe (represented in yellow in Figure 1). For one subject (Subject 10) no temporal connections of SLF were originally detected in the right hemisphere since the entire tract appeared to curve laterally toward the parietal cortex. Subsequent WMT performed using lower anisotropy and angle threshold revealed a small, highly curved bundle terminating into superior temporal gyrus. An asymmetry index was calculated for the temporal connections as: ai=[lv-rv]/[lv+rv], with lv the left and rv the right volume. The ai values varied between 0.307 and 0.740 (median=0.235). A paired t-test of left versus right normalized volumes showed that the difference is statistically significant (Z=3.33, p=0.005). Discussion In this study, WMT was used to segment and investigate the hemispheric asymmetry of right and left SLF volumes. A larger volume of temporal connections was measured for the left fasciculus for 14 out of 15 subjects. This result is consistent with previous findings of higher white matter density of the left SLF [3] and may be related to the predominant dominance of the left hemisphere for language function. One limitation of the study is the inclusion in measured volume of both the tract trunk and the branches that connect laterally to the cortex. Future studies aim a detailed classification of the trajectories that compose SLF as a function of the cortical areas they connect as well as measurements of trunk-only volume and diameter. References: [1] Mori, 2002; [2] Woods,1998; [3]Paus,1999.

e91

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1. Tractograms of the left and right SLF for six of the subjects. A larger volume of temporal connections (in yellow) is apparent for the left SLF for all subjects except Subject 12. Tractograms were generated by first projecting the superior SLF bundles (in red), and then the temporal connections (in yellow) onto fractional anisotropy maps.

e92

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 54 Hemispheric Asymmetry in Syntactic Processing: an event-related fMRI study Byeong-taek Lee 1 , Jeong-sug Kyong 2 , Kyoung-Min Lee 1,2 , Cheongtag Kim 3 1 Department of Neurology, Seoul National University, 2 Interdisciplinary Program in Cognitive Sciences, Seoul National University, 3 Department of Psychology, Seoul National University Previous investigation in our laboratory showed that facilitative effects of TMS on syntactic processing vary with the level of parsing load. Furthermore, we observed that TMS on the Brocas area hom*ologue in the right hemisphere was more effective than that on Brocas area in facilitating comprehension of more complex sentences with the OSV/SO word order (Kyong & Lee, manuscript in preparation). From these findings, we developed a hypothesis that the right hemisphere gets involved in syntactic processing only when the parsing load is higher. The goal of the current study is to examine this hypothesis further by using the event-related fMRI technique. Ten healthy volunteers were imaged using 1.5T Signa MR scanner while they read relative sentences for three blocks. Each block consisted of 3 conditions: Conjunctive sentence condition as control condition, SOV/SS (relatively low parsing load), and OSV/SO (high parsing load) condition. All the sentences in 3 conditions were constructed to have the same number of propositions in order to control a differential load due to manipulating propositions. The areas activated in the SOV/SS condition in comparison with the conjunction condition were found in the left Broca’s area, such as the left inferior frontal gyrus (BA 47) and the middle frontal gyrus (BA 46). The areas activated in OSV/SO condition in comparison with conjunctive condition included bilateral Broca’s area and other areas such as the left fusiform gyrus (BA 37). A direct contrast between OSV/SO and SOV/SS showed that the area more activated in OSV/SO condition was at the right insula, which extended into Broca’s area. These findings are quite consistent with the hypothesis that syntactic processing recruits the Broca’s hom*ologue of the right hemisphere when its parsing load is higher. This pattern of activation presumably underlies the facilitative effect observed in our previous study using sub-threshold TMS on these areas.

Figure 1. Main effects of SOV/SS > Conjunctive (red), OSV/SO > Conjunctive (green). Yellow represents overlap of red and green. (n = 10, p < .05, extent threshold = 10)

e93

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 55 Korean-English Bilingual’s Brain Activation: Age of Acquisition of the Second Language Seungbok Lee 1 , EunKyoung Yeon 2 , KwanJin Chung 3 , HyoWoon Yoon 4 1 Dept. of Psychology, Chungbuk National University, Korea, 2 Korean Educational Development Institute, Korea, 3 Brain Imaging Research Center, Pittsburgh, USA, 4 Brain Research Center, KAIST, Korea We examined the images on fMRI in early(experiment 1) and late(experiment 2) Korean-English bilinguals during the inherent production task. The subjects were instructed to describe events that occurred during a specified period of the near past. Unlike the Kim et al.(1997)s task, we did not train the subjects before the scanning sessions, because the planning process itself was emphasized to be important(Levelt et al., 1987). The SPM99 was used to analyze the results. The activation area during production in the six early bilinguals(experiment 1) involved the precentral gyrus and the middle frontal gyrus(BA 6) in both languages, and the right posterior cingulate and the right precentral gyrus in English. The activated areas seemed be overlapping in the two languages. To compare this result with the late bilinguals, we recruited the 11 late bilinguals, who were moderately proficient in L2(English), and applied the fMRI scanning while doing the same task as the one used in experiment 1. Quite different regions were activated in these late bilinguals. The left middle frontal gyrus(BA 10), the bilateral lentiform nucleus, the left cuneus(BA 18), the left cingulate gyrus(BA 32), and the left lingual gyrus(BA 18) were involved in their L1(Korean) production. The activated regions during L2(English) production were the left middle(BA 10, 46) and the inferior frontal gyrus(BA 45), and the bilateral lentiform nucleus. The frontal areas were more activated in L2. The lentiform nucleus was activated in both languages, but other areas were different. We conclude that in early bilinguals, the same or similar regions were involved in producing the two languages. They seemed to focus more to the speaking act. However, the activated areas were quite different in the late bilinguals. The frontal areas are more involved in these late learners when speaking L2. These results suggest that frontal areas play a more important role in L2 production in the case of the late bilinguals.

e94

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The activation areas in early (exp.1) and late bilinguals(exp.2) Early bilinguals(Exp.1) L1(Korean)

Late bilinguals(Exp.2)

L2(English)

L1(Korean)

L2(English)

x,y,z(mm)

cortical areas(BA)

x,y,z(mm)

cortical areas(BA)

x,y,z(mm)

cortical areas(BA)

x,y,z(mm)

cortical areas(BA)

-12,-11,8

L thalamus

12,-63, 12

posterior cingulate(6)

20,-7,6

R lentiform nucleus

-50,22,8

L inferior frontal(45)

-46,-3,48

L precentral(6)

-32,5,59

L middle frontal(6)

-22,4,0

L lentiform nucleus

-18,6,9

L lentiform nucleus

-36, -8, 43

L middle frontal(6)

-4,-17,10

L Thalamus

-2,29,30

L cingulate gyrus(32)

-22,13,-4

L lentiform nucleus

-51,2,44

L precentral(6)

-46,-3,48

L precentral(6)

-2,-74,6

L cuneus(18)

-18,-17,16

L thalamus

42,-6,32

L precentral(6)

-38,-6,33

L precentral(6)

-12,-54,5

L lingual gyrus(18)

20,6,9

R lentiform nucleus

38,-8,30

R precentral(6)

-8,-14,-4

L thalamus

-36,47,12

L middle frontal

-38,49,12

L midbrain

-42,45,7

L middle frontal

22,8,0

R lentiform nucleus

-4,-84,26

L cuneus

The activation areas in early K-E bilinguals: red blots(Korean, L1), green blots(English, L2)

e95

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The activation areas in late bilinguals: red blots(Korean, L1), green blots(English, L2)

e96

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 56 The Neural Basis of Bilingual Sentence Comprehension in Korean-English Late Bilinguals Seungbok Lee 1 , EunKyoung Yeon 2 , HyoWoon Yoon 3 1 Dept. of Psychology, Chungbuk National University, Korea, 2 Korean Educational Development Institute, Korea, 3 Brain Research Center, KAIST, Korea Recent neuroimaging studies have drawn a heterogeneous picture on the cerebral organization of a first language(L1) and a second language(L2). There had been some evidences that L1 and L2 are supported by identical brain regions, whereas contradictory results indicate a differential cerebral organization for L1 and L2. Concerning the sentence processing, neuroimgaing studies of syntactic and lexico-semantic processing had produced different results as to the cortical regions(left prefrontal cortex[Homae et al, 2002; Ni et al., 2000] or superior-temporal region[Lee H. et al., 2003; Kuperberg et al., 2000]). These contradictions may be caused from the difference of the language studied. Bilingual brain representation could provide a good suggestion on the processing of the linguistic aspects. We applied fMRI to examine the bilingual brain activation at semantic aspects of sentence processing, in order to compare our previous study on syntactic processing(Lee S. et al. 2003). In semantic decision task, the visually presented sentences were judged based on the semantic appropriateness. The cortical activations were compared with the control task, which was ordering judgment task with the sentences of the same length. The SPM99 was employed to analyze the results obtained. The activated areas were calculated by the substraction method. While processing the Korean sentences, the left inferior frontal gyrus, the bilateral lingual gyrus, the left cuneus, and the left superior and middle temporal gyrus were activated. While doing the tasks in English, the areas were the left inferior frontal gyrus, the left medial frontal gyrus, the left lingual gyrus and the left cuneus. There were some overlapping regions in frontal and occipital regions. However, it was noted that the superior and middle temporal gyrus were activated in Korean(L1). The frontal region was more activated in English(L2). We(Lee S. et al, 2003) have found the same result in the syntactic judgment task. Both of the previous and present results suggest that the L1 processing was more involved in the temporal region and L2 in frontal region. Further studies on bilingualism would be focused on these two regions to examine whether the acquisition age difference really exists or not. Table.1. . The activation areas in the semantic judgment task in L1 and L2 L1(Korean) cortical areas(BA)

-48, -54, 12

Left anterior cingulate(32)

-50, -69, 11

Left Lentiform nucleus

-51, -65, 18

Right Lentiform nucleus(putamen)

-16, -100, 11 -16, -78, -5 -4, -66, 3

e97

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The activation areas in the semantic judgment task: red blots(Korean, L1), green blots(English, L2)

The activation areas in the syntactic judgment task: red blots(Korean, L1), green blots(English, L2)

e98

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 57 Functional MR Brain Imaging of Cantonese Rhymes Wing-kit Lee 1 , Sze-wing Tang 2 , Kwok-wing Tang 3 , Shing-shun Lo 4 , Kenneth K. Kwong 5 , Suk-tak Chan 1 1 Department of Optometry and Radiography, The Hong Kong Polytechnic University, Hong Kong., 2 Department of Chinese and Bilingual Studies, The Hong Kong Polytechnic University, Hong Kong., 3 Department of Diagnostic Radiology and Imaging, Queen Elizabeth Hospital, Hong Kong., 4 Department of Radiology and Organ Imaging, United Christian Hospital, Hong Kong., 5 Department of Radiology, Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Harvard Medical School, U.S.A. Introduction: The determination of hemispheric dominance for language is an important component in neurosurgical planning. Previous functional studies have demonstrated that major language processing areas at the inferior frontal (Brocas area) and posterior temporal gyri (Wernickes area) were activated by auditory stimuli with the basic component of syllables (English vowels). The purpose of this study was to investigate the feasibility of using auditory stimuli with the rhymes in Cantonese, a major southern Chinese dialect, in the determination of language processing areas, and hence the language dominance, in native Cantonese speakers. Materials and Method: 12 right-handed native Cantonese speakers were studied by functional magnetic resonance imaging (fMRI) for language processing areas with an event-related auditory recognition task consisting of 2 conditions: Cantonese rhymes and filtered pitch (i.e. rhymes filtered to retain the duration and the tone of the rhymes only). The semantic processing areas activated by an event-related visual language task, comparing Chinese / Cantonese characters and Korean characters, were used as a reference to compare with those activated by the auditory task. Results: Direct comparison between the two conditions (Cantonese rhymes vs. filtered pitch) showed significantly more functional activations with Cantonese rhymes at the left superior temporal gyrus. This difference was not seen in the visual language task (Chinese / Cantonese characters vs. Korean characters). Our result is consistent with previous reports, which proposed that superior temporal gyrus was involved in phonology. Brain activations at the inferior frontal gyrus, also observed with Cantonese rhymes, were however not as strong as those activated by visual stimuli using Chinese / Cantonese characters. Conclusion: Our findings suggest that superior temporal gyrus has more activations in auditory task than in visual task. Differences of the language processing network between auditory and visual stimuli need to be taken into account when interpreting the language dominance from such brain activations. A larger sample size with different auditory language conditions would be needed for further investigation.

e99

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 58 Cortical correlates of translation performance in Finnish-Norwegian bilinguals Minna Lehtonen 1 , Matti Laine 1 , Jussi Niemi 2 , Tormod Thomsen 3 , Kenneth Hugdahl 3 1 Department of Psychology, Åbo Akademi University, Turku, Finland, 2 Department of Linguistics, University of Joensuu, Joensuu, Finland, 3 Department of Biological and Medical Psychology, University of Bergen, Bergen, Norway Introduction Translating a sentence from one language to another is a demanding task that requires encoding the original sentence into working memory, meaning analysis, and retrieval of the corresponding words and sentence structure in the other language. An executive control system must mediate the whole process. In the very few studies published on the subject, translation has elicited activation in various brain areas, including the left frontal cortex [1,2], the anterior cingulate, subcortical structures [3] and the inferior temporal cortex [1,2]. Translating has been mostly studied by using single words as stimuli [1,3], but in the present study, we employed a more natural task, sentence translation. Methods The 11 participants (10 females, age 22 41) were native Finnish-speakers who had learned Norwegian in adulthood. The tasks they performed were 1) translating sentences from Finnish to Norwegian and 2) reading sentences in Finnish (control task), and then deciding, depending on the task, whether the later presented probe sentence was a correct translation equivalent or exactly the same as the first one. Two types of test sentences were included in both tasks: one requiring a word order change in translation, the other one not. The design was blocked so that 3 sentences (time for both the test sentence and the probe 10 s) were included in each block following a 20-second rest between blocks, with 18 blocks for each of the two runs (translation and control). The fMRI data were collected with a 1.5 T Siemens Vision Plus scanner. 300 BOLD sensitive EPI volume measurements were done for each of the two runs. Each volume measurement consisted of 30 axial slices covering most of the cerebrum (TR = 3 s., FA = 50 deg., TE = 60 ms., Slice thickness = 4 mm., FOV = 230 mm., Matrix = 64x64). A fixed-effects analysis was performed using SPM2 thresholded at p < .001 (uncorrected) and p < .05 (corrected) for selected contrasts. Results and discussion Behaviorally, the translation task was more demanding, eliciting longer reaction times and higher error rates than the control task, but both tasks were manageable for the subjects. The translation task (in contrast to the control) resulted in increased activation in the left inferior frontal gyrus (BA 47) and in the left basal ganglia, but the word order manipulation in translation did not have any effects on activation patterns. The left inferior frontal gyrus has been suggested to be part of a semantic executive system that controls the retrieval of semantic information [4]. Basal ganglia activation may be related to selection of lexical alternatives or to a more general action control function [5]. These findings are in line with views that emphasize the role of the left frontal and subcortical areas in the translation process. References 1. Klein, D., Milner, B., Zatorre, R.J., Meuer, E. & Evans, A.C. Proc. Natl. Acad. Sci. USA 1995, 92:2899 2. Rinne, J.O., Tommola, J., Laine, M., Krause, B.J, Schmidt, D., Kaasinen, V., Teräs, M., Sipilä, H., Sunnari, M. Neurosci. Lett. 2000, 294:85 3. Price C., Green D.W. & von Studnitz R. Brain 1999, 122:2221 4. Fiez, J.A. Hum. Brain Mapping 1997, 5:79. 5. Crosson, B., Zawacki, T., Brinson, G., Lu, L. & Sadek, J.R. J. Neuroling. 1997, 10:277

e100

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Results of the translation - control contrast

e101

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 59 Functional MRI study of frequency and regularity effects in Chinese Lifei Ma 1 , Xiangjie Tan 1 , Xiaoyi Wang 1 , Wei Yu 2 , Zhaoqi Zhang 2 , Zhuangwei Xiao 3 , Xuchu Weng 1 1 Laboratory for Higher Brain Function, Institute of psychology, Chinese Academy of Sciences, Beijing 100101, China, 2 Department of Radiology, Anzhen Hospital, Beijing Capital Medical University, Beijing, China, 3 Molecule Imaging Research Center, Medical College of Shantou University, Shantou 515041, China INTRODUCTION A number of neuroimaging studies have been conducted to investigate the frequency and regularity effects in reading alphabetic languages, such as English (1), whereas very few studies have been reported concerning Chinese, a logographic language. Chinese script differs markedly from alphabetic script in that its characters have square configuration and its orthography maps onto morphemes (meaning) rather than to phonemes. Therefore, one expects to find its orthography-phonology conversion process to be different from that of alphabetic languages. For this purpose, we studied the neural correlates for reading Chinese characters. METHODS Eight college students participated in the present study. MRI scans were conducted on a 1.5T Siemens Sonata scanner. Four EPI sequences (TR=2s, TE=60ms, 20 6mm axial slices with 1.5mm gap) were acquired while subjects silently read 4 types of Chinese characters (40 for each type): high frequency regular characters (1000-2500 per million, phonetic component and the whole character having the same sound), low frequency irregular characters (100-30 per million, phonetic component and the whole character having different sounds), high frequency irregular characters, and low frequency regular characters. Deconvolution analysis based on multiple linear regression were used to estimate hemodynamic responses and to generate activation maps for each condition. RESULTS As seen in Figure 1, a large distributed brain regions showed activation in reading Chinese characters, involving left superior temporal lobe, left middle frontal lobe, left inferior frontal lobe, supplementary motor area(SMA), bilateral posterior parietal lobe, right insula lobe, right middle frontal lobe and anterior and posterior cingulate cortex. Among them, left superior temporal lobe, SMA, left inferior frontal lobe and left middle frontal lobe showed frequency effect, and bilateral posterior parietal lobe, right insula lobe and right middle frontal lobe showed regularity effect. Frequency and regularity interaction was found in insula and cingulate cortex. CONCLUSIONS Most brain regions identified in the present study are similar to those in studies with English. However, it appears that some regions, including left middle frontal and bilateral posterior parietal cortices, were activated only during reading Chinese characters. These regions are suggested to be engaged in attentional and visual spatial processing, possibly related to the unique orthography of Chinese characters (2). More important, we found that the frequency effect, the regularity effect and their interaction seem to have different neural substrates across Chinese and alphabetic languages. References 1. Fiez JA et al. (1999) Neuron 24: 205-218. 2. Tan LH et al. (2001) NeuroReport 12: 83-87. This research was supported by the National Natural Science Foundation of China (No. 30000054, 30170325), the Chinese Ministry of Science and Technology (No. G1999054000).

e102

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Activation maps for different types of Chinese characters

e103

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 60 Language-related asymmetries of the arcuate fasciculus in right- and left- handers: a DT-MRI and fMRI combined study Roberto Martuzzi 1,2 , Leila Cammoun 2 , Patric Hagmann 2 , Philippe Maeder 1 , Stephanie Clarke 3 , Jean-Philippe Thiran 2 , Reto Meuli 1 1 Department of Diagnostic and Interventional Radiology, University Hospital, Lausanne, SWITZERLAND, 2 Swiss Federal Institute of Technology, Signal Processing Institute, Lausanne, SWITZERLAND, 3 Department of Neuropsychology, Universitiy Hospital, Lausanne, SWITZERLAND Introduction We use DT-MRI and statistical fiber tracking to quantify the left-right asymmetry of connectivity between the posterior part of the superior temporal gyrus (Wernickes area) and the hom*olateral pars opercularis of the frontal inferior gyrus (Brocas area). Material and Methods We investigated 7 left-handers (LH) and 5 right-handers (RH), at 1.5 T. We got from each volunteer a DT-MRI dataset (b = 1000 s/mm 2 ), a high resolution MPRAGE, and a block-designed fMRI study with a silent word generation paradigm to detect Brocas area and a sentence recognition paradigm to detect Wernickes area. For RH, two box-shaped ROIs were centered on the fMRI activations corresponding to Wernickes and Brocas areas. Similar ROIs were placed symmetrically, according to anatomic landmarks. For LH, ROIs were centered on the fMRI activations present bilaterally. In the fiber tracking technique, briefly, the fibers were grown by a random walk algorithm that propagates according to the local diffusive properties, giving a statistical estimate of brain connectivity (see [1] for a detailed description of the method). For each subject, the fiber tracking algorithm was applied twice, on the real DT-MRI data and on what we call a water brain. A subjects water brain is his DT-MRI data where all the tensors in the white matter have been set isotropic with some noise added. Fiber tracking in a water brain depends only by the brain geometry therefore it guaranties that the bilateral placement of ROIs is unbiased. For each subject and condition, the number of fibers connecting the two ipsilateral ROIs were estimated (Nleft and Nright) and the Normalized Left-Right Difference (NLRD), defined as: NLRD = (N left N right )/(N left + N right ), was computed. Results Hypotheses were evaluated by Student testing with a rejection level set to 0.05. First we determined whether the RH population show a significantly higher connectivity in the left than in the right hemisphere between the frontal and the temporal ROIs. This was done by testing the NLRD in real data versus water brain. The result showed a statistical difference in these two densities of connection (p = 0.0098). Similarly, we tested whether the LH population is lateralized to the right using an approach similar to the previous test. The result reached the limit of significance (p = 0.0522), suggesting that the LH population is less lateralized on the right. Finally, we compared the real data of the RH versus LH population, and we noticed that the language connectivity lateralization is significantly different between the two populations (p = 0.0019). Discussion Results show that in right-handers the association pathways between the superior temporal gyrus and the hom*o-lateral pars opercularis of the frontal inferior gyrus are significantly stronger left and that this asymmetry is fairly constant over that population. Left-handers seem to be more heterogeneous in terms of lateralization. References [1] Hagmann et al. Neuroimage, 2003; 19:545-54.

e104

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 61 The role of the right Wernicke in processing novel metaphorical expressions: Cortical correlations evaluated with principal component analysis and fMRI Nira Mashal 1,3 , Miriam Faust 1 , Talma Hendler 2,3 Leslie and Susan Gonda (Goldschmied) Multidisciplinary Brain Research,Bar-Ilan University, 2 Sackler Faculty of Medicine, Tel Aviv University, 3 Functional Brain Imaging Unit, Wohl Institute for Advanced Imaging, Tel Aviv Sourasky Medical Center

1 The

Some research indicates that the right hemisphere (RH) has a unique role in comprehending the figurative meaning of metaphors whereas the results of other studies do not support the notion of a selective role for the RH in accessing metaphorical meanings. The current research used fMRI technology to test a theoretical explanation of the above conflicting findings. This theoretical account is derived from the Graded Salience Hypothesis (Giora ,1997). According to the Graded Salience Hypothesis (GSH), the degree of meaning salience rather than literality or nonliterality primarily affects differences between the LH and RH in linguistic processing. Thus, the GSH predicts a selective RH involvement in comprehension of novel, nonsalient metaphoric meanings and LH involvement in the comprehension of conventional, salient metaphoric meanings. Fifteen normal adults participated in a block designed fMRI experiment that compared the patterns of brain activation induced by processing the meanings of literal, conventional metaphoric, novel metaphoric and unrelated word pairs. The subjects had to decide which kind of relation exists between the two words (literally related , metaphorically related or not related). The literal word pairs (i.e., broken vase) and the dead metaphors (i.e., bright student) represented the expressions with salient meanings whereas the novel metaphors, taken from poetry (i.e., crystal river) and the unrelated word pairs (i.e., junction laundry) represented the expressions with nonsalient meanings. We applied the principal component analysis (PCA) technique in order to find different functional networks (for definition see Horvitz, Tagamets, and McIntosh, 1999, page 92) corresponding to the different stimuli. In order to do so we generated fMRI time series segments from each subject at each region separately for each of the four conditions, such that each subject produced a 20x19 matrix. The columns of this matrix are the concatenated segments of the time series of each of the nineteen regions, and the rows are the twenty time points of each condition (for details see Honey et al., 2002, page 576). We averaged all the subjects matrices into one matrix for each condition and then decomposed those matrices by PCA. Our results, obtained from group analysis and from PCA of the fMRI data, indicate that the right Wernicke has a special role in processing novel metaphors. We suggest that a unique network, consisting of the right Wernicke, the right and left premotor areas, the right and left Insula and the left Broca, is recruited for the processing of novel metaphors but not for the processing of conventional metaphors. Giora, R. (1997). Understanding figurative and literal language: The graded salience hypothesis. Cognitive linguistics , 7, 183-206. B.Horwitz, M-A. Tagamets and A. R. McIntosh (1999). Neural modeling, functional brain imaging, and cognition. Trends in cognitive sciences, 3, 91- 98. G.D. Honey, C.H.Y. Fu, J. Kim, M.J. Brammer, T.J. Croudace, J.Suckling, E.M. Pich, S.C.R. Williams and E.T. Bullmore (2002). Effects of verbal working memory load on corticocortical connectivity modeled by path analysis of functional magnetic resonance imaging data. NeuroImage, 17, 573-582.

e105

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e106

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 62 Bilateral activation in the inferior frontal cortex during accent judgment: an fMRI study Kayako Matsuo 1,2 , Yuko Ohgami 1,3 , Keiichiro Toma 2 , Kenichi Oishi 4,5 , Chika Sumiyoshi 6 , Toshiharu Nakai 1,2 1 National Institute of Advanced Industrial Science and Technology, 2 Institute of Biomedical Research and Innovation, 3 Ochanomizu University, 4 Kobe University, 5 Kyoto University, 6 f*ckushima University Introduction Brain activation during accent judgment was examined using functional magnetic resonance imaging (fMRI). Accent patterns determine lexical meaning of Japanese words, one of pitch-accent languages. For example, /sa-ku-ra/ for cherry blossom is pronounced in a low pitch high pitch high pitch accent pattern in standard Japanese. The same phonetic sequence pronounced in a highlowlow accent pattern usually does not mean cherry blossom but a location name or personal name. Lexical knowledge is indispensable for the accent processing. Therefore, Brocas area in the left inferior frontal gyrus (IFG) might be involved in an accent judgment because this area is crucial for the propositional (lexical) aspect of language [1]. On the other hand, an accent judgment might also activate the right IFG because of a pitch processing of sounds [2] included in the accent patterns. Materials and Methods Ten native Japanese-speaking volunteers participated in this study (age 21-39, F/M = 5/5, all right-handed, all gave written informed consent). An fMRI series consisted of rest (R), target (T) and control (C) periods (RTCTCTCTCR, 30 sec each, total 5 min). Three experiments were conducted in separate fMRI series; accent (A), semantic (S) and phonological (P) judgments. The experiments S and P were for comparisons. During the target periods (T) of the experiment A, subjects judged whether Japanese words written in 3 syllabic characters had a high-low-low accent pattern. In the experiment P, subjects detected sounds including /e/ represented in the same 3-syllable words. In the experiment S, subjects judged whether the words meant food or not. In the control periods (C), subjects detected $$$ among other nonsense strings. All imaging was performed on a 3T MR scanner (GE Signa VH/i 3.0T). A spiral sequence [3] was used for the functional studies (TE/TR/FA = 30/3000/15, FOV = 220 mm, 30 axial slices, 4 mm thick interleaved). The fMRI data were analyzed using SPM2. Activation maps were generated for the T > C contrast in each experiment using a random-effect group analysis (p = 0.001, uncorrected). Results and Discussion As shown in Fig. 1, all experiments activated Brocas area (BA 44/45). The accent judgment specifically activated the right IFG extensively. The activation in the right IFG suggested a pitch processing included in the accent judgment. These results are consistent with a previous study concerning pitch judgments of perceived and imagined songs [2] and traditional neuropsychological reports of prosodic disorders [1,4]. Predominant activation in the left IFG during the accent judgment suggested the propositional or lexical processing combined with accent patterns. A discrimination of Chinese words, a tonal language, also predominantly activated the left hemisphere in Chinese speakers [5]. Conclusions Our data demonstrated bilateral involvement of IFG in accent judgment of Japanese, one of pitch-accent languages. Our findings provide evidence of a right hemisphere specialization for pitch processing and a left hemisphere specialization for lexical processing. References [1] Ross, E. D., Arch Neurol, 38, 561-569, 1981 [2] Zattore, R.J. et al., J Cognit Neurosci, 8, 29-46, 1996 [3] Glover, G.H. et al., Magn Reson Med, 3, 361-368,1998 [4] Monrad-Krohn, G.H., Brain, 70, 405-415, 1947 [5] Klein, D. et al., NeuroImage, 13, 646-653,2001

e107

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Rendered activation maps and a coronal section including Broca’s area

e108

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 63 Is it a Wird? Is it a Plane? fMRI of the Orthographic Neighbourhood Effect. Adam J McNamara 1 , Michal Lavidor 2 , Martin Meyer 3 , Padraic Monaghan 4 , Richard C Shillco*ck 5 , Enrico Simonotto 6 , Andrew W Ellis 4 1 Department of Neurology, University Hospital Schleswig-Holstein, Germany., 2 Department of Psychology, University of Hull, UK., 3 Department of Neuropsychology, University of Zurich, Switzerland., 4 Department of Psychology, University of York, UK., 5 Department of Psychology, University of Edinburgh, UK., 6 MRI Devices, Waukesha, WI, USA. Introduction: If an English word can be changed to another legal English word by changing just one letter, i.e. ’cover’ and ’lover’, then these words are described as orthographic neighbours. The number of neighbours denotes the N value of a word and is a useful tool for studying the underlying processes of word recognition. The Orthographic Neighbourhood (N) effect describes the facilitation of the recognition of English words that are orthographically similar to many others, but inhibited recognition of pseudo words that are orthographically similar to many real English words. The N effect has been demonstrated to occur when words are presented in the left visual field (LVF) but not the right, (RVF), (Lavidor and Ellis 2002). We used a divided visual field paradigm presenting words and pseudo words of high and low N and have shown that regions of both the fusiform gyri and hippocampus are sensitive to N. Methods: Whilst scanning, twelve male, right handed participants undertook a lexical decision task identifying presented stimuli as either ’English word’, or ’not a word’ conveying their response by button press, (counterbalanced). Stimuli were displayed for 150ms each, pseudo-randomly in the left or right visual field. Inter-trial interval was jittered, (mean 2929ms) and a centred ’+’ displayed. Scanning was conducted with a 1.5T GE scanner at the Western General Hospital, Edinburgh, (T1-weighted structural scan, and two 581sec T2*-weighted functional scans, EPI, matrix 64x64, FOV 19.2cm, TR=1.5s, TE=40ms, 14 slices, interleaved acquisition). Pre-processing: (slice time and motion artifact correction, alignment with structural scans, spatial normalization to MNI template, resampling to 2mm 3 voxels, and 6mm FWHM smoothing). GLM based fixed-effects event-related analysis was performed using SPM99. The design matrix coded five conditions (’High-N Pseudo Words (LVF)’, ’High-N Pseudo Words (RVF)’, ’Low-N Pseudo Words (LVF)’, ’Low-N Pseudo Words (RVF)’, and ’All Words’) using canonical HRFs (plus movement regressors). T-maps were constructed using a minimum cluster size of 20 voxels. Only the effects of pseudo words are considered here. Results: Figure 1: The main effect of pseudo words (threshold P = 0.05 corrected), yielded robust left hemisphere activation of the left fusiform gyrus, and left inferior frontal gyrus BA44 and BA47. Activation in the right hemisphere was noted in the right inferior frontal gyrus BA45 and the cerebellum. Figure 2: Significant increases, (P<0.001 uncorrected) in activity were found between high-N and low-N pseudo-words in the splenium, (red blobs). This contrast limited to LVF stimuli revealed bilateral anterior fusiform activity (yellow blobs), and Low-N Pseudo Words > High-N Pseudo Words (RVF) revealed left hippocampus activation, (green blobs). Conclusions: Although activation was mainly lateralized to the left hemisphere no N based effects were found, supporting the results of Binder, (2003). However when stimuli were presented in the left visual field then N effects correlated with BOLD signal were elicited. We suggest this may be due to increased relative load on the non practiced hemisphere. Bilateral activation clusters at the nexus of the parahippocampal gyri with the corpus callosum splenium support this view of trans-hemispheric activity and suggests it is greater for high N pseudo words than low N pseudo words. References: 1. Lavidor, .M., Ellis, A. W. (2002) Brain. Lang. 80: 63-76. 2. Binder .J.R et al (2003), J, Cogn. Neurosci. 15:372-393

e109

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e110

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 64 Double dissociation for irregular and pseudo word reading in semantic and phonological frontal areas Andrea Mechelli 1 , Steven Long 1 , Jenny Crinion 1 , Matt Lambon-Ralph 2 , Karalyn Patterson 3 , Jay McClelland 4 , Cathy Price 1 1 FIL, 12 Queen Square, London, WC1N 3BG (UK), 2 Department of Psychology, Oxford Road, Manchester, M13 9PL (UK), 3 MRC Cognition and Brain Sciences Unit, 15 Chaucer Road, Cambridge, CB2 2EF (UK), 4 CNBC, 115 Mellon Institute, 4400 Fifth Avenue Pittsburgh, PA 15213-2683 (USA) Cognitive models of reading typically envisage multiple mechanisms for translating orthography to phonology. For instance, pseudo words (e.g. ledan) rely on direct links between orthography and phonology; words with irregular spellings (e.g. yacht) rely on an indirect link from orthography to phonology via semantics. Finally words with regular spellings (e.g. lamp) can be read by both direct and indirect links between orthography and phonology 1 . The aim of the present study was to dissociate these different mechanisms at the neuronal level, using fMRI. A Siemens 1.5T scanner was used to acquire 360 T2*-weighted echoplanar images, with a TR of 3.15 seconds, from 22 healthy right-handed subjects. We presented each subject with blocks of regular, irregular or pseudo words, alternating with blocks false font strings. Regular and irregular words were matched for imageability, familiarity and frequency 2 . Regular, irregular and pseudo words were also matched for number of letters, number of syllables and bigram frequency. Stimuli of the same type were presented in blocks of 21 seconds, with a duration of 750 milliseconds and an onset asynchrony of 3 seconds. Subjects were instructed to read the words and pseudo words silently as soon as they appeared on the screen, and to look at the strings of false fonts. An eye-tracker was used to monitor the eye-movements of the subjects and ensure that they attended to the stimuli. Data were analyzed with SPM2 using standard procedures. We analyzed each subject independently and then performed a second level ANOVA to make inferences at the population level (p<0.05 corrected for multiple comparisons). Reading words and pseudo words relative to viewing false fonts increased activity in a distributed network including the left inferior prefrontal cortex, the superior temporal cortex and the anterior fusiform gyrus consistent with previous studies. Three distinct regions, within the left inferior prefrontal cortex, expressed differential patterns of activation. The pars triangularis showed increased activation for irregular relative to pseudo words (x=-52 y=32 z= 4; Z-score=5.4); this region is therefore more responsive to reading via semantics than via direct links between orthography and phonology. The reverse effect was found in the dorsal premotor cortex, which showed increased activation for pseudo relative to irregular words (x=-56 y=-8 z=38; Z-score=5.6); this region is therefore more responsive to reading via direct links between orthography and phonology than via semantics. Finally, the pars opercularis showed increased activation for irregular relative to regular words and for pseudo relative to regular words (x=-52 y=6 z=22; Z-score=6) consistent with a previous report that this region is sensitive to reading difficulty 3 . The double dissociation between irregular and pseudo words, which activated the pars triangularis and the dorsal premotor cortex respectively, parallels the double dissociation between semantic and phonological processing previously reported in the very same areas 4,5,6 . References 1) Plaut et al. Psych Review (1996). 2) MRC Psycholinguistic Database (www.psy.uwa.edu.au). 3) Feiz et al. (1999) Neuron 24, 205-218. 4) Mummery et al. (1999) JOCN 10(6), 766-777. 5) Roskies et al. (2001) JOCN 13, 829-843. 6) Devlin et al. (2003) JOCN 15(1), 71-84.

e111

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 65 Dissociating Reading and Picture Naming with Repetition Priming Andrea Mechelli , Cathy Price FIL, 12 Queen Square, London, WC1N 3BG (UK) Previous neuroimaging studies have identified a common neuronal network for word reading and picture naming. The aim of the present investigation is to test whether, within this common system, different neuronal populations activate during word reading and picture naming. This might be achieved using repetition priming: when two similar stimuli are presented within a short interval, reduced neuronal response is observed for the second (target) relative to the first (prime). Neuronal populations involved in reading but not picture naming might therefore be primed with words (but not pictures). Conversely, neuronal populations involved in picture naming but not reading might be primed with pictures (but not words). We presented each subject with pairs of stimuli, i.e. a prime and a target. Each stimulus was either a word or a picture, resulting in 4 types of pairs: word-word, picture-picture, word-picture and picture-word. In addition, the prime and target referred to items which were identical (e.g. dog-DOG) or unrelated (e.g. RADIO-glove), resulting in a fully factorial design. Perceptual priming was minimised by using different fonts and letter cases and presenting pictures of identical objects in different views. The prime was presented for 600 milliseconds, followed by a fixation cross for 200 milliseconds and the target for another 600 milliseconds. Pairs of stimuli were presented in an event-related fashion, with a stimulus onset asynchrony of 3200 milliseconds, and intermixed with null events. Subjects were instructed to read/name the words/pictures overtly as soon as they appeared on the screen, and to look at the fixation cross during the null events. A Siemens 3T scanner was used to acquire 814 T2*-weighted echoplanar images, with a TR of 2.275 seconds, from 16 healthy right-handed subjects. During the scanning, a microphone was used to record vocal responses and the statistical analysis was performed on correct trials only, using SPM2 software. Each subject was analyzed independently and then a second level ANOVA was used to make inferences at the population level (p<0.05 corrected for multiple comparisons). Preliminary results from 12 subjects indicate that word reading and picture naming (relative to fixation) activated the same neuronal system which included the bilateral premotor cortex, the inferior prefrontal cortex, the superior temporal cortex, and the occipito-temporal cortex - consistent with previous studies of word and picture processing. Priming effects (i.e. reduced response when the prime and target are identical relative to when they are unrelated) were detected for both naming and reading in the left medial extra-striate cortex (x= -10 y= -60 z=0). In addition, priming effects were detected during word reading in a left posterior middle temporal region (x= -62 y= -38 z=2); and during picture naming in a set of right lateralized motor regions including the SMA (x=2 y= -6 z=58), the anterior cingulate (x=2 y=22 z=30) and the cerebellum (x=8 y= -74 z= -22). All these effects were observed irrespective of whether the prime was a word or a picture. In summary, we found priming effects for reading and naming in a common perceptual area, and in different sets of speech production areas.

e112

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 66 Common cortical representations of mirror neurons and sign language in deaf signers Ingo G Meister 1,2 , Katrin Weier 1,2 , Walter Huber 2,3 , Babak Boroojerdi 1,2 1 Dept of Neurology, University Hospital Aachen, Germany, 2 Interdisciplinary Center CNS, University Hospital Aachen, Germany, 3 Section Neurolinguistics, University Hospital Aachen, Germany Mirror neurons denote a cortical neuron population which is active both in the execution and the observation of movements. They are supposed to be involved in imitation and learning by imitation. Furthermore, it is suggested that they played a key role during the evolution of language. According to this theory, the observation and execution of subject/object interactions (action observation) could be an early form of communication. If so, there should be overlapping cortical representations of the language and mirror neuron network. The present study investigated sign language, a natural language which shares both manual output and visual input with action observation. We studied perception of action observation and sign language in deaf and hearing adults to disentangle common and distributed cortical networks for action observation and language processing. Using functional magnetic resonance imaging (fMRI), cortical areas which mediate sign language and regions which are active during the observation of subject-object-interactions were investigated in deaf signers and normal subjects which were naive to sign language. Both experimental conditions require the observation of a person performing hand actions within space but linguistic processing was solely required by deaf subjects perceiving the sign lexems. We showed videos of sign language lexems which are non-iconic, i.e. the hearing control subjects could not assign a certain meaning to the signing movements, and videos of hand-object interactions like grasping a cup. The task was to look carefully at the videos and report after the session what could be seen. fMRI analysis was done using SPM2. In both groups and both tasks visual and parietal cortical areas were activated bilaterally due to the spatial processing of the stimuli. In deaf subjects compared to hearing subjects, a robust activation on the temporo-parietal junction and the adjacent posterior temporal lobe was found additionally both for naming of sign language lexems and observation of subject/object interactions in both hemispheres (Fig.1). Furthermore, we found a left hemispheric inferior parietal activation for sign language processing and a right hemispheric inferior parietal activation for action observation in deaf signers compared to hearing subjects. The present study demonstrated that the cortical networks for action perception and sign language processing overlap to a great extent in deaf signers. Interestingly, comparison with hearing sign language naive control subjects revealed a language-specific activation of the left inferior parietal cortex. The finding that the temporo-parietal junction and the adjacent posterior temporal cortex region is essential for the processing of sign language is consistent with prior studies. Furthermore, this region is involved in the observation/execution matching system in deaf signers which may indicate that the cortical representation of sign language in deaf subjects is based in part on mirror neurons.

Fig. 1: Regions which were active in deaf but not in hearing subjects for action observation (right) and sign language processing (left). In the latter condition there was an additional activation of the left inferior parietal cortex (p<0.001, uncorrected).

e113

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 67 An fMRI study of reading of ancient writings Naoki Miura 1,2 , Jobu Watanabe 3,2 , Kazuki Iwata 3,2 , Yuko Sassa 3,2 , Jorge Riera 3,2 , Hideo Tsuchiya 4 , Makoto Takahashi 1 , Masaharu Kitamura 1 , Ryuta Kawashima 2 1 Graduate School of Engineering, Tohoku University, 2 NICHe, Tohoku University, 3 LBC Research Center, Tohoku University 21st Century Center of Excellence Program in Humanities, 4 The vice president of the Japan Promotion Society for the Spread of Education by ’KANJI’ Introduction It is difficult to read an ancient writing compared with reading modern writing. In order to understand ancient writings, we need the special vocabulary and grammatical techniques. Therefore, it is thought that the language processing within a brain when reading an ancient writings differs from the processing when reading modern writings. In this study, we investigated the brain activity involved during reading of Japanese ancient writing using functional magnetic resonance imaging (fMRI). Methods Twenty-three right-handed normal subjects participated in the study. All subjects were healthy and native speakers of Japanese. The brain activity during sentence reading was examined block design. In this study, a session consisted of two tasks: reading of Japanese ancient writings, and Japanese modern writings. In the case of reading of ancient writings, famous ancient writings that are carried by the textbook of a high school were used for the sentence stimuli. In the case of the reading of modern writings, sentences that translated the ancient writings into modern writings ware used for the sentence stimuli. In each reading task, one sentence was displayed on the screen. During the resting condition, the subjects were instructed to look at a fixation-cross. During the reading tasks, the subjects were instructed to read the sentences, considering a meaning without moving their mouth. Each reading condition lasted 40s, and the resting condition lasted 30s, and subject underwent five sessions during fMRI scanning. fMRI time series data were acquired by gradient-echo echo-planer MRI (42 slices, slice thickness = 2.2 mm, gap = 1.1 mm, TR = 5000 ms), on a 1.5T Siemens Magnetom Symphony scanner. Statistical analysis of fMRI time series data was carried out using SPM2. Results The contrast of reading of ancient writings vs. resting condition activated the bilateral premotor area, bilateral superior frontal gyrus, right inferior frontal gyrus, the left superior temporal gyrus, right intraparietal sulcus, left middle occipital gyrus, bilateral lingual gyrus, right middle temporal gyrus, left thalamus, and bilateral cerebellum. The contrast of reading of modern writings vs. resting condition activated the left premotor area, bilateral superior frontal gyrus, bilateral middle frontal gyrus, left inferior frontal gyrus, the left superior temporal gyrus, left inferior occipital gyrus, right lingual gyrus, left inferior temporal gyrus, bilateral middle temporal gyrus, bilateral thalamus, and right cerebellum. The contrast of reading of ancient writings vs. modern writings activated the left premotor area, left inferior frontal gyrus, and the left superior temporal gyrus. The contrast of reading of modern writings vs. ancient writings activated the left middle occipital gyrus, and the bilateral lingual gyrus. Conclusion When reading ancient writings, the activity of left inferior frontal gyrus and the left superior temporal gyrus that related to syntactic and semantic processing was increased. Therefore, it is suggested that the ancient writing was read considering the structure of a sentence rather than modern writing.

e114

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 68 Neurocognitive aspects of bilingualism: the case of interpreters C. Momaur 1 , P. Péran 1 , J-L. Nespoulous 2 , J-F. Démonet 1 , D. Cardebat 1 1 Inserm U455, Toulouse, France, 2 Laboratoire Jacques Lordat, Toulouse, France Introduction Several neuroimaging studies devoted to bilingualism have shown the preponderant impact of proficiency on language representations in the brain 1 . This study aimed at comparing brain activation patterns in two types of subjects mastering two languages, bilinguals and interpreters. Based on previous literature on bilingualism we hypothesized that lexical search in interpreters would be associated with less brain activation than in bilinguals as interpreters would resort to more automatized and effortless procedures. Methods Fourteen proficient French-English bilinguals (8 bilinguals and 6 interpreters) were scanned while generating words in French (L1), English (L2) or alterning L1/L2. All subjects were right-handed, had French as mother tongue, and used French and English daily. The subjects were instructed to generate verbs upon presentation of heard nouns; languages were alterned during the runs. The fMRI block-design paradigm included 4 runs interleaved with resting blocks: 2 runs hom*ogeneous for stimulus and response language (L1/L1, L2/L2) and 2 switching runs (L1/L2, L2/L1). Functional data were acquired using a 1.5 T imager and data were analysed using SPM2. Threshold was set at P<0.001 (cluster extent of 20 voxels) using a random-effect analysis. Results All subjects performed the tasks at ceiling. hom*ogeneous conditions: In French, the interpreters-bilinguals contrast showed a large, mainly left-sided pattern involving the posterior inferior frontal, inferior parietal, superior temporal and basal ganglia regions. The reverse contrast did not show consistent activation. In English, the bilinguals-interpreters contrast showed higher activation in the left BA4 and SMA while the interpreters-bilinguals contrast showed increased activity in the left head of caudate and BA44. Switching conditions: In French to English, interpreters bilinguals contrast showed mainly activation in BA 6 bilaterally. The reverse contrast did not reveal consistent activation. In English to French, group comparisons showed contrasted effects since the interpreters bilinguals contrast showed mainly a right-sided pattern involving the premotor and the superior temporal regions, associated with activation in the thalami. The bilinguals-interpreters contrast showed a left-sided pattern involving BAs 40, 4, 7, 45, and the SMA, associated with the medial extra-striate cortex bilaterally. Discussion The above mentioned hypothesis of less activation in interpreters relative to bilinguals is not supported by these results when considering experimental conditions implicating stimuli in mother tongue. In spite of bilingualism, these results show that L1 and L2 are not equivalent in terms of neural substrates as more activity was seen in both groups when source language is not the mother tongue, especially in switching conditions. Moreover, patterns differed in lateralization between groups, with involvement of the right hemisphere in interpreters and of the left hemisphere in bilinguals. This hemispheric balance would may reflect group difference in terms of language pragmatics, skilled interpreters using L2 in a conscious, metalinguistic way, while bilinguals managed the task as they process both languages, rather unconsciously, in daily life. 1 Pallier, C et al. (2003). Imagerie cérébrale du bilinguisme. In N. Tzourio & O. Etard (Eds) Bases Neurales du langage. Traité de Sciences Cognitives. Hermès Science, Paris.

e115

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 69 An event-related fMRI study of task comparisons for word identification in reading Dina L. Moore 1 , Stephen J. Frost 1 , W. Einar Mencl 1, 2 , Rebecca Sandak 1 , Stephanie A. Mason 2 , Jay G. Rueckl 1,3 , Leonard Katz 1,3 , Kenneth R. Pugh 1,2 1 Haskins Laboratories, 2 Yale University School of Medicine, 3 University of Connecticut Introduction: Previous neuroimaging studies have identified a set of left hemisphere areas critical for word identification. This reading circuit includes ventral (occipitotemporal), dorsal (temporoparietal) and anterior (inferior frontal gyrus) components. Developmental studies suggest that with increased reading skill, word identification becomes increasingly reliant on the ventral system relative to dorsal and anterior sites. Moreover, studies of skilled adult readers suggest that dorsal and anterior areas play a stronger role in coding unfamiliar stimuli, whereas the ventral system appears to constitute a fast word-form system particularly well-suited to processing familiar stimuli. The present study was designed to further explore the roles of the major circuits involved in written word identification by contrasting the neurobiological response patterns associated with changing task demands. Reading tasks that place progressively greater processing and computational demands on the reader should differ with respect to the involvement of these cortical systems. In the present experiment, we used simple naming, go/no go naming (implicit lexical decision), and go/no go semantic naming (implicit semantic classification). Importantly, the tasks are matched on the nature of the stimuli presented to the participants (pronounceable letter strings) and on the nature of the response (pronouncing stimuli aloud), but vary in their demands on lexical and semantic processing. Subjects: Twenty (10 males, 10 females) native English speakers served as subjects. All reported normal or corrected-to-normal vision and no history of neurological impairment. Methods: In each task subjects were presented with a letter string and instructed to name it aloud. In simple naming, subjects were shown both words and pseudowords and were instructed to name each one aloud. In go/no go naming, subjects were shown both words and pseudowords (as in simple naming), but were instructed to name the stimulus aloud only if it was a real English word. In go/no go semantic naming, subjects were shown words that were from one of two categories, living objects and non-living objects, and were instructed to name the stimulus aloud only if it belonged in the non-living category. Using an event-related design, each of the three tasks were presented in a separate run with four runs of each task type. Results and Discussion: Analyses of reaction time data collected outside of the magnet in a behavioral pilot study [outside the magnet] revealed an average 100 msec increase in naming latency from simple naming to go/no go naming, and another 100 msec increase from go/no go naming to go/no go semantic naming, thus these tasks did make progressively greater language processing demands as predicted. Functional imaging analyses identified a set of reading-related areas that modulate with increased task demands. These findings allow a more precise specification of the nature of information processing within each circuit as well as their interactions in coping with progressive demands on component processes in reading.

e116

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 70 Architectonic mapping of human posterior superior temporal gyrus: Brodmann area 22 revisited Patricia Morosan 1 , Jörg Rademacher 1,2 , Nicola Palomero-Gallagher 1 , Axel Schleicher 3 , Hartmut Mohlberg 1 , Katrin Amunts 1 , Karl Zilles 1,3 1 Institute of Medicine, Research Center Jülich, Jülich, Germany, 2 CEA Service Hospitalier Frédéric Joliot, Orsay, France, 3 C.&O. Vogt Institute of Brain Research, University of Düsseldorf, Düsseldorf, Germany It is generally accepted that Brodmann area 22 (BA22) occupies a major part of human posterior superior temporal gyrus, but little is known about the exact position and extent of this huge cortical area. Such knowledge, however, is relevant for the analysis of structure-functional relationship, e.g., with respect to semantic processing and the identification of morphologic information of sentences (1). Here, we present individual and probabilistic maps of a superior temporal area (area Te3), which for the most part corresponds to the posterior portion of BA22. In contrast to BA22, area Te3 does not occupy the depth of the superior temporal sulcus or the posterior bulge of the middle temporal gyrus. Cytoarchitectonically, Te3 was reliably defined in ten normal adult brains by using an observer-independent method for localization of areal borders (2). These borders were validated by combined receptorarchitectonic studies, which revealed that the cytoarchitectonically defined borders of Te3 closely matched abrupt changes in the laminar distribution pattern of several receptors from different neurotransmitter systems, such as those for noradrenaline, GABA, glutamate, serotonine and acetylcholine (3). Individual maps of Te3 show a high degree of variability among subjects and hemispheres, e.g., right Te3 is located more anteriorly than left Te3 by up to 10 mm. To account for topological variability of Te3, a 3-D probabilistic map was created by superimposing all individual maps of Te3 to the spatially normalized, T1-weighted MR single-subject brain of the MNI (4). We conclude that this 3-D probabilistic map of Te3 represents a more appropriate neuroanatomical reference for functional imaging and clinical research settings than depictions of BA22 in stereotaxic atlases derived from a single hemisphere, e.g., the Talairach and Tournoux (1988) atlas (5). 1. Friederici & Kotz (2003) Neuroimage 20: 8-17. 2. Schleicher et al., (1999) Neuroimage 9: 165-177. 3. Zilles et al., (2002) Brain Mapping: The Methods. Elsevier, 573-602. 4. Evans et al., (1993) IEEE-NSS-MI Symposium, 1813-1817. 5. Talairach & Tournoux (1988) Coplanar Stereotaxic Atlas of the Human Brain. Thieme, Stuttgart. This Human Brain Project/Neuroinformatics research was funded jointly by the National Institute of Mental Health, of Neurological Disorders and Stroke, of Drug Abuse, and the National Cancer Center.

e117

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 71 fMRI study on the cerebral activation associated with the semantic processing in Korean and English spoken word production Kichun Nam 1 , Wichoi Choi 2 , Kyungduk Cho 3 of Psychology, Korea university, South Korea., 2 Department of Psychology, Korea university, South Korea., 3 Division of Humanities, Paichai University, Daejeon, South Korea.

1 Department

The present study was designed to examine the cerebral activation areas associated with the semantic processing in Korean and English spoken word production in Korean-English unbalanced bilingual speaker. The picture involving a word was presented and subjects were instructed to name the picture name covertly. The English distractor involved in the picture was presented at the around of the picture. There were five experimental conditions; (1) meaningless picture and subjects were asked to say "moo moo", (2) only picture without a word, (3) picture with a semantically unrelated word, (4) picture with a semantically related word, (5) picture with the picture name word. In the previous study(Choi et al, 2003) which used Korean word distractor, the overall cerebral activation of Korean naming condition is different from that of English one. In the condition used semantic related Korean distractor, there was cerebral activation in right frontal lobe in Korean naming, but no activation in the same region in English naming. In present study, the English word was used as the semantic distractor. The difference of cortical activation was shown between Korean and English naming. But the activation of the condition used semantic related English distractor was not observed in any conditions. That is, because the English proficiency of Korean-English unbalanced bilingual is very poor, the English distractor could not be processed automatically.

e118

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 72 Mental representation of Korean inflectional and derivational affixes: An fMRI study Kichun Nam 1 , Yu Mi Hwang 2 , Hyojeong Sohn 1 , Myung-Yoon Kang 2 1 Department of Psychology, Korea University, 2 Department of Linguistics, Korea University Background The present study was planned to investigate the cortical activation correlated with producing morphologically complex Korean morphemes by using fMRI technique. In this study three derivational affixes and one inflectional affixes were selected: pre-final ending for inflectional affix and passive, causative, nominal affix for derivational affix. Methods 12 healthy right-handed adults were participated in two Experiments. Each experiment was consisted of 2 sessions(12 sec dummy scan, 6 activations with 7 control blocks per session, total 2 sessions). Two control tasks were used in Experiment 1, ’vowel transformation task(C1)’ and in Experiment 2, ’seeing Arabic character(C2)’. Experimental stimuli were as follows: two kinds of pre-final ending(tense and honorific pre-final ending) for inflectional affix and three kinds of affix (causative, passive, and nominal affix) for derivational affix. The data acquisition was performed using an ISOL 3.0 Tesla forte MR scanner with EPI sequence(TR/TE = 3000/35ms, 5 mm no gap, 20 slices, 64x64 matrix, FOV=240mm, Flip angle 80). All data processing and statistical analysis were performed using SPM 99 and statistical significance level was p<0.001. Results In Experiment 1, there was a similar activation area between and . The activation of occipital lobe, inferior frontal gyrus, superior temporal gyrus was observed. The pattern of cortical activation between and was also quite similar. The areas including occipital lobe, inferior frontal gyrus, superior temporal gyrus, and precentral gyrus were activated. In Experiment 2, , and had the same activation areas: occipital lobe, temporal lobe, and supra parietal lobule. Conclusion The results of two Experiments showed that Pre-final ending had a quite similar activation pattern to derivational affixes.

e119

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 73 GRAMMATICAL, MORPHOLOGICAL, AND NARRATIVE STRUCTURE IN AMERICAN SIGN LANGUAGE: AN FMRI STUDY 1 Brain

Aaron J. Newman 1 , Ted Supalla 1 , Peter Hauser 2 , Elissa Newport 1 , Daphne Bavelier 1 & Cognitive Sciences, University of Rochester, Rochester, NY, USA, 2 Psychology, Rochester Institute of Technology, Rochester, NY, USA

A central question in neuroscience is the extent to which the neural organization of information processing is determined by the types of processing required, in contrast to the form in which the information is conveyed. Signed languages such as American Sign Language (ASL) present an opportunity to investigate this. Signed languages are natural human languages sharing all of the linguistic characteristics and complexity of spoken natural languages, such as grammatical structure. At the same time, they are conveyed through the visual-manual, as opposed to aural-oral, medium. To the extent that similar networks of brain regions are engaged in a particular linguistic process in signers and speakers, we can conclude that the demands of processing the underlying linguistic structure drive the brain organization. At the same time, the unique demands and capacities of each modality may lead to some differences in the brain areas recruited for their processing. Neuropsychological studies have indicated that ASL is dependent on the same left hemisphere (LH) brain areas as are spoken languages. In contrast, some but not all neuroimaging studies have suggested greater right hemisphere (RH) involvement in processing signed than spoken languages. However, as the materials used and linguistic complexity and structures tested have varied across studies the reasons for these differing results are unclear. In the present study we aimed to assess separately the effects of morphological, syntactic, and narrative structure on the brain activation in native ASL signers. 13 right handed congenitally deaf, native signers were scanned with BOLD fMRI while viewing ASL sentences with similar semantic content and three different types of linguistic marking: marking of grammatical information based solely on word order (SVO); grammatical information marked by inflectional morphology, including the syntactic use of space (IMS); and grammatical marking plus narrative/discourse structure (NAR). Blocks of sentences alternated with blocks in which the same sentences were played backwards, and three movie clips were overlaid simultaneously. This manipulation obliterated signers ability to comprehend the sentences while preserving the visual features of biological motion. Each block of three ASL sentences or backwards clips alternated with a block in which a still frame of the signer was viewed. Participants identified exemplars of semantic categories in ASL conditions and instances of bimanual symmetry in the backwards-overlaid conditions. All sentence types activated classical LH language areas, including Broca’s area, dorsolateral prefrontal cortex and the extent of the superior temporal sulcus (STS). Activation in RH hom*ologue areas was also noted, but was predominant only for NAR sentences - a result also found in spoken language comprehension studies. Backwards-overlaid stimuli also activated the STS bilaterally, including some overlap with the activation for ASL sentences. However, meaningful sentences uniquely activated the more anterior STS while incomprehensible biological motion activation was restricted to the posterior STS. Importantly, results indicated that syntactic use of space in a signed language does not obligatorily recruit the RH, as IMS sentences did not recruit the RH to a greater degree than SVO sentences.

e120

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 74 Are category-selective responses mediated by bottom-up or top-down mechanisms? Uta Noppeney , Karl Friston , Will Penny , Cathy Price Wellcome Department, Institute of Neurology, UCL, London Introduction The sensory-motor theory of semantics predicates category-selective responses on the functional organisation of the sensory-motor system. Tools enhance activation in a left lateralized visuo-motor system encompassing left middle temporal (LPMT), left supramarginal (SMG) and premotor (PM) areas. Activations within the fusiform gyrus have been found medially for tools and laterally for animals. This study segregates category-selective regions depending on whether their responses were modulated by stimulus modality or task context. Using effective connectivity analyses (Dynamic Causal Modelling, DCM), we then investigated whether the context-sensitive, category-selective activations can be mediated via bottom-up or top-down mechanisms Methods 22 normal right-handed volunteers participated in this fMRI study (Sonata 1.5 T scanner, GE-EPI, TE=50ms, 33 slices, TR=3 s). The 2X3X3 factorial design manipulated: category (animals, tools), stimulus modality (written words, spoken words, pictures) and task (one-back-tasks on (i) stimulus identity, (ii) action or (iii) real life size). The activation conditions were interleaved with 5.5 s fixation. GLM analysis: To allow for a random-effects analysis (SPM2), contrast images (single condition > fixation) for each subject were entered into a second level ANOVA, which modeled the 18 effects in our 2 x 3 x 3 design. We tested for the main effects of category and the interactions of category with stimulus modality and task context. Results (at p<0.05 corrected): Tools > Animals: fronto-temporo-parietal action system: (i) left posterior medial fusiform and middle occipital gyri for pictures > words; and (ii) LPMT, SMG, and prefrontal areas for explicit > implicit semantics (action and real life size > identity). Animals > Tools: right middle occipital and fusiform gyri for pictures > words. DCM analysis: For each subject, several DCM models were constructed to test whether (i) modality dependent tool-selective (L. medial post. fusiform & middle occipital gyrus) and animal-selective (R. middle occipital & lateral fusiform) responses were mediated by bottom up mechanisms and (ii) task dependent tool-selective responses (SMG & LPMT) by top-down mechanisms. The individual connection strengths were entered into a random effects analysis (t-test, p<0.05). Results: Increased connectivity (i) from occipital to left fusiform & occipito-temporal areas for tool pictures and to right fusiform & middle occipital areas for animal pictures irrespective of task; and (ii) from prefrontal to SMG and LPMT for tools during explicit semantic tasks irrespective of modality Conclusions Our results suggest two types of category-selectivity: In the first (SMG, LPMT), category-selectivity is task-dependent and emerges during explicit semantic tasks. According to our DCM analyses, this type is mediated via increased backward connections from the left prefrontal cortex. In the second (posterior fusiform, occipito-temporal), category-selectivity is modality dependent, increased for pictures and possibly mediated by increased forwards connections from early visual areas. Thus, from a cognitive perspective, category-selectivity lies in the interaction between task and stimulus-bound factors. In terms of neural mechanisms, it might emerge from context-sensitive modulation of both forward and backward connections.

e121

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e122

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 75 Reading irregular and regular English words in Italian subjects Chiara Nosarti 1 , Andrea Mechelli 2 , Cathy J Price 2 1 Division of Psychological Medicine, Institute of Psychiatry, De Crespigny Park, London SE5 8AZ (UK), 2 Wellcome Department of Imaging Neuroscience, Institute of Neurology, 12 Queen Square, London WC1N 3BG (UK) Reading is slower and more error prone when words have irregular (e.g. yacht) than regular (e.g. ferry) spelling-to-sound correspondences. This effect could arise because for irregularly spelled words there is conflict between semantic and non-semantic reading pathways. Likewise, pseudoword (e.g. yitch) reading is slower than regular word reading, but in this case the effect is attributed to the lack of contribution from semantics. In English subjects, we have demonstrated a double dissociation in frontal activation for irregular and pseudoword reading which illustrates the differential reliance that these word types place on semantic and non-semantic reading 1 . In the study reported below, we investigated whether this double dissociation would also be observed when Italian subjects read regular and irregular words in English. Participants were 17 healthy right-handed Italians (mean age 31 years) who learnt English at a mean age of 11 and lived in the UK at time of assessment. Using a Siemens 1.5T scanner, (3.15s TR), subjects were scanned while reading blocks of English regular and irregular words, their Italian translations and pseudo words. Each block lasted 21 seconds (3s SOA) and was alternated with strings of false fonts which the subjects were instructed to view. Words were matched for imageability, familiarity and frequency 2 , number of letters, syllables and bigram frequency. After scanning, all subjects read each English word again. Correct and incorrect responses were then used as a regressor in the statistical analysis of the imaging data (SPM2). Individual analysis of each subject identified the effect of reading each word type (correct responses only) relative to false fonts. A second level ANOVA was then used to make inferences at the population level (p<0.05 corrected for multiple comparisons). Reading Italian, English and pseudo words relative to false fonts activated a distributed left occipito-temporal and frontal network including premotor cortex and the SMA. In addition, reading English irregular words or pseudo-words relative to English regular or Italian words activated the right cerebellum (x= 28 y= -68 z= -28), left inferior frontal regions (x= -56 y= 8 z= 10; x= -48 y= 0 z= 36; x= -36 y= 30 z= -4; x= -46 y= 46 z= 0; x= 4 y= 14 z= 50), the SMA and anterior cingulate (x= -2 y= 2 z= 62; x= 4 y= 12 z= 54). There were no other effects and no evidence of a double dissociation in semantic and non-semantic reading (all regions were activated by pseudo and irregular words > regular words). The similarity between reading English irregular words and pseudo words that place contrasting demands on the reading system can be explained by a common demand on executive processing in fronto-cerebellar regions. This might arise in Italian subjects due to conflict between simultaneously activated phonological codes from semantic and phonological pathways in both English and Italian. References 1) Mechelli et al (2004). Abstract for Human Brain Mapping conference. 2) MRC Psycholinguistic Database (http://www.psy.uwa.edu.au).

e123

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 76 Possible role of the left anterior insula in articulation Yukiko Nota , Kyoshi Honda ATR Human Information Science Laboratories A lesion-based analysis by Dronkers (1996) has shown that the left anterior insula is associated to motor coordination of articulation. Recent brain imaging studies on normal subjects have reported both activation and non-activation of the insula during speech. In this study, we measured brain activity during speech production using fMRI to investigate task-depending activities of the insula. Methods 17 normal right-handed Japanese participated in the examination. In the speech condition the subjects produced CVCVCV syllables in a small voice following a visual cue presented every three seconds; The cues 1, 2, and 3 indicate utterance types katata, takata, and tataka, respectively. In the control condition they looked at a cue and judged which syllables were indicated without any articulation. Each subject performed two sessions; in a Random session, the three cues in both speech and control conditions were presented in random orders and in a Repeating session, only one cue was presented repeatedly. Subtraction results of speech condition minus control condition in Random session (RanS RanC), speech condition minus control condition in Repeat session (RepS RepC), and speech condition in Random session minus speech condition in Repeat session (RanS RepS) were assessed using SPM99. Results In the contrast of RepS RepC, the activated regions were the primary motor cortex (BA 4), premotor cortex (BA 6), auditory cortex, superior and middle temporal gyrus, putamen, thalamus, and cerebellum (H VI, VIIIA). In the contrast of RanS RanC, activated regions were the primary motor cortex (BA 4), premotor cortex (BA 6), inferior parietal lobule, posterior superior temporal gyrus, auditory cortex, the left anterior insula, putamen, thalamus, and cerebellum (VI, VIIIA). In the contrast of RanS RepS, activation was detected in the premotor area and supplementary motor area (BA6), left inferior parietal lobule, the left anterior insula, Brocas area, putamen, thalamus, and cerebellum (VI, VIIIA). Activation in the left anterior insula was detected in the contrast of RanS RepS and RanS RanC but not in RepS RepC (Fig. 1). Discussion According to Levelt (1989), execution of articulation consists of four phases: 1) assembling a motor program (stages of phonological encoding and storing phonetic plans in the articulatory buffer), 2) retrieving the motor programs from the buffer, 3) unpacking subprograms, and 4) executing motor commands. In the present study, articulatory movements performed in a session were the same across the two sessions. In the speech condition of Repeating session, phonetic plans are fed into the articulatory buffer only once at the first utterance in the session, while in Random session encoding is required at every utterance. We infer that the left anterior insula may be involved in phonological encoding and storing the phonetic plans in the articulatory buffer. References Dronkers, N. (1996) Nature 384, 159-161. Levelt, WJM. (1989) Speaking MIT press, Massachusetts, 413-457 Murphy, K. et al. (1997) Journal of Applied Physiology 83, 1438-1447. Riecker, A. et al. (2000) Brain and Language 75, 259-276. Wise, RJ. et al. (1999) Lancet 343, 1057-1061.

e124

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Activation of the left anterior insula was detected in the contrast of RanS - RanC (a) and RanS - RepS (c), but not in the contrast of RepS - RepC (b) (z = 6).

e125

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 77 Correlates of Speech Sound Extraction in Anterior Superior Temporal Cortex an auditory fMRI study Jonas Obleser 1 , Henning Boecker 2 , Alexander Drzezga 2 , Carsten Eulitz 1 , Josef P. Rauschecker 3 1 Departments of Psychology and Linguistics, University of Konstanz, 2 Department of Nuclear Medicine and Zentrum funktionelle Bildgebung, Technical University Munich., 3 Georgetown University, School of Medicine, Washington, DC Neurophysiological studies in nonhuman primates and functional imaging studies in humans suggest that auditory object processing is accomplished by neural networks within anterior superior temporal gyrus (aSTG). Auditory object processing includes the decoding of species-specific communication sounds, for instance speech sounds in humans. The present study investigated the functional anatomy of vowel processing using functional magnetic resonance imaging (fMRI; 1.5 T). We compared attentive auditory perception of natural German vowels to non-speech band-passed noise stimuli in 13 subjects. More specifically, we explored the cortical mapping of the vowels spectral properties in terms of first and second formants (F1, F2), which are closely linked to the phonological features tongue height and place of articulation. Acoustically variant exemplars of natural German vowels were aligned in cycles of vowel stimuli that either comprised changes along the first formant (e.g., [u]-[o], [i]-[e]) or along the second formant (e.g., [u]-[i],[o]-[e]). Vowel cycles lasted 2 s and alternated with 2.5-s clustered acquisition periods (TR=4500 ms, TA=2505 ms, TE=50 ms, flip angle 90°, 28 slices with 5mm thickness and 0.5 mm gap, field of view 64x64, 192 mm, pixel size 3x3 mm, transversal orientation parallel to AC-PC, slice order interleaved). Ninety-six volumes of each condition were submitted to statistical analysis. Before statistical inference, all images were slice-time corrected, realigned to the first image acquired, co-registered with subjects individual high-resolution anatomical scans and normalized using a Talairach-transformed template image. All images were smoothed using an 8x8x12 mm³ Gaussian kernel. In both random-effects and fixed-effects analyses, vowel cycles consistently elicited significant activation in the left anterior superior temporal (STG, BA 22) and the middle temporal gyrus (BA 21) compared to non-speech noise. Not as consistent across vowel conditions, activation was also found in the right anterior STG and the inferior frontal cortex. These findings strongly support the notion of an auditory what stream, with highly object-specialized areas anterior to primary auditory cortex and direct anatomical links with inferior frontal regions. In particular, our results add to the growing evidence that speech sounds, as one of the behaviorally most relevant classes of auditory objects, are analyzed and categorized in aSTG.

e126

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 78 Using functional Magnetic Resonance Imaging (fMRI) to Explore Brain Function: Cortical Representations of Language Critical Areas 1 dept

mohammad ali oghabian 1 , hooshang saberi 2 , ali mahdavi 3 , nader riahi 1 , alireza rezvanizadeh 3 of medical physics-tehran univ of medical science, 2 dept of neurosurgery-tehran univ of medical science, 3 tehran univ of medical science

Introduction Pre-operative determination of the dominant hemisphere for speech and speech associated sensory and motor regions has been of great interest for the neurological surgeons .This dilemma has been of at most importance ,but difficult to achieve ,requiring either invasive (Wada test) or non-invasive methods(Brain Mapping).In the present study we have employed functional Magnetic Resonance Imaging(fMRI) to observe and delineate regional brain activations during execution of language related tasks. Subjects and Methods Our healthy volunteers comprised of 10 right-handed males (handedness being ascertained by Snyder-Harris Handedness Inventory) all speaking Persian as their native language. All the subjects performed two consequent language tasks namely ; Word Generation and Reverse Word Reading. The visual stimulation was performed employing a video projector and the presentation software (Neurobehavioral Systems), while the brain activity was monitored and studied by fMRI . The stimuli were given during the activation period ,while the asterix had been used as the blank (rest period) .The subject response was internal speech . The fMRI method employed ,was Echo Planar Imaging (EPI) using FSL (FMRIB Software Library) as the analyzing software .The Hardware comprised of 1.5 Tesla GE brand MRI scanner. Results The brain regions involved language processing could be successfully and prominently activated while the aforementioned tasks(percent activity 1.2% and P<0.005) .in all ten subjects being examined these regions were exclusively located in the left hemisphere corresponding to Broca ,Wernicke and exner areas . Discussion These promising results may be of value to determine the dominant hemisphere by a non-invasive method , or as an adjunct to conventional methods of Electrocortical Mapping and Wada test .Besides ,the presented visual tasks could specifically activate the traditional language centers even those known to be involved in writing (Exner area).Estimation of the exact sensitivity and specificity of the methods ,however requires employing a gold Standard methods and larger subject populations.

e127

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 79 Visual word form recognition in the posterior inferior temporal cortex. a MEG study using symbols and Japanese charactersKatsuya Ohta 1,4 , Yasuhiro Shirahama 2 , Atsuko Takashima 3 , Eisuke Matsushima 4 , Yoshiro Okubo 5 Hospital, 302 Kanegasaku, Matsudo, Chiba 270-2251, Japan, 2 Sangenjaya Hospital, 1-21-5 Sangenjaya, Setagaya-ku, Tokyo 154-0024, Japan, 3 F.C. Donders Centre for Cognitive Neuroimaging,Trigon 181, P.O. Box 9101 NL-6500 HB Nijmegen, The Netherlands, 4 Section of Liaison Psychiatry and Palliative Medicine, Graduate School of Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8519, Japan, 5 Department of Neuropsychiatry, Nippon Medical School, 1-1-5 Sendagi, Bunkyo-ku, Tokyo 113-8603, Japan 1 Onda-daini

A standard model of word reading postulates that the posterior inferior temporal cortex is involved in the processing of written words. This processing probably occurs within 200 ms after stimulus presentation. In order to characterize this process more precisely, we conducted a MEG study during a reading task in nine right-handed healthy Japanese subjects (3 females and 6 males). The subjects were required to respond to a word pertaining to the human body (Figure 1) so that all stimuli would be subject to the same semantic processing. The trials for non-target conditions, such as kanji words, kana words, kana pseudowords and symbols were analyzed to avoid possible P300 effect. The symbols were recruited from the Symbol and Wingdings font of Microsoft Word. Recordings of event-related magnetic fields (ERFs) were carried using a Magnes 2500, 148-channel, whole-head system (Biomagnetic Technologies). The magnetic response peak of around 200 ms for symbols was smaller than any of the other three letter conditions (Figure 2 and Figure 3), but there were no differences between actual words and pseudowords or between kanji and kana. More than half of the sources of M200, which showed satisfactory dipole solutions on the left hemisphere, were localized in the vicinity of the fusiform gyrus (Figure 4). The location did not differ significantly between any two of the four conditions. No significant difference of RMS (Root mean square) for the M400 period was found out among four conditions whereas RMS for M400 was greater in the left hemisphere than in the right hemisphere under all conditions. One possible explanation for the finding that M200 for symbols was smaller than for any other condition is that the processing of words may activate the neural substrates that subserve visual word form recognition. Consistent with this perspective, the present study revealed greater activation in PIT during the processing of real words and pseudowords relative to symbols. An alternative explanation would be that at least a part of M200 is involved in phonological processing and that, in the case of symbols, little phonological processing occurs. However, recent studies using fMRI have revealed that the left inferior prefrontal cortex plays a critical role in phonological processing, inconsistent with lesion-deficit studies with neurological patients. Therefore the plausibility of the second interpretation seems very slight. This result may suggest that M200 reflects the word-specific process such as visual word form recognition. The reason for M400 being clearly observed not only by words but also by symbols would be that the trials for words or symbols that are not categorized as parts of the human body were analyzed. The results that symbols, which have no visual word form but have semantics, elicited similar M400 as the other word conditions while M200 was smaller suggested that M200 may reflect the prelexical process.

e128

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1. Examples of kanji words (one character), kana words (two characters), kana pseudowords (two characters) and symbols. Target stimuli are shown in the lower portion. The trials of these stimuli were excluded from the analysis.

Figure 2. Grand-averaged (n=9) event-related field (ERF) waveforms elicited by four non-target conditions (kanji words, kana words, kana pseudowords and symbols) during the semantic task recorded from 38 sensors in the temporo-parieto-occipital regions on each hemisphere. Three magnetic components can be detected (M150, M200 and M400).

e129

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 3. Grand-averaged root mean square ( RMS ) waveforms recorded from the whole head are shown separately for the kanji word condition (red), kana word condition ( yellow ), kana pseudoword condition ( green ) and symbol ( blue ) condition. The M200 component in the symbol condition is smaller and later than in the other three conditions.

Figure 4. White circle with a bar indicates one representative equivalent current dipole(ECD) source estimated during the M200 time window in the kanji word condition from one subject superimposed onto horizontal and coronal MRI scans. There was no difference among the four conditions.

e130

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 80 Comparison of EEG and fMRI for the detection of early word-related processes Hauk Olaf , Pulvermüller Friedemann , Johnsrude Ingrid MRC, Cognition and Brain Sciences Unit, Cambridge It cannot be assumed that all brain areas showing activation in fMRI data also contribute to the EEG/MEG signal, and vice versa. In particular, EEG and MEG are sensitive to very short-lived changes in electrical brain activity on a millisecond scale, possibly too short-lived to be detected by fMRI. We here compare word-evoked brain activity revealed by event-related fMRI with that obtained from multi-channel ERPs. The same 10 subjects participated in the fMRI and EEG experiments. They were presented with words and length-matched strings of hash marks, among other stimuli [1]. Two brain areas were predominantly activated in the fMRI experiment: One in left inferior frontal gyrus, the other in left fusiform gyrus ("LFG") (fig. 1). For each individual subject, the mean activation of the LFG cluster was extracted from the fMRI data, and the neural activity distributed over the whole brain surface was derived from 65-channel ERP recordings using minimum norm estimates [2]. The intensities of the sources were correlated across subjects with the LFG cluster activation. The time points 96ms, 126ms and 168ms correspond to peaks in the RMS time course of the ERP difference waveform (words-hashes), while 138ms lies in the onset slope of the peak at 168ms. Words produce stronger activity in left posterior areas at 96ms and 138ms (fig 2). The later activation focus appears superior-anterior to the earlier one. In both cases follow time points exhibiting stronger activation for strings of hash marks in posterior regions, at 126ms and 168ms, respectively. The earliest significant correlation between source estimates in left posterior areas with the LFG-cluster activation occurred at the peak at 138ms, accompanied by another cluster at left mid-temporal sites (fig 3). The early posterior-anterior shift of word-related activity in the ERP source estimates within ~40ms indicates distinct processes in different areas, similar to the findings of [3]. Though a direct matching of localization results between fMRI and EEG/MEG is problematic due to uncertainties both in image normalization and source estimation, the correlation of metabolic activity of the LFG cluster with electric source strengths in left posterior regions around 138ms hints at a common process for these two in early word processing. In general, the source estimates appear to reveal more sources and possibly more processes than the fMRI in posterior brain areas. In particular, the finding of an additional mid-temporal correlation focus may indicate the involvement of more areas at that processing stage than imaged by fMRI alone. The relationship between the left inferior frontal activation present in the fMRI with sources estimated from the ERP has yet to be determined. [1] Hauk O, Johnsrude I, Pulvermüller F (2004). Somatotopic representation of action words in human motor and premotor cortex. Neuron, 41. [2] Hauk O (2004). Keep it simple: A case for using classical minimum norm estimates in the analysis of EEG and MEG data. Neuroimage, in press. [3] Pulvermüller F, Shtyrov Y, Ilmoniemi R J (2003). Spatio-temporal patterns of neural language processing: an MEG study using Minimum-Norm Current Estimates. Neuroimage 20, 1020-1025.

e131

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig. 1: Surface rendering of fMRI activation for the contrast words>hash marks (p<0.001 uncorrected).

Fig. 2: Grand average across subjects of noise-normalised minimum norm estimates. Top row: left view. Bottom row: Right view. Red colour indicates more activation for words (W), blue colour more activation for hash marks (H).

e132

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig. 3: See figure 2.

Fig. 4: Distribution of correlation coefficients for the correlation across subjects between mean fMRI activation in the LFG cluster and neural source strengths estimated from ERPs at 138ms.

e133

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 81 Cortical dynamics of letter-string perception in beginning readers Tiina Parviainen 1 , Päivi Helenius 1 , Elina Poskiparta 2 , Pekka Niemi 2,3 , Riitta Salmelin 1 1 Brain Research Unit, Low Temperature Laboratory, Helsinki University of Technology, Espoo, Finland, 2 Department of Psychology, University of Turku, 3 Centre for Reading Research, Stavanger University College Introduction We have previously used MEG to characterize the spatiotemporal organization of brain activation during written word perception in adults [1,2,3]. The activation proceeded from visual feature analysis in the bilateral occipital cortex at about 100 ms to the analysis of letter-strings in the left occipitotemporal cortex at about 150 ms and finally to the analysis of semantic information within temporal cortices. In dyslexic subjects, the stage of visual feature analysis was normal but the letter-string-specific activation was absent or abnormal, and the later activation at about 400 ms for comprehension was delayed [2,4]. Adult brain organization for written word perception is a result of years of development and experience. Both success and failure in learning to read are manifested during early school years. Here, we studied the perception of written words in children with normal reading development . Methods We measured 18 first grade students (7-8 yrs, 10 females). Stimuli were words embedded in four different levels of Gaussian noise and symbol strings with no noise, presented randomly to the subject once every 2 seconds [3]. Subjects task was to detect animals among symbols in the target strings. The magnetic signals were measured with a 306-channel whole-head neuromagnetometer and the activated brain areas were modeled using equivalent current dipoles. To identify sources specifically activated for analyzing visual features or letter-strings, the time-behavior of each source was tested for differences in the strength or latency between noisiest words, noiseless words and symbols [3]. Sources showing significant differences were selected for further analysis. In addition to the MEG measurement, the subjects were tested for general linguistic and non-linguistic abilities, phonological awareness, reading skills and rapid naming. Results The children (16/18) showed significantly stronger responses to the noisiest than to the noiseless words in the occipital cortex bilaterally (Fig. 1A), with the mean strength and latency of the response for the noisiest words 25±4 nAm and 159±6 ms, respectively. Sources with significantly stronger responses to noiseless than to the noisiest words or to symbol strings were clustered in the occipitotemporal cortex, predominantly in the left hemisphere (Fig. 1B). This response was detected in 10/18 subjects, with the mean strength and latency for noiseless words 24±4 nAm and 269±10 ms. In 10/18 subjects we also found a later letter-string specific activation clustering in left (and right) temporal cortex (Fig. 1C and D). Of the subjects with an occipitotemporal source 80% had also a left temporal source. In the preliminary analysis the neural responses were not clearly correlated to the behavioral measures. Conclusion In children, the occipitotemporal letter-string-specific response was detected less frequently than in adults. However, when it was present, also the anterior letter-string source was found, resembling the activation sequence in adults. The latency of both visual feature and letter-string specific activation was clearly delayed in children. References 1. Salmelin et al. Ann Neurol 1996, 40:157. 2. Helenius et al. Cereb Cortex 1999, 9:476. 3. Tarkiainen et al. Brain 1999, 122:2119. 4. Helenius et al. JOCN 1999, 11:535.

e134

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Location and time-behavior of sources specifically activated by visual features (A) or letter-strings (B,C,D).

e135

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 82 RECONCILING IMAGING BASED NEUROLINGUISTICS WITH MONSIER BROCA Eraldo Paulesu 1 , Gilda Pesenti 1,2 , Tiziana Savio 3 , Emanuela Bricolo 1 , Giorgia Silani 1 , Choi Deblieck 1 , Nadia Colombo 4 , Marco Minella 5 , Vincenzo Branca 6 , Antonella Costa 6 , Margherita Zito 6 , Andrea Boghi 1,7 , Giuseppe Scialfa 4 , Gabriella Bottini 2,8 1 Dipartimento di Psicologia, Università degli Studi di Milano-Bicocca, Italy, 2 Laboratorio di Neuropsicologia, Azienda Ospedaliera Niguarda Ca Granda, Milano, Italy, 3 Dipartimento di Medicina Sperimentale, Università degli Studi di Genova, Italy, 4 Servizio di Neuroradiologia, Azienda Ospedaliera Niguarda Ca Granda, Milano, Italy, 5 Servizio di Fisica Sanitaria, Azienda Ospedaliera Niguarda Ca Granda, Milano, Italy, 6 IRCCS Policlinico di Milano, Milano, Italy, 7 Dipartimento di Neuroscienze, Università di Torino, Torino, Italy, 8 Dipartimento di Psicologia, Università di Pavia, Pavia, Italy Introduction:In the assessment of language lateralization, neuroimaging studies have repeatedly shown a larger involvement of the right hemisphere than one would have expected from lesion data even in strong right-handers. Observation of such redundancy creates heuristic problems in cognitive neuroscience and practical problems in the clinical use of fMRI. However, task by hemisphere interaction effects have been seldom assessed formally in imaging studies. In this paper we address this limitation: activation data collected in a multi-task fMRI language protocol were treated with two different kinds of random effect analyses: conjunctions of activation patterns and hemisphere by task interaction effects. By doing so, we show a substantial reconciliation between imaging data and lesion data. Methods: 15 healthy right handed volunteers (f/m ratio =7/8; age range: 22-32), underwent an fMRI study which included eight linguistic tasks: automatic speech, phonemic fluency, semantic fluency, rhyme detection for visually presented letters, word reading, auditory word comprehension, object naming, auditory sentence comprehension. EPI/BOLD fMRI scans were collected using the Marconi/Philips Infinion or Eclipse 1.5T scanner (TR3000; TE 60msec; Flip angle 90°). After image reconstruction, raw-data visualization and pre-processing were performed with MRICro, all subsequent data analyses were performed using SPM2b. fMRI scans were realigned, stereotactically normalised into the standard symmetrical MNI space using an EPI-MRI template. The data were analysed according to a random-effect model implemented in SPM2b. Hemispheric lateralization of activations was determined with a paired t-test analysis in which neurologically oriented contrast images were compared with the same images flipped to be radiologically oriented. The conjunction analysis was masked onto the individual simple effects thresholded at p≤ 0.05. Activation images were also imported in MRIcro and saved as ROIs. ROIs were then compared by looking at their intersection in stereotactic space. Results: The conjunction analysis of the task by hemisphere interaction effects revealed a common area of activation across all tasks in Brocas area (left BA 44; see figure below). Although a number of regions were activated in the right hemisphere in the various tasks in regions mirror to the language areas of the left hemisphere, none of such right hemispheric regions was more active than the left ones. Discussion:This data provide an explicit reconciliation between functional imaging data and lesion based neurolinguistics and indicate an avenue for the assessment of hemispheric specialization for pre-surgical fMRI. Our data also indicate that Brocas area is a suitable indicator of hemispheric specialization for language across a broad range of language tasks.

e136

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e137

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 83 Paroxetine-induced modulation of cortical activity in language processing. Patrice Péran 1 , Emmanuelle Cassol 2 , David Tombari 1 , Isabelle Loubinoux 1 , Jean François Démonet 1 , Dominique Cardebat 1 1 INSERM U455, Hopital Purpan, Toulouse (France), 2 Department of Diagnostic and Therapeutic NeuroRadiology, Hopital Purpan, Toulouse (France) Introduction Previous studies have shown that paroxetine, a selective serotonin reuptake inhibitor (SSRI) affects cerebral motor activity in healthy subjects in single-dose trial 1 . This study aimed at assessing the effects of paroxetine on activity of cortical area network implicated in language processing related to action using chronic treatment. Methods A double-blind, cross-over, randomized paradigm was used comparing one month of paroxetine treatment (20 mg /day), vs placebo, with 3 months of wash-out. 21 healthy subjects were randomized (11 men; mean age 57,2 ± 6,5 years) among which functional neuroimaging data from 12 subjects were analysed (reasons for excluding patients were claustrophobia, imager failure, unexpected brain lesions). FMRI images were acquired with a 1.5 T imager at the end of each treatment, while performing 3 tasks: to repeat aloud verbs, to generate aloud verbs, and to mentally simulate actions presented in a block design. The stimuli were the same for the 3 tasks, 24 human action verbs (e.g. to row, to hammer) with auditory presentation. For generation, subjects were instructed to produce one verb semantically related to the input verb. For mental simulation of action, subjects were asked to imagine themselves performing the motor act denoted by the heard verb, avoiding overt movement. Verbal fluency and noun generation were assessed off line. Neuroimaging data were analyzed using SPM2 and a random-effect model (threshold of P<.05, k> 40 voxels). Results We did not find any difference on behavioral performance. We assessed the drug effects i.e. Paroxetine - Placebo>0 (hereafter called hyperactivation), and Placebo Paroxetine>0 (hypoactivation) on the activation-rest contrast for each task. For verb generation, a large bilateral pattern of hypoactivation in the upper lateral frontal cortex was found. For mental simulation of action, hypoactivation concerned the left middle and inferior frontal and supramarginal gyri. By contrast, an hyperactivation was found in BA 6 bilaterally for the repetition task. Discussion These results suggest that a chronic treatment of paroxetine modulates differentially the neural substrates of language tasks in spite of shared cognitive components (i.e. identical input in the 3 tasks, verb output for repetition and generation, semantic of action for generation and imagination). The main difference among these tasks concern the cognitive load that is much lower in the repetition task, a rather automatic procedure, compared to generation and imagination that require subjects to access word meaning, explore semantic features, and select and generate either a word or an inner representation of action. Our results show that paroxetine induces a relative enhancement of cortical activity for a low-level language task, and a relative decrease of cortical activity for the two demanding tasks. The latter finding suggests that paroxetine could have a facilitatory impact on neural networks subserving higher-order language tasks and concur with results described by Furey et al. (1997) 2 who dealt with modulation of the neural correlates working memory by physostigmine. References 1- Loubinoux I. et al., NeuroImage 2002;15:26-36. 2- Furey ML et al., Science. 2000 290:2315-9.

e138

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 84 Metaphoric thought and idiom processing in Chinese Conrad Perry 1 , Zhen Jin 2 , Li Hai Tan 1 1 Joint Laboratory for Language and Cognitive Neuroscience. The University of Hong Kong, 2 Beijing 306 Hospital In this study, we used fMRI to investigate the neural mechanisms underlying the processing of two common types of Chinese idioms, semi-literal and metaphoric idioms. Semi-literal idioms have a partially overlapped literal and idiomatic meaning, and the literal meaning is also metaphorically congruent with the idiomatic meaning. Alternatively, the literal and idiomatic meaning of metaphoric idioms has no overlap, although the literal meaning is also metaphorically congruent with the idiomatic meaning. Compared to control sentences with only a literal meaning, the results showed that a) the literal meaning of semi-literal idioms was processed to a greater extent than the literal meaning of the metaphoric idioms; b) the extent that the idioms were processed like metaphors was very limited; and c) there appeared to be brain areas that were activated more when processing idioms compared to literal sentences. These areas included the thalamus, cerebellum, and the anterior cingulate/superior frontal gyrus. Based on previous findings, we hypothesized that activation in the thalamus may have been elicited due to the retrieval of idiomatic meaning from smaller linguistics units (i.e., words and sentence parts) and that activation in the cerebellum may have been elicited due to the discrimination of idiomatic from literal meaning. Alternatively, the anterior cingulate/superior frontal gyrus may have be activated due to the abstract nature of idiom processing. Together, these results show that both cortical and subcortical areas can be activated from learnt (i.e., idiomatic meaning) rather than inferred relationships in higher order language tasks. This research was supported by a grant from the Research Grants Council of the Hong Kong Government (HKU 3/02C) awarded to L.H. Tan. Conrad Perry is supported by a UDF grant from the University of Hong Kong

e139

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 85 Brain connections of language and actions: Evidence from multimodal imaging Friedemann Pulvermuller 1 , Olaf Hauk 1 , Yury Shtyrov 1 , Vadim V. Nikulin 2 , Risto J. Ilmoniemi 3,4 1 MRC Cognition and Brain Sciences Unit, Cambridge, UK, 2 Karolinska Institutet, Stockholm, Sweden, 3 BioMag Laboratory, Engeneering Centre, Helsinki University Cenral Hospital, Finland, 4 Nexstim Inc., Helsinki, Finland If words are learned in conjunction with actions, the neurons controlling the actions frequently fire together with the neurons involved in word processing. Hebbian correlation learning predicts that the action- and language-related neurons in motor/premotor cortex and in perisylvian language cortex become associated. This suggests that the neuronal ensembles that represent and process action words such as lick, pick and kick (which refer to face-, arm- and leg-actions, respectively) reflect the somatotopy of the motor and premotor cortex (Somatotopy-of-action-words (SAW) model, 1-3). FMRI: Subjects were instructed to move their tongue, index finger and foot. The same subjects silently read English face-, arm- and leg- words matched for word length, frequency and imageability. Event-related fMRI revealed somatotopic activation along motor and premotor cortex elicited by face-, arm- and leg-words. Arm-word and finger activation overlapped with each other, as did leg-word and foot activation. EEG: Subjects silently read matched English face-, arm- and leg- words. RMS of 64-channel EEG showed activation peaks around 150 and 220 ms after stimulus onset. At 210-230, there was a significant topographical difference between face- and leg words. Minimum Norm Estimates (4) showed stronger inferior frontal activation for face- than leg-words and the reverse for superior central areas. MEG: The Finnish words for eat (hotki) and kick (potki) were presented as deviant stimuli (p=.17) in an oddball design against the background of frequently occurring pseudowords sharing the first syllable with the deviants but ending in π. Minimum-Norm Estimates (5) calculated from 306 channel MEG signals for the magnetic Mismatch Negativity elicited by the word final syllable indicated similar superior temporal activation for the two words, but stronger inferior fronto-central activation for the face-word than for the leg-word (~170ms) and the reverse in superior central areas (~200ms). TMS was applied to a locus around C3/4 where the first dorsal interosseus muscle of the hand could be stimulated and to a more dorsal site close to Cz (leg area stimulation). In the LH, hand area TMS tended to produce faster arm- than leg-word responses in a lexical decision task and leg-word processing was speeded relative to arm-words when stimulation was to the leg area. Conclusions: Experiments using spoken and written words carried out with 4 neuroimaging methods (fMRI, EEG, MEG, TMS) and implementing 3 different tasks (lexical decision, silent reading, watching video while ignoring words) indicated that 3 semantic subtypes of action words (face-, arm- and leg-words) are related to different areas in the fronto-central cortex, consistent with the somatotopy of the motor and premotor cortex. This is support for the SAW Model (1-3). Data also show good correspondence between fMRI results and early local (~200ms) EEG/MEG sources and indicate that activation of meaning-related areas can influence word processing. References: 1 Pulvermüller, F., Härle, M., and Hummel, F. (2000). Neuroreport 11, 2789-2793. 2 Pulvermüller, F. (2001). Trends in Cognitive Sciences 5, 517-524. 3 Pulvermüller, F. (2003). The neuroscience of language (Cambridge, Cambridge University Press). 4 Hauk, O. (2004). Neuroimage, in press. 5 Uutela, K., Hamalainen, M., and Somersalo, E. (1999). Neuroimage 10, 173-180.

e140

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 86 An MEG correlate of word recognition Friedemann Pulvermuller 1 , Yury Shtyrov 1 , Risto J Ilmoniemi 2,3 , William Marslen-Wilson 1 1 MRC Cognition and Brain Sciences Unit, Cambridge, UK, 2 BioMag Laboratory, Engineering Centre, Helsinki University Central Hospital, Helsinki, Finland, 3 Nexstim Inc., Helsinki, Finland A fundamental challenge for the cognitive neuroscience of language is to capture the spatio-temporal patterns of brain activity that underlie critical functional components of the language comprehension process. We combine here psycholinguistic analysis, whole-head magnetoencephalography (MEG), the Mismatch Negativity (MMN) paradigm, and source localization techniques (Equivalent Current Dipole and L1-Minimum-Norm Current Estimates) to locate the process of spoken word recognition in space and time. Methods: MEG signals were recorded with a 306 MEG system (Vectorview, Elekta Neuromag, Helsinki) while subjects instructed to ignore acoustic stimuli watched a silent movie of their own choice while being presented with spoken words of Finnish. Words were presented in an oddball design, either as frequent standard stimuli (p = .17) or as rare deviant stimuli (p = .83). Two spoken words starting with the same consonant and vowels, but ending in different consonants, were submitted to psychophysical and psycholinguistic experiments to determine (a) when they became perceptually different and (b) at which exact point in time after their onset each of them could be recognized 1. Stimulus triggered averaged event related fields, ERFs, were calculated for each subject and each standard and deviant stimulus word. The magnetic Mismatch Negativity, MMNm, was calculated by subtracting the averaged standard ERF from the deviant ERF. Source localization was performed using equivalent current dipoles and minimum-norm current estimates 2. Results: The MMNm to words peaked 100-150 ms after the information in the acoustic input was sufficient for word recognition. At ~130 after the point in time when a word could first be recognized from the acoustic input, the left-superior-temporal generators of the MMNm showed their first activation peak, whereas those in left inferior-frontal areas did so with an additional delay of 22 ms 3. The latencies with which words were recognized corresponded to those of an MMN source in the left superior temporal cortex. There was a significant correlation (r = 0.66) of latency measures of word recognition in individual study participants with the latency of the activity peak of the superior temporal source. Conclusions: The word recognition point of different words and of individual subjects was reflected in the peak latency of the left superior-temporal source of the MMNm, thus providing a tentative brain correlate of word recognition. The results are used to draw careful conclusions on the cortical organization and activity dynamics of cortical memory traces for words 4. References: 1. Marslen-Wilson, W. D. Functional parallelism in spoken word-recognition. Cognition 25, 71-102 (1987). 2. Uutela, K., Hamalainen, M. & Somersalo, E. Visualization of magnetoencephalographic data using minimum current estimates. Neuroimage 10, 173-180 (1999). 3. Pulvermüller, F., Shtyrov, Y. & Ilmoniemi, R. J. Spatio-temporal patterns of neural language processing: an MEG study using Minimum-Norm Current Estimates. Neuroimage 20, 1020-1025 (2003). 4. Pulvermüller, F. The neuroscience of language (Cambridge University Press, Cambridge, 2003).

e141

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 87 Time course of word meaning processing as revealed by magnetoencephalography Friedemann Pulvermuller 1 , Yury Shtyrov 1 , Risto J. Ilmoniemi 2,3 1 MRC Cognition and Brain Sciences Unit, Cambridge, UK, 2 BioMag Laboratory, Engineering Centre, Helsinki University Central Hospital, Helsinki, Finland, 3 Nexstim Inc., Helsinki, Finland Associative learning theory predicts that word form and semantics are tightly interwoven in the cortex. Action words referring to leg movements (e.g., to kick) should therefore activate cortical areas dorsal to those activated by words referring to actions performed with the upper body parts (to eat) 1. In earlier studies, written action words activated different areas in the fronto-central cortex 2-4. Here, we look at spoken words and ask when the word-category-specific activity spreads toward fronto-central areas, and whether this differential activation is automatic and persists when subjects are instructed to ignore stimulus words and watch a silent video film. Methods: 16 right-handed native speakers of Finnish heard the Finnish words hotki (to eat) and potki (to kick) presented as rare (p=.17) deviant stimuli in trains of frequent standard stimuli (SOA=1.4s). Standard stimuli were pseudowords (hotpi, potpi). 100-250ms after the critical stimulus divergence point (onset of second syllable), deviant stimuli elicited the magnetic Mismatch Negativity (MMNm) 5. To reveal the cortical sources of the MMNm, Minimum Current Estimates (MNCEs) 6 were performed on individual subjects data. ROIs for statistical analysis were defined so that they included superior temporal, inferior frontal, inferior central (face/hand) and superior central (leg) areas. Results: MCEs indicated that in the left hemisphere the two words activated the four ROIs to different degrees; F(3,45)=4.3, p<0.009. Planned Comparison Tests confirmed stronger activation to potki than hotki in the leg ROI, F(1,15)=23.8, p<0.0001, whereas the inferior frontal and central ROIs revealed stronger activation to the face/arm word hotki compared with the leg word, F(1,15)=8.0, p<0.005. Time course analyses showed differences in the activation latencies of left hemispheric ROIs, F(3,45)=4.8, p<0.003. Superior temporal sources sparked at 160ms after the divergence point, followed by inferior frontal (172ms) and inferior central (arm) sources (176ms), F(1,15)=3.9, p<0.03, and, finally, the superior central (leg) ROI (200ms), F(1,15)=17.1, p<0.0005. Conclusions: 150-200 after information about stimulus words is present in the input, a leg word activated leg areas in superior central motor and/or premotor cortex, and a face-/arm-related word such as to eat activated inferior frontal areas anterior to the motor cortex, probably in premotor and prefrontal areas. Aspects of word meaning are reflected in early quasi-automatic word-evoked cortical activity . References: 1. Pulvermüller, F. Brain reflections of words and their meaning. Trends in Cognitive Sciences 5, 517-524 (2001). 2. Hauk, O., Johnsrude, I. & Pulvermüller, F. Somatotopic representation of action words in the motor and premotor cortex. Neuron in press (2004). 3. Hauk, O. & Pulvermüller, F. Neurophysiological distinction of action words in the fronto-central cortex. Human Brain Mapping in press (2004). 4. Pulvermüller, F., Hummel, F. & Härle, M. Walking or Talking?: Behavioral and neurophysiological correlates of action verb processing. Brain and Language 78, 143-168 (2001). 5. Näätänen, R. The perception of speech sounds by the human brain as reflected by the mismatch negativity (MMN) and its magnetic equivalent (MMNm). Psychophysiology 38, 1-21. (2001). 6. Uutela, K., Hamalainen, M. & Somersalo, E. Visualization of magnetoencephalographic data using minimum current estimates. Neuroimage 10, 173-180 (1999).

e142

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 88 Processing Native vs. Foreign Languages: Universals and Differences seen using fMRI Shirley-Ann Rüschemeyer 1 , Christian J. Fiebach 2 , Angela D. Friederici 1 1 MPI of Human Cognitive and Brain Sciences, Leipzig, Germany, 2 Helen Wills Insititute of Neuroscience, U of C, Berkeley, U.S.A. In the Chomskian view of language acquisition, the ability to learn, master and use a natural language is an innate and uniquely human ability. The underlying principles of language systems are thus assumed to be universal, and a set of given principles and parameters must be shared by all natural languages in their most basic forms. Not clear is the question of whether or not a universality of languages in terms of syntactic structure is reflected in universal language processing strategies on the neuronal level. In other words, do humans from different linguistic backgrounds employ the same cerebral areas to support the processing of their typologically very different native languages? Recent electrophysiological studies have shown that robust language-related event-related potentials (ERPs) are elicited from participants across a multitude of languages. This indicates that at least some aspects of language processing probably do rely on universal strategies, at least in their temporal parameters. Part 1 of the current study set out to investigate whether or not similar neural correlates also underlie the processing of different native languages. To this end brain activation correlated with the processing of acoustically presented syntactically and semantically anomalous sentences vs. correct sentences was compared in two groups of participants with differing native languages (German and Russian). The results show that participants with different native languages (L1) indeed rely on similar neural networks to process their respective L1. Specifically, syntactic anomalies vs. correct sentences brought on increased levels of activation in mid and anterior portions of the left lateral superior temporal gyrus (STG). The processing of semantic anomalies, on the other hand, elicited increased activation in the inferior portion the left inferior frontal gyrus (IFG; BA45/47). Furthermore electrophysiological studies have indicated that, while the processing of different native languages is probably quite similar, differences do exist in the temporal dynamics of L1 processing vs. second language (L2) processing. Part 2 of the current study therefore investigated whether or not different neural correlates of L1 vs. L2 processing could be detected using the same paradigm as described above. Identical sentence stimuli were presented to a group of native speakers and a second group of non-native speakers. The results indicate that different portions of the fronto-temporal language network support native and non-native language processing (see Figure 1). Specifically native speakers seem to rely greatly on superior temporal regions to quickly and efficiently decipher incoming acoustic speech stimuli. Non-native speakers, not as adept in the analysis and categorization of incoming speech signals, appear to resort to frontal cortical regions in order to make a final analysis.

e143

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Direct comparison of German participants listening to correct sentences in German (L1) vs. non-native speakers of German listening to the same sentences (L2). Native speakers show greater levels of increased activation in STG, while non-native speakers show greater involvement of IFG.

e144

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 89 The Contribution of the Basal Ganglia and the Cerebellum to Speech Motor Control Mechanisms. Axel Riecker 1,2 , Dirk Wildgruber 1,2 , Wolfgang Grodd 2 , Herrmann Ackermann 1 of Neurology and, 2 Department of Neuroradiology, University of Tuebingen

1 Department

Introduction. Only sparse data are available on the cerebral organization of motor aspects of speech production and on the pathomechanism of dysarthric deficits following brain lesions and diseases. Whereas bilateral lesions of the corticobulbar tracts ultimately may result in anarthria and/or aphonia, speech production does not fall below a level of about 3 Hz in cerebellar disorders [1]. In contrast, subgroups of patients with Wilson‘s or Parkinson‘s disease exhibit ‘speech hastening‘, i.e. involuntary acceleration of speaking rate, while speech tempo otherwise is found largely unimpaired in these disorders [2,3]. To clarify the differential contribution of these various cortical and subcortical structures to the pathomechanisms of dysarthric deficits subsequent to central motor disorders, brain activation was evaluated during syllable repetitions at fine-grained production rates from 2.0-6.0 Hz. Methods and Results. 8 right-handed subjects performed the acoustically paced syllable repetitions with 6 different frequencies: 2.0, 2.5, 3.0, 4.0, 5.0, 6.0 Hz. In the baseline condition (passive listening) subjects were instructed to pay close attention to the stimuli and to refrain from any responses. fMRI-data were acquired across the whole brain by Echo Planar Imaging. A T1-weighted MPRAGE served as an anatomical reference. Post processing was performed using SPM99. Statistical analysis considered a height threshold of T > 3.10 (p < 0.001) at voxel level and an extension threshold of k > 59 (p < 0.05) at cluster level to correct for multiple comparisons. The analysis of the obtained fMRI data comprised: (1) Parametric analysis to determine the influence of stimulus frequency upon hemodynamic response (fig. 1,2). This analysis models three different rate-to-response functions: main effect, linear and nonlinear in-/decrease during syllable repetitions. As two further steps of analysis various brain structures (fig. 1) were considered as volumes of interest in order to determine (2) the correlation coefficients across the time series (functional connectivity; fig. 3) and (3) the temporal dynamics (fig. 4). Discussion. Syllable repetitions yielded bilateral activation at supplementary motor area, sensorimotor cortex, putamen / pallidum, thalamus and cerebellum, whereas anterior insula and dorsolateral frontal cortex, including Broca’s area, showed a left-sided response. Assuming that the organization of speech motor control within the basal ganglia is more efficient at higher frequencies, we found a negative linear relationship between syllable repetition rate and BOLD signal change at the level of the striatum. In contrast, the cerebellar activation spots at either side exhibited a positive linear relationship. Most noteworthy, the analysis of fMRI time series and temporal dynamics revealed that the various brain regions engaged in speech motor control are organized into two separate networks subserving motor preparation on the one hand, and motor execution on the other. References. [1] Ackermann H. et al. (2000) J Neuroling 13: 95-116. [2] Ackermann H. et al. (1997) Brain and Language 56: 312-319. [3] Hefter H. et al. (1993) Acta Neurol Scand 87: 148-160.

e145

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1. Parametric analysis.

Figure 2. Relationship between BOLD signal intensity and stimulus frequency.

e146

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 3. Results of quantitative functional connectivity analyses.

Figure 4. The time course of hemodynamic activation.

e147

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 90 fMRI neuronal activation to single phonemes in left temporal cortex Lars M. Rimol 1 , Robert Savoy 2 , Karsten Specht 3 , Susanne Weis 4 , Kenneth Hugdahl 1,5 1 Department of Biological and Medical Psychology, Division of Cognitive Neuroscience, University of Bergen, Norway, 2 Harvard University and Massachusets General Hospital, 3 Institute of Medicine, Research Centre Jülich, Jülich, Germany, 4 Department of Epileptology, University of Bonn, Germany, 5 Haukeland University Hospital, Bergen, Norway Introduction To approach the question of how the brain transforms sound into speech, we used fMRI to map brain activation at very early stages of the speech perception process by using the smallest contrastive unit of a speech sound, namely single phonemes. Brain activation to single phonemes was contrasted with activation to consonant-vowel syllables, and duration-matched noise. Method The stimuli used were 3 CV syllables (/pa/, /ta/, /ka/) and 3 phonemes (/p/, /t/, /k/) that were extracted from the CV syllables. To control for general auditory activation we used 3 types of noise (white, brown and pink). The noise stimuli were matched in duration and intensity to phonemes and CV syllables. Thus, there were 3 short and 3 long noise stimuli, matching the duration of the phonemes and the CV syllables, respectively. 17 male, right-handed subjects with normal hearing participated in the study. A clustered-volume-acquisition paradigm was used, with TR =12 s and TA = 2,5 s. This allowed us to present the phoneme stimuli in a "silent gap" unconfounded by scanner noise. This accomplished two things: 1) The perception of the stimuli was not disturbed by the scanner noise. 2) The silent gap was long enough that the BOLD response to the scanner noise did not confound the BOLD response to the stimuli. In the silent gap of each trial/image acquisition 10 stimuli of one category (phonemes, CV, or noise) was presented in random order. ISI was 500 ms. For each condition there were 40 trials. Results Areas that showed speech specificity, i.e. significantly more activation to phoneme and CV than duration-matched noise, were found in the lateral parts of the upper, left temporal lobe (MTG and STG). The t-values were higher for the phoneme - duration-matched noise comparison than for the CV - duration-matched noise comparison, and the cluster was larger and extended further medially for phoneme - duration-matched noise. An asymmetry analysis, which compared the left with the right half of the brain for each condition, revealed that all stimuli were lateralised to the upper, left STG (planum temporale). But there was more extensive (larger clusters) and more consistent (higher t-values) lateralization for the speech sound conditions than for the respective noise conditions. And again, the lateralization of the speech sounds was stronger laterally, while as in more medial parts of the planum temporale noise was more strongly lateralised. Discussion The results have demonstrated that lateralization of speech is observable even at the phonemic level, i.e. at the most basic level of speech sounds. Furthermore, the results demonstrate that phonemes may be more effective speech stimuli than CVs in fMRI brain imaging. Finally, the results demonstrate that single phonemes can be used as auditory stimuli in fMRI research.

e148

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 91 Hemodynamic Effects of Morphosyntactic Operations and Polarity Reversal in a Tag Judgment Task Miguel A. Rubio 1,2 , Surina Basho 1,2 , Charity C. Johnson 1,2 , Shawn Sennett 1,2 , Holly Little 1,2 , Beverly Wulf*ck 3 , Judy S. Reilly 2 , Ralph-Axel Müller 1,2,4 1 Brain Development Imaging Laboratory, San Diego State University, 2 Dept. of Psychology, San Diego State University, 3 Center for Research on Language, University of California, San Diego, 4 Dept. of Cognitive Science, University of California, San Diego BACKGROUND Tag questions offer a window into morphosyntactic operations and into their prolonged acquisition. Previous studies [1,2,3] have identified various markers of morphosyntax that must be acquired in order to correctly produce tag questions (subject agreement, number agreement, inversion and polarity). All studies [1,2,3] have identified polarity as problematic, suggesting processes in addition to purely linguistic processing [1] and the recruitment of cerebral regions outside those previously identified for language processing. The present study examined processing of tag questions, focusing on effects specific to judgment of polarity violations. METHODS Ten healthy right-handed subjects (4m, 6f, 18-26y) were studied on a 3T scanner. For each subject, 184 time points were acquired. Subjects were asked to provide judgment on 140 items (30 no violation, 30 polarity violations and 30 subject, agreement and inversion violations). Also included were 20 baseline trials of the same length and prosody as tag questions, presenting semantically void syllable chains. Thirty 1-second null trials were included for temporal jittering. A model based on timing of separate items was applied for tag trials overall and specifically for polarity items. Imaging data were analyzed using a general linear model employing multiple regression to provide an estimation of hemodynamic response related to judgment of specific trial types. RESULTS For judgment of tags overall, activation was found in regions typically associated with syntactic processing in adults [4]. Effects were seen in the inferior frontal (Fig. A, B), bilateral medial frontal (Fig. C) and left middle temporal gyrus (Fig D.). Inferior frontal activations were bilateral, but more extensive and pronounced in the left hemisphere. Activations specific to polarity included bilateral middle frontal (Fig. E), bilateral medial frontal and anterior cingulate gyri (Fig. F, G). Activation in inferior frontal gyrus was mostly right hemispheric (Fig. H). CONCLUSION First, our findings support the notion that polarity in tag questions requires the use of resources beyond purely linguistic processes. Secondly, regions specific to these additional resources (middle frontal, medial frontal, anterior cingulate gyri) are similar to those activating during response inhibition and set shifting [5,6,7], indicating the nature of processes at work in polarity. Moreover, the prefrontal regions identified mature late compared to other cortical regions [8]. In conclusion, our hypothesis that polarity reversal would require cognitive processes beyond pure language was supported by the findings that show recruitment of brain regions more commonly associated with executive function. REFERENCES 1. Dennis M et al. Child Development 53: 1254-1257. 2. McGrath CO et al. J of Speech Hearing Research 16: 498-512. 3. Weeks LA J Psycholinguistic Research 21: 31-40. 4. Caplan D et al. NeuroImage 9: 343-351. 5. Konishi S et al. Nat Neurosci 1: 80-84. 6. Menon V et al. Human Brain Mapping, 12: 131-143. 7. Aron AR et al. Nat Neurosci, 6: 115-116. 8. Klingberg T et al. Neuroreport 10: 2817-2821. Supported by 1R01-NS43999. e149

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e150

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 92 Dissociations of expectancy and priming during effortful semantic processing Fred w. Sabb 1,2 , Susan Y. Bookheimer 2 1 Interdepartmental Program in Neuroscience, 2 Brain Mapping Center, Department of Psychiatry and Biobehavioral Science, UCLA Introduction. Semantic access involves both automatic and effortful processes, which can be dissociated using the semantic priming paradigm. One can examine effortful semantic processing in semantic priming studies through variation of the stimulus onset asynchrony (SOA) as short times between prime and target dont allow for effortful processing (1). Extending the SOA allows for the examination of variations in the expectancy of related information and other effortful processes (1). This variation in the probability of related information allows subjects to engage effortful processing to speed their responses. In a previous semantic priming fMRI study using pictures (2), we identified two regions that showed opposite blood flow responses to related stimuli across SOA. Here, using an event-related fMRI design, we present data that show manipulation of expectancy modulates the response in this same region. Method. 24 subjects performed one of two lexical decision semantic priming studies of printed words. The first study manipulated the expectancy of related targets. The second study kept expectancy constant, while manipulating SOA. In each study, we acquired 16 AC-PC slices over 6.5mins of scanning (GE; TR:2.5s; TE:54ms; 64matrix). Data were motion corrected, warped into Talairach space (3), and analyzed using SPM99. Regions of interest were derived from a previous study (2). We interrogated these regions with the SPM ROI toolbox, and performed random-effects ANOVAs on time-series data using unixSTAT. Results. A conjunction of regions significantly active during unrelated trials compared to baseline in both the expectancy and SOA studies yielded increased blood flow in the bilateral insula, both inferior and superior aspects of the left inferior frontal gyrus, supplementary motor area, and inferior parietal lobule on the superior edge of the supramarginal gyrus. The superior portion of the inferior frontal gyrus showed an increase response to unexpected trial types regardless of the expected trial type. However, it was not specifically sensitive to changes in SOA without an expectancy bias. A neutral block, when compared to rest, also produced the areas mentioned above with the addition of the right middle frontal gyrus. This right middle frontal gyrus region was sensitive to both expectancy changes and SOA changes. Discussion. This suggests a complex anatomical and temporal network for evaluating incoming semantic stimuli. Both unique areas for priming and expectancy exist, as well as anatomical areas where expectancy and priming information is temporally distinct. References. 1-Neely JH (1991) in Basic progresses 2-Sabb FW & Bookheimer SY, (2003) SFN abstracts 3-Woods R, et al (1999), HBM 8.

e151

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e152

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 93 Correlated Functional Changes of the Prefrontal Cortex in Twins Induced by Classroom Education of Second Language Kuniyoshi L. Sakai 1,2 , Kunihiko Miura 2,3 , Nobuko Narafu 2,3 , Yukimasa Muraishi 2,3 of Cognitive and Behavioral Science, the University of Tokyo, Komaba, 2 CREST, JST, 3 The Secondary Education School Attached to the Faculty of Education of the University of Tokyo, Japan

1 Department

Introduction In contrast to the first language (L1), a second language (L2) can be mastered at any time in life, though the L2 ability becomes rarely comparable to L1 if it is acquired beyond the hypothesized critical period around puberty. The critical experiments would be to clarify brain plasticity that represents new acquisition of L2, i.e., the dynamic processes of L2 acquisition and associated cortical changes. Subjects and Methods Participants in the present fMRI study were 14 native Japanese speakers (8 females and 6 males, all aged 13). They consisted of six monozygotic and one dizygotic twin pairs. For two months, the subjects participated in intensive training in English verbs as part of their standard classroom education. The twins completed two sets of functional magnetic resonance imaging (fMRI) sessions, one before the training (Day 1) and one after the training (Day 2). Four tasks were used in fMRI sessions: an English verb-matching (EM) task, an English past tense (EP) task, a Japanese verb-matching (JM) task, and a Japanese past tense (JP) task. The fMRI scans were conducted using a 1.5 T scanner. Using a gradient-echo echo-planar imaging sequence (TR = 3 s, TE = 50.5 ms, resolution, 3 x 3 mm), we scanned 16 horizontal slices, each 6 mm thick and having a 1-mm gap, covering from z = 49 to 62 mm. We performed random effects analyses on fMRI data using SPM99. Results In parallel with the performance improvements for English past tense verbs, we observed clear activation increases in cortical regions (Figure 1). When EP and EM blocks were contrasted in a random effects analysis for Day 2, major activations were found in the following regions: the dorsal region of the left inferior frontal gyrus (IFG), the left F3t/F2, and the right cerebellum. All of these activations were absent in EP - EM for Day 1. The left IFG exhibited significantly correlated activation increases within each pair of twins, and the increases were positively correlated with individual performance improvements. Moreover, the cortical plasticity for L2 acquisition was guided toward the L1 specialization of the left dorsal IFG (1). Conclusion These findings suggest a cortical mechanism underlying L2 acquisition, which critically depends on shared genetic and environmental factors for each twin in a surprisingly predictive manner (2). To our knowledge, the present study is the first direct demonstration that classroom education, if properly executed, can change the function of the prefrontal cortex. References 1. Hashimoto, R., and Sakai, K. L., Neuron 35, 589-597 (2002). 2. Sakai, K. L. et al., Cereb. Cortex, in press (2004).

Figure 1. The neural systems for processing L2 past tense verbs.

e153

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 94 Two routes for naming objects, one for naming actions Riitta Salmelin 1 , Mia Liljeström 1 , Antti Tarkiainen 2 , Tiina Parviainen 1 , Jan Kujala 1 , Jussi Numminen 1 , Jaana Hiltunen 2 , Matti Laine 3 1 Brain Research Unit, Low Temperature Laboratory, Helsinki University of Technology, Espoo, Finland, 2 AMI Centre, Helsinki University of Technology, Espoo, Finland, 3 Department of Psychology, Åbo Akademi University, Turku, Finland Introduction Category-specific impairments in aphasic patients suggest that names of actions (’to light’ a candle) and objects (’a candle’) are retrieved using dissociable cerebral networks [1,2]. In the naming tests used, objects and actions are usually named from different sets of pictures (objects only vs. actions). In recent fMRI [3] and MEG [4] studies, however, when nouns and verbs were named from an identical set of pictures, no category-specific dissociation was found in healthy subjects. Behavioural observations suggest that nouns can be retrieved via two routes, a nominal route (confrontation naming of objects) or a propositional speech route (normal language production) [5]. We investigated this idea using fMRI. Methods The task was to silently name actions (Act) or objects (ObjAct) from line drawings illustrating a simple event (e.g. to write with a pen [4]) and objects (Obj) from these images when the action had been dissolved into arbitrary lines in the background. In a block design, 30-s task and 21-s rest periods alternated. Following an instruction to name nouns or verbs, 10 images were shown (duration 300 ms, interval 1.8-4.2 s). There were 10 blocks per task. The whole brain was imaged with 3x3x3 mm3 voxels, using a GRE-EPI sequence on a GE 3T system (TR = 3 s, TE = 32 ms, flip angle = 90 deg). The random-effects analysis of 9 subjects (19-30 yrs, 5 females) was done using SPM2. Results All naming tasks resulted in activation of bilateral occipitotemporal cortex and left inferior parietal and frontal lobes (Fig. 1, left). The pattern of activation was practically identical for naming actions (Act) and objects (ObjAct) from the same pictures (Fig. 1, right). Activation was enhanced in this entire network when objects were named from action pictures (ObjAct) rather than from stand-alone object images (Obj). On the other hand, when objects were named from images which did not depict an event (Obj) the left superior temporal cortex emerged as a relatively more strongly activated area. Conclusion The overall pattern of activation [2,4,6,7] and the similarity of this pattern to naming actions and objects from the same pictures [3,4] are in line with earlier reports. The cerebral network of noun retrieval was affected by context. Verbs are an integral part of normal sentence formulation and, therefore, prime candidates for the propositional speech production route. The present data suggest that (i) the cerebral dissociation of object and action naming, or the lack of it, is determined by the current framework and goal of language production rather than by stimulus category as such, and (ii) the parieto-frontal network is associated with production of propositional speech whereas naming via the nominal route additionally engages the left temporal cortex. References 1. Caramazza A, Hillis A. Nature 1991, 349:788 2. Damasio A, Tranel D. PNAS 1993, 90:4957 3. Hernandez et al. Neuroimage 2001, 14:510 4. Sörös et al. Neuroimage 2003, 19:1787 5. Manning L, Warrington E. Neuropsychologia, 1996, 34:809 6. Murtha et al. JOCN 1999, 11:399 7. Salmelin et al. Nature 1994, 368:463.

e154

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Results of SPM2 analysis. Act = action, ObjAct = object in action, Obj = object alone.

e155

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 95 Brain activation during rapid listening with an fMRI study Yuko Sassa 1,2 , Yuko Akitsuki 2,3 , Nobuo Usui 4 , Masato Taira 5 , Ryuta Kawashima 2 , Taka-aki Tanaka 6 , Kisou Kubota 7 1 LBC Research Center, Tohoku University 21st Century Center of Excellence Program in Humanities, 2 NICHe, Tohoku University, 3 Department of Psychiatry, Tohoku University Graduate School of Medicine, 4 Japan science and technology agency, 5 Nihon university Graduate school of medical science, 6 SSI Corporation, 7 Department of Social and Information Sciences, Nihon f*ckushi University Introduction The analyses of the relationship between brain activity and speed (frequency) have been investigated by PET and fMRI during motor-related tasks. However, the relationship between brain activity and speed of cognitive processes has remained unclear. In addition, although there are many functional imaging studies about brain activity for listening to sound, especially speech sound, those studies did not investigate brain activities during listening to different speeds of speech. Therefore, the purpose of our study was to investigate the brain activity correlated to the speed of speech sounds using fMRI analysis. Methods Twenty-five right-handed normal subjects participated in this study. They were all native speakers of Japanese. We prepared simple declarative Japanese sentences as the stimulus, but half of them were acceptable sentences and the rest were sentences including a grammatical anomaly. We prepared digitized sounds of a trained female speaker who read those sentences at standard speed, and re-digitized them into three types of different speed: one and a half times as fast as standard speed, twice as fast as standard speed, two and a half times as fast as standard speed. We did not change frequency and loudness modulations. We used white noise as a baseline condition. The subjects were introduced to judge whether or not a sentence was acceptable and to respond to every sentence by pressing one of two buttons with right hand. We used experimental design of event-related fMRI (Siemence Vision 1.5 T; TR = 5000 msec, 24 slices, 4 mm thickness (0 mm gap)). We excluded the data of MR images if answers were not correct. Data analysis was performed using SPM99. In order to investigate the effect of the difference of the listening speed, we used the speed rate as a covariate. Simple regression analysis was done to detect the activated regions correlated with the speed of speech sounds. Results and Conclusion Compared with the white noise, all conditions activated the left precentral gyrus, the bilateral inferior frontal gyri, the right prefrontal regions, posterior part of the left medial inferior temporal gyrus. In addition, the activated regions correlated with the speed of speech sounds were the bilateral inferior frontal gyri and the right prefrontal regions. We showed that activity of the right frontal cortices and the left inferior frontal gyrus had positive correlations with the speed of speech sounds. These results indicate that activation of these areas may be related to the increases of cognitive load and /or increasing load for attention during rapid listening.

e156

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 96 The role of the left frontal cortex during judgment of grammatical violation : an fMRI study Yuko Sassa 1,2 , Motoaki Sugiura 2 , Jobu Watanabe 1,2 , Yuko Akitsuki 2,3 , Kazuki Iwata 1,2 , Naho Ikuta 2,4 , Naoki Miura 2,5 , Hideyuki Okamoto 2,6 , Shigeru Sato 4 , Ryuta Kawashima 2 1 LBC Research Center, Tohoku University 21st Century Center of Excellence Program in Humanities, 2 NICHe, Tohoku University, 3 Department of Psychiatry, Tohoku University Graduate School of Medicine, 4 GSICS, Tohoku Univ., 5 Graduate School of Engineering, Tohoku University, 6 Graduate school of Medicine,Tohoku University Introductions It has been reported that the left inferior frontal gyrus (LIFG), especially the Brocas area, is activated during comprehension of syntactic complex sentences by positron emission tomography (PET) and functional magnetic resonance imaging (fMRI). In addition, these areas are also activated during grammaticality judgment task using short sentences including a grammatical violation. From these studies, it has been thought that the LIFG are related to syntactic processing. Although there were many functional imaging studies investigated brain regions associated with syntactic processing, most of these studies investigated brain regions involved in only a type of syntactic processing. Therefore, in this study, we tried to use different types of grammatical violations and to investigate brain regions involved in specific syntactic processing, using region of interest (ROI) analysis of fMRI. Methods We prepared simple declarative Japanese sentences as the auditory stimulus. Five types of sentences that were contained a grammatical violation and normal (well-formed) sentences were presented through a pair of headphone in random order. Twenty-seven right-handed normal subjects participated in this study. They were asked to judge whether or not a sentence was acceptable. The subjects were instructed to respond to every sentence by pressing one of two buttons with right hand. We used event-related fMRI (Siemence Vision 1.5 T ; TR = 5000 msec, 20 slices, 4 mm thickness (2 mm gap)). We excluded the data of MR images if answers were not correct. We performed data analysis using SPM99. To investigate which region of the LIFG was related to each type of grammaticality violation, we set ROIs on subtracting images from hearing normal sentences to grammatically anomalous sentences. Results The LIFG and the premotor area were activated in judgment of each types of anomalous sentences. In particular, anterior part of the LIFG, inferior part of the LIFG, posterior part of the left prefrontal and the left premotor area were strongly activated. Anterior and inferior regions of the LIFG significantly activated in tense disagreement, violation of case particles position and sentence without case particles comparing with other types of anomalous sentence, whereas posterior part of the left prefrontal significantly activated in verb conjugation disagreement and violation of case particles position comparing with other types of anomalous sentence. Discussion Our results showed that the magnitude of activation in the LIFG was different according to different types of syntactic processing by ROI analysis. Furthermore, they showed tendency that anterior and inferior regions of the LIFG were involved in grammaticality judgment associated with sentence meaning, whereas posterior part of the left prefrontal was involved in grammaticality judgment associated with morphological. These results indicate that different brain regions in the LIFG are related to processing of different types of grammaticality judgment.

e157

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 97 The Bilateral Nature of Language Processing: An fMRI Study of Anomaly and Complexity Robin J. Schafer , R. Todd Constable Yale University Syntactic complexity and anomalous sentences have been independently examined in neuroimaging studies. Previous results from fMRI studies of these two phenomena show similar areas of differential activation, particularly in inferior frontal gyrus (IFG) and superior or medial temporal gyrus (STG, MTG). This study examines the two factors simultaneously with the goal of isolating differences in the degree or kind of activations that are observed with the two phenomena. fMRI data were collected on 12 healthy control subjects exposed to sentence stimuli in blocks that varied in complexity and in degree of anomaly. Sentences were presented visually and auditorily. Imaging was performed using a 1.5T GE SignaLx with a standard quadrature head coil: 16 axialoblique slices (6 mm thick, 0 gap) with gradient echo planar imaging (64x64 matrix, TE=45ms, a= 80, TR=1800ms). A whole brain analysis of activations was conducted. Because a factorial design was not employed, a regression on anomaly and complexity measures per block was calculated. Areas of activation fitting the regression model defined by beta weights per block for these two predictors were examined. Results suggest that an increase in activation in a subpart of IFG (BA 44) can be attributed more to complexity than to anomaly, whereas that in the STG (BA 22) is influenced predominantly by anomaly. We interpret this as a basic difference in the processing of the two types of language phenomena. Interestingly, we found that in both cases the activations were bilateral. Activations that pattern with complexity are similar across modality, and the bilateral increase in activation in BA 44 occurs in both modalities. We consider this an important finding, as all subjects show a dominant left language processing. Activations that pattern with anomaly differ across modality, and bilaterality was observed only in the visual modality. Finally, we report on variations by condition and modality in frontal cortex (BA 9, 10, 46) and in the anterior and posterior cingulate cortex (BA 24, 32, 23, 31).

e158

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 98 An fMRI study of lexical and compositional semantics Robin J. Schafer , R. Todd Constable Yale University Functional MRI studies of semantic knowledge concern themselves almost exclusively with lexical semantics, the interpretation of words. Particularly these studies examine the associations among words or their classification into semantically influenced categories. In the latter type of task, subjects are presented with lists of words that are prototypical members of one of two classes, such as abstract or concrete noun, and are asked to classify the items. The work here builds on these previous studies in two ways. First we examine the processing of phrases as well as that of lexical items in two parallel tasks. The semantics of phrases, compositional semantics, is theoretically quite distinct from lexical semantics. This aspect of our work probes the question of whether left inferior prefrontal cortex plays the crucial role in compositional semantics that it does in lexical semantics. Second we expand on the classification tasks by including task items which are marginal, as well as those which are prototypical members of a class. The classification used is the abstractconcrete noun classification. We present prototypical abstract and concrete nouns (idea, teacher) as well as marginal examples (swish, group). Both types of nouns are presented in three formulations: as single, lexical items and as members of two types of noun phrase, intersecting and nonintersecting. Both noun phrase types consist of an adjective plus a noun. Intersecting noun phrases are formed with adjectives that are intersecting: they form subcategories of the category named by the noun. Examples include tall or cool: tall teacher is a subcategory of the class of teachers, cool idea is a subcategory of the class of ideas. Nonintersecting noun phrases are formed with adjectives that are nonintersecting: they do not clearly form subcategories of the category named by the noun. Examples include dead or incomplete: it is not clear that a dead teacher is a teacher or that an incomplete idea is an idea. N=16 healthy, native English speaking subjects participated in a 2x3 factorial, block design fMRI study. Materials were presented visually over 10 runs of six blocks each. Each block contained a mix of abstract and concrete nouns or noun phrases of one of the six conditions: prototypical lexical, intersecting, or nonintersecting, marginal lexical, intersecting, or nonintersecting. Adjectives were chosen so that they could be used with both abstract and concrete nouns of both the prototypical and marginal types. Subjects were asked to indicate whether each item was abstract or concrete. fMRI images were collected on a 3T Siemens Trio with an 8-channel phased array head coil using 22 axialoblique slices (4 mm thick, 0 gap) with gradient echo planar imaging (64x64 matrix, TE=30ms, a= 80, TR=1500ms). A general linear model analysis is used to compare whole brain activation patterns across the six conditions. Main effects for semantic processing (lexical v. compositional) as well as for noun classification (prototypical v. marginal) serve to clarify the role of IFG in a semantic network.

e159

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 99 The neural substrates of speech components an fMRI study Ralph Schnitker 1 , Armin Thron 2 , Walter Huber 3 1 Interdisciplinary Centre for Clinical Research Neurofunctional Imaging Lab, University Hospital Aachen, 2 Department of Neuroradiology, University Hospital Aachen, 3 Neurolinguistics at the Department of Neurology, University Hospital Aachen Introduction fMRI studies dealing with language processing often use covert speech rather than overt speech, in order to avoid susceptibility and motion artifacts caused by articulation during scanning. Recent fMRI studies have demonstrated that this problem can be solved by using event-related designs. Thereby a direct comparison of overt and covert articulation can be made. In addition, isolated phonation without the movement of any articulators is of interest to disentangle components of speech (which is important for future investigation of patients with motor speech disorders). In the present study, we systematically compared the neural substrates of phonation, covert and overt speech using a simple word repetition task. The stimuli consisted of German pseudo words and pronounceable non words to exclude semantic and lexical components of speech processing. Methods 9 male and 6 female healthy, right-handed volunteers took part in this study on a 1.5T Philips Gyroscan Intera MRI system. We used a T2*-weighted EPI sequence (TR 2800ms, TE 50ms, FA 90?); 33 transversal slices with a thickness of 3.3 mm were acquired covering the brain without the cerebellum. The experiment was conducted in an event-related design with three sessions representing the three speech modalities in the following order: covert articulation, phonation and overt speech. The same stimuli were used for each modality in pseudo randomized order and jittered around an inter-stimulus interval (ISI) of 6s. The stimuli were presented over headphone. Imaging data was analyzed using SPM2. Results All three conditions yielded significant bilateral activations of the auditory association cortex (BA21/22) extending into the anterior part of Wernicke?s area and its hom*ologue on the right side (BA22). The extent of the activations decreased across speech modalities in the order of presentation from covert articulation to phonation to overt articulation. Significant activations of the speech motor areas (BA6/4) were found in the left hemisphere during all three conditions. In the right hemisphere, activations of the pre-central gyrus and the hom*ologue of Broca?s area (44) were only found during covert articulation and phonation. The left putamen was involved in phonation and overt articulation only. Discussion Bilateral activations of the auditory association cortex as input to speech were reported by several authors before. The decrease of the activations may be caused by an increasing familiarity of the participants with the stimuli. The speech motor areas and Broca hom*ologue on the right hemisphere were only active in covert articulation and phonation, i.e. when specific components of speech had to be inhibited. We therefore assume that the motor speech areas of the right hemisphere are responsible for inhibition of phonatory and articulatory components of speech, but do not contribute naturally to complex programming and execution of speech. The findings of other studies in which overt speech showed more extensive motor cortex activations than covert speech could not be confirmed in the present study.

e160

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 100 Early Category-Specific Semantic Processing In The Brain As Shown By Somatotopic Event-Related Responses To Auditorily Presented Action Words: ERP And L2 Minimum-Norm Source Analysis Yury Shtyrov , Friedemann Pulvermuller Medical Research Council (MRC) Cognition and Brain Sciecnces Unit, Cambridge, UK A growing body of clinical and neuroimaging evidence suggests that not only the core language areas in the brain but also other cortical areas may participate in the processing of linguistic information. This evidence speaks in favor of the association learning theory which postulates that neurons that are active at the same time become strongly linked [1]. Thus, if a word is learned in conjunction with an object or action, a neuronal assembly will be formed comprising neurons in both language cortical areas and visual or sensorimotor cortices. Such neuronal network will function as a memory trace for the word in question and become active whenever the word is perceived [1]. A more fine-grain prediction would be that even sub-category differences may lead to differential cortical activation patterns. For action-related words, words referring to movements of different body parts should incorporate neurones in different parts of sensorimotor cortices whose topological organization of these cortices is well-known (Fig. 1). To test this hypothesis, we recorded mismatch negativity (MMN), a well-known index of experience-dependent memory traces in the brain [2], to investigate the processing of action-related words in the human brain. Responses to auditory presented movement-related English words were recorded in non-attend odd-ball paradigm using a high-density EEG set-up. MMN was calculated using responses to the same words presented as standard and deviant stimuli in different sessions to avoid contamination from phonetic-acoustic differences. The spatio-temporal patterns of the activation were analyzed by calculating ERPs as well as computing cortical source estimates (L2 minimum norm) for each subject; the data were then subjected to ANOVA. We found (Fig. 2) that: 1. The topography of the mismatch negativity to the action words showed unusual centro-posterior distribution of the responses suggesting that activity was at least in part generated posterior to usually observed frontal MMNs. 2. MMN responses to the hand-related word stimulus (pick) had a more spread-out lateral distribution, whereas the leg-related stimulus (kick) elicited a more focal dorsal negativity. These differences, remarkably reminiscent of the sensorimotor cortex somatotopy, were confirmed by the source analysis (L2 minimum-norm current estimates). 3. The latency of these word-specific MMN responses was in the range of 140-180 ms and could possibly be related to the recognition of the individual words. The present results are best explained in terms of distributed neuronal assemblies which function as category-specific memory traces for words and might involve sensorimotor cortical structures for encoding action words. The current data (see also [3]) are also confirmed by MEG [4] and fRMI [5] findings. The observed effects occurred early in time suggesting that semantic processing may commence in the brain as early as ~140ms after the word onset. 1.Pulvermuller F. Progr Neurobiol 67: 85-111 (2002) 2.Naatanen R et al. Nature 385:432-434 (1997) 3.Shtyrov Y et al. European J Neurosci (2004) 4.Pulvermuller F. et al. This volume. 5.Hauk O et al. Neuron 41: (2004)

e161

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig. 1. Illustration of our somatotopy-of-action-words (SAW) model.

Fig.2. Results: ERP voltage maps (top view) and source estimates (left view) of the MMN to leg- and arm-related words.

e162

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 101 Spatio-Temporal Pattern Of Word-Elicited Activity In The Human Neocortex Yury Shtyrov , Friedemann Pulvermuller Medical Research Council (MRC), Cognition and Brain Sciecnces Unit, Cambridge, UK Identical acoustic contrasts are known to produce larger mismatch negativity (MMN) response when incorporated in word than in non-word context [1-4]. This word-related enhancement of the MMN was linked to activation of long-term memory traces for words [1,3]. Word-elicited MMN responses were consecutively suggested as a tool for investigating spatio-temporal patterns of neural memory traces for words, words’ individual signatures in the brain. In our previous study, we used this tool to demonstrate for the first time the pattern of language-related activation spreading in the left hemisphere [5]. Acoustically presented words activated superior temporal cortex at about 140ms after the relevant information was present in the auditory input; at ~20ms later, the activation could be registered in the inferior-frontal areas. It remained unclear, however, whether such activation patterns are characteristic of any phonological stimulus processing or are elicited only by meaningful members of the lexicon. To further detail spatio-temporal profile of language-related cortical activity, we performed another study. Using state-of the-art high-density MEG (Neuromag Vectorview, 306 channels) and passive odd-ball paradigm (visual attention task, no attention to the auditory stimulation), we recorded MMN responses elicited by identical acoustic change (presence/absence of final inflectional affix [t]) in Finnish verb, noun and a pseudoword. All stimuli were closely matched for their physical, acoustic and phonological properties; the MMN was computed as a difference between responses to the same words presented as deviant (i.e., rare unexpected) and standard (frequent) stimuli to avoid any contamination from phonetic-acoustic differences. We found (Figs. 1-2) that: 1. The MMN responses to both words were larger then that to the matched pseudoword. This difference was most pronounced in the left hemisphere. 2. The MMNs to words were larger in the left than in the right hemisphere. The MMN to the matched pseudoword stimulus was less lateralized than those to words. 3. In the left hemisphere, the MMN responses to words first peaked in the superior temporal lobe. This was followed, with a delay of ~20ms, by left inferior frontal activation. 4. In the right hemisphere, no latency differences between the activation of temporal and frontal areas could be seen. 5. The pseudoword did not demonstrate a similar activation pattern; the smaller pseudoword response peaked near-simultaneously in both temporal and frontal channels. These data confirm the previously reported MMN enhancement for word stimuli. They suggest that MMN can be used to document specifically spatio-temporal patterns of cortical activity related to word processing: the spread of activation from temporal to inferior-frontal cortices could only be seen for words but not for matched pseudoword stimulus. The current results also indicate early lexical and/or semantic processing in the brain which commences before ~150ms after the relevant information is available in the auditory input and is largely independent of focused attention. 1. Pulvermuller F et al. NeuroImage 14:607-616 (2001) 2. Korpilahti P et al. Brain Lang 76:332-339 (2001) 3. Shtyrov Y, Pulvermuller F. NeuroReport 13:521-525 (2002) 4. Endrass T et al. This volume. 5. Pulvermuller F, Shtyrov Y et al. NeuroImage 20:2 1020-1025 (2003)

e163

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig.1. MMN responses to words (but not to pseudoword) peaked later in the left inferior frontal than in temporal recordings. Responses to words are larger than that to the matched pseudoword.

Fig.2. In the right hemisphere, no delay between the temporal and frontal activation can be seen. Word-related MMN enhancment is substantially less pronounced than in the left cortices.

e164

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 102 Making sense of nonsense texts: An fMRI study on coherence building via strategic inference processes Florian Th. Siebörger 1 , Evelyn C. Ferstl 1, 2 , D. Yves von Cramon 1, 2 1 Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany, 2 Day Clinic for Cognitive Neurology, University of Leipzig, Germany A central process for text comprehension is the establishment of coherence. The antero-dorsal fronto-medial cortex (adFMC, medial BA9/10) has been shown to be crucial for coherence building (Ferstl & von Cramon, 2001, 2002). Because of its involvement in other higher cognitive demands (Theory of Mind, emotional processing, etc.) we hypothesize that adFMC activation during comprehension of coherent sentence pairs reflects a domain independent, non-automatic process rather than stimulus properties. The goal of the present event-related fMRI study was to show that, depending on the performed task, adFMC activation can be elicited also by incoherent trials. 14 participants were scanned while they listened to 30 coherent and 90 incoherent sentence pairs and judged the perceived strength of coherence on a 4 point scale (pragmatical relationship graded from directly evident = 1 to unimaginable = 4). Since even incoherent sentence pairs can be considered coherent depending on the listeners creativity, we expected that especially for incoherent trials a search for coherence is performed requiring adFMC activation. The behavioural data show that the participants indeed used their creativity: 61 % of the incoherent sentence pairs were rated as somewhat related and the response times for those trials were significantly prolonged. For a first analysis of the functional data we contrasted all trials against a jabberwocky sentence control condition time-locked 1 second before the end of the second sentence. Along with several fronto-temporal areas, text processing involved the adFMC as expected. A time-course analysis of the latter region revealed that the amplitudes of the signal change peaks were comparable for all four response conditions. Instead the activation was due to differences in the peak latencies, which varied in the same way as the response latencies. For comparisons between the 4 response conditions we therefore time-locked the BOLD signal 1 second before the individual response to cover activation during the supposed search for coherence instead of online processes. While medial BA9/10 was equally activated in all 4 conditions, trials with weak or no coherence (response 2, 3 or 4), compared to trials with directly evident coherence (response 1), activated medial BA 8 more strongly. This area has been associated with decision making under uncertainty (Volz et al., 2003). In line with this interpretation the timelines indicate that the intermediate categories 2 and 3 drive this effect. The comparison of successful and unsuccessful coherence building within the incoherent trials (contrast: response category 2 & 3 vs. 4) yielded no medial activation differences. Orbital, fronto-lateral and parietal activations indicate that evaluative, executive and memory processes are involved in this task. In summary, we showed that the adFMC activation is not stimulus dependent but task related.

e165

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 103 Your brain says what it sees: motor mechanisms of audiovisual speech perception Jeremy I. Skipper 1 , Howard C. Nusbaum 1 , Virginie van Wassenhove 2 , Frederic Dick 1,3 , Steven L. Small 1 1 The University of Chicago, 2 The University of Maryland, College Park, 3 Birkbeck College, University of London Introduction We have proposed a model where the observation of a speakers mouth movements affects a listeners speech perception. This occurs via an internal and covert production process, one mirroring the speakers motor behavior in the listener (Skipper et al., in review). If this is true, then seeing and hearing a speaker produce a syllable should activate the same brain regions underlying the hearers own production of that syllable. Moreover, these production-related regions should be sensitive to specific phonetic visual information in audiovisual speech. To test this hypothesis, we examined the distribution of brain activity resulting from incongruent audio and visual phonetic information, for example, hearing an audio pa while simultaneously seeing a mouth pronouncing ka. Participants typically hear this syllable as a ta (the McGurk-MacDonald effect). Methods Right-handed native English speakers participated in whole brain spiral functional magnetic resonance imaging (fMRI) at 3 Tesla. There were three presentation conditions. In the audiovisual (AV) condition participants saw ka and heard the incongruent syllable ta (we refer to this as pk); participants also saw and heard audiovisually congruent syllables pa, ka, and ta. In audio-alone (A) and video-alone (V) conditions, participants saw or heard pa, ka, and ta. Each stimulus was heard an equal number of times. No behavioral response was required during the AV, A, and V conditions. Following these three runs, a second AV condition was performed; here participants used a 3-button mouse to classify the syllables as pa, ta, or ka. Finally, in a speaking condition (S) participants spoke pa, ka, and ta. This condition identified brain regions active during syllable production. A deconvolution/regression analysis was used to reveal voxel-wise signal change for each condition. Resulting activation patterns were analyzed using cortical-surface-based, cross-participant ANOVAs. Results Audiovisual and visual-only conditions - but not the audio-alone condition -produced cortical activity that overlapped with cortical activity during speech production in several regions (Figure 1) including the pars opercularis of Brocas area, premotor and primary motor cortices, the postcentral gyrus, and posterior superior temporal gyrus. In particular, the incongruent AV syllable produced differential activation in these areas relative to the congruent AV syllables. The incongruent syllable also evoked changes in regions associated with visual processing, including the middle temporal and fusiform gyri. Conclusions These results support a model where auditory speech comprehension is affected by visual information about speech production through activation of a motor network. Visual speech cues affect speech comprehension by making use of the listeners speech motor knowledge and experience. Furthermore, our results suggest that novel audio-visual pairings may prime or "re-tune visual cortices to incorporate the novel incongruent AV stimulus into what the brain already knows about audio-visual contingencies.

e166

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Distribution of activation (p<.00001) demonstrating motor-related frontal activation in audiovisual, video-alone, and speaking, but not audio-alone conditions in the left (LH) and right (RH) hemisphere. Statistics are collapsed over syllables.

e167

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 104 Inter-regional covariances in a fronto-temporal system for speech and language. Emmanuel A. Stamatakis 1 , William Marslen-Wilson 2 , Lorraine K. Tyler 1,3 , Paul Fletcher 4 1 Department of Experimental Psychology, University of Cambridge, Cambridge, UK, 2 MRC Cognition and Brain Sciences Unit, Cambridge, UK, 3 Wolfson Brain Imaging Unit, University of Cambridge, Cambridge, UK, 4 Department of Psychiatry, University of Cambridge, Cambridge, UK The ways in which language processes and representations are instantiated in the brain constitute a key area of research in cognitive neuroscience. The distinction between regular and irregular verb inflections provide an important input to this research since evidence from behavioural dissociations in patients with lesions suggests that processing regular forms involves a frontal neuronal circuitry and irregulars engage more temporal regions (Marslen-Wilson & Tyler, 1998; Tyler et al, 2002). This general distinction has been supported in a recent neuroimaging study using efMRI, in which we contrasted regular (stayed/stay) and irregular (teach/taught) past tense forms with pseudo past tense forms matched on phonological complexity (jade/jay, peach/port). All words were matched on number of syllables, familiarity, lemma and wordform frequencies. There were 56 word-pairs in each condition. 18 Subjects performed a timed same-different judgement task on spoken word-pairs for regulars/irregulars and pseudo regulars/irregulars. We showed that a fronto-temporal network, linking anterior cingulate (ACC), left inferior frontal cortex (LIFG) and bilateral superior temporal gyrus (STG), is preferentially activated for regular past tense forms (Tyler et al., submitted). We propose that this reflects the additional processing demands posed by regular inflected forms, requiring modulation of temporal lobe lexical access processes by morpho-phonological parsing functions supported by the LIFG. We carried out a complementary re-analysis of the data, in order to investigate this system in terms of inter-regional covariances, which are taken as an index of functional connectivity. We identified cortical regions in which activity was predicted by LIFG and ACC, and critically, by the interaction between these two regions. Furthermore, we determined the extent to which these inter-regional correlations were influenced differentially by the four experimental conditions. Using this approach, we expressed brain activity in terms of inter-regional correlations and of modulations of these correlations by other brain regions and by the experimental conditions. We found that a functional connection between LIFG and left fusiform gyrus (LFG) BA20 (p=0.029 corrected cluster level, Fig. 1a) was significantly stronger for processing real regulars than pseudo regulars. A similar result was found for connectivity between LIFG and RSTG BA42 (p=0.046 corrected cluster level, Fig. 1b). This finding was not replicated in the case of real irregulars vs. pseudo irregulars (or in other pair-wise comparisons). Additionally, we found that functional connectivity between LIFG and left MTG (Fig. 1c) is positively modulated by activity in ACC and that this effect is significantly modulated by experimental condition (greater for regular than irregular items). This latter finding constitutes a three-way interaction that may be explicable in terms of three integrated effects: first, a positive influence of LIFG on temporal activity; second a modulatory influence of ACC activity upon this fronto-temporal connectivity; third a condition-specificity of this modulatory effect such that the cingulate influence upon fronto-temporal connectivity is greater when processing regular items. These findings point to a complex network underlying the processing of spoken words differing in morpho-phonological complexity.

e168

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Functional connection between the LFG (a) and LIFG was significantly stronger when processing real regulars than pseudo regulars. Similarly, functional connection between the RSTG (b) and LIFG was significantly stronger when processing real regulars than pseudo regulars. Finally, a three way interaction was found between the ACC, LIFG and LMTG (c) that was greater for regular than irregular items.

e169

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 105 Brain Activities in the Memory Process of Phonologically Ambiguous Graphemes: An fMRI Study Using Japanese Speakers 1 f*ckushima

Chika Sumiyoshi 1 , Kayako Matsuo 2 , Yuko Ohgami 2,3 , Toshiharu Nakai 2 University, f*ckushima, Japan, 2 National Institute of Advanced Industrial Science and Technology, Ikeda Osaka, Japan, 3 Ochanomizu University, Bunkyo-ku, Tokyo, Japan

Introduction There are not independent phonological categories for /r/ and /l/ in Japanese language system, assimilating into one substitutive phoneme. The lack of independent phonological categories for these sounds causes one-to-many ambiguous association between the phonemes and the relevant graphemes. Similarly, in Japanese writing system, there are a few irregular kana scripts in which grapheme-to-phoneme association is one-to-many as found in "r" and "l" graphemes for Japanese speakers. The purpose of the present study was to investigate whether different brain activity was observed between the memory process of "r" and "l" graphemes and that of ambiguous kana graphemes in Japanese speakers. Method Ten right-handed native Japanese speakers (F/M=6/4, age 21-43) who gave written informed consent participated in the experiment. Visual short-term memory task [1, 2] was employed using either alphabetical syllables or kana (hiragana) scripts. First, three target stimuli were visually presented one at a time. After 1.5 second, a probe appeared requesting subjects to decide whether the probe was present or absent by pressing buttons with their left hand. A task block contained five trials and each sequence contained 4 task blocks interleaving the rest periods. The experiment consisted of two English syllable task sequences and two hiragana task sequences. In the ambiguous kana task sequence (AK) and the r/l task sequence (RL) sequence, grapheme-to-phoneme association was one-to-many contrasting to the normal kana task sequence (NK) and the b/n task sequence (BN). A gradient recalled echo EPI sequence was employed for functional studies on 3T MR scanner (GE, Signa VH/i 3.0T). The imaging parameters were TR 3000 msec, TE 30 msec, FA 90 degree, FOV 22 cm, and slice thickness 3mm plus 1mm gap. Thirty axial slices were obtained. The fMRI data were analyzed using SPM99 (Wellcome Department of Cognitive Neurology, London, UK). A random effect model was applied with the height threshold p=0.005 at voxel level (uncorrected), and extent threshold p=0.05 at cluster level. One sample t-test was conducted for the each task sequence while the paired t-test was done for the comparison between the r/l- and the ambiguous kana task sequences. Results and Discussion Extensive activation in the right MFG (BA 9) was observed both for the r/l task sequence and for the ambiguous kana task sequences compared with the b/n task sequence and the normal kana task sequence, respectively (Figure 1; pink arrows). The greater activation in the area indicates that the working memory load was heavier [3] for these phonologically ambiguous graphemes. The paired t-test revealed the superior activation in the left SMG (BA 40) for the ambiguous kana task contrasting to the r/l task sequence (Figure 1; yellow arrow). The area has been reported to be responsible for the storage the phonological information [4]. The result suggests that subjects tried to memorize ambiguous kana as phonological codes, even though the graphemes lack definite association with relevant phonemes. References [1] Paulesu, E. et al., Nature, 362, 342-345, 1993 [2] Sumiyoshi, C. et al., Neurosci Lett, 336;50-54, 2003 [3] Collette, F. et al., Brain Res Cogn Brain Res, 7, 411-7, 1999 [4] Paulesu, E. et al., Brain, 119, 143-157, 1996

e170

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1

e171

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 106 fMRI Evidence for Regularity Effects in High-Frequency Words Li Hai Tan The University of Hong Kong, Hong Kong Regularity effects are often demonstrated with low-frequency, but not high-frequency, words. Thus, the interactivity of regularity and frequency has been widely regarded as the most crucial evidence in evaluating major cognitive models of lexical processing. In the present event-related fMRI study with covert naming, the regularity x frequency interaction is found to hinge on other factors, such as the relative frequency of wholes (i.e. words) and parts (e.g., rimes in alphabetic words or phonetic components in Chinese characters). For high-frequency words, if the frequency of their constituents (rimes or phonetic components) is higher than the frequency of whole words, there is a robust regularity effect, as illustrated by much stronger brain activity for irregular words in bilateral middle temporal gyrus, left precuneus, posterior cingulate, and right extrastriate cortex. These cortical regions are known to mediate orthography-to-phonology transformation and monitoring of competition. When the frequency of whole words is higher than that of parts (phonetic components), there is no regularity effect. This pattern of findings supports the proposal that word recognition abides by a statistical principle, according to which, a parallel processor for parts and wholes ranks access to different linguistic units by their conditional probability. (The author thanks Zhen Jin for her assistance with image acquisition. This research was supported by a grant HKU 3/02C from the Research Grants Council of the Hong Kong Government.)

e172

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 108 Neuroimaging guided rTMS interferes with semantic priming in normal subjects Alexander Thiel 1,3 , Birgit Habedank 1 , Hans-Joachim Markowitsch 2 , Walter F Haupt 1 , Karl Herholz 1,3 , Wolf-Dieter Heiss 1,3 1 Department for Neurology, University of Cologne, Germany, 2 Physiological Psychology, University of Bielefeld, Germany, 3 Max-Planck-Institute for neurological research, Cologne, Germany Background: Neuroimaging studies of right-handed normal controls under semantic word generation tasks have consistently reported left lateralized activation of the anterior inferior frontal gyrus (ifg) which decreased during task repetition (1,2). This repetition related activation decrease has been interpreted as the neurophysiological correlate of a semantic priming mechanism which reflects implicit memory for initial semantic processing. Objectives: In order to put this hypothesis to the empirical test, we tried to interfere with left lateralized ifg activation, as identified by O-15-water PET activation using repetitive transcranial magnetic stimulation (rTMS) in five right-handed male normal subjects. rTMS was performed twice, once using new nouns and once using known nouns for the procedure. Methods: We used O-15-water PET during an auditory verb generation task to identify the respective areas of the left ifg in 5 normal right-handed male subjects. Optimum stimulation sites for rTMS over the activated areas were determined on 3D-renderings of the subjects head obtained from T1-weighted MR images. rTMS at 4Hz and at individual motor threshold was performed during verb generation over the respective sites (left ifg, right ifg and vertex) once using lists of new nouns and once using known nouns. Results: All five subjects exhibited clear left lateralized activations of the triangular part of the left ifg in the PET studies. In all subjects reaction time latencies were significantly longer during rTMS over the activation sites in the left ifg, as compared to latencies off stimulation. Latencies during stimulation of the right ifg or over the vertex were not affected. These effects were observed within the group and in each individual, only if lists of known nouns were used in the verb generation task. Conclusion: These results support the hypothesis that the anterior part of the left ifg is essential for semantic priming since successful interference with rTMS was only observed if known word lists were used for the generation task. This however also outlines a discrepancy with respect to neuroimaging results: A TMS effect is induced over a region with the same task which causes a decrease of the hemodynamic response within this region. It may thus be assumed that the priming effect is not caused by a reduction of neuronal activity in the left ifg but by a shorter duration of neuronal activity which may also lead to a decrease in the hemodynamic response (3). 1 Petersen et al., PNAS Volume 95 Number 3, 1998 2 Buckner et al., Brain, Vol. 123, 2000 3 Henson et al., Neuropsychologia 41, 2003

e173

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Decrease of reaction time latencies in 5 subjects during verb generation with new and known lists of nouns. Latencies increase again if rTMS is applied over Broca’s Area

e174

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 109 Brain activation during rapid listening with an fMRI study: Effects of training Nobuo Usui 1,2 , Tomoki Haji 1 , Izuru Nose 1 , Yuko Sassa 3 , Ryuta Kawashima 3 , Masato Taira 1 , Taka-aki Tanaka 5 , Kisou Kubota 6 1 Nihon university Graduate school of medical science, 2 Japan science and technology agency, 3 NICHe, Tohoku University, 4 Bunkyo Gakuin University, 5 SSI Corporation, 6 Nihon f*ckushi University Introduction In our daily life, we experience various speeds of speech sounds in conversation. Although it is hard to understand a rapid speech, a training of listening to the rapid speech seems to have an effect on listening comprehension. In this study, we focus on the brain areas related to the listening comprehension at various speech speeds and investigate of which areas are correlated with the training career (This is a series of experiments of our group. See also Sassa et al.). Methods Twenty-four right-handed normal subjects (7 females and 17 males, mean age 31.6 ± 7.2 years old) participated in this study. They were all native speakers of Japanese. We prepared simple declarative Japanese sentences as the stimulus, but half of them were acceptable sentences and the rest were sentences including a grammatical anomaly. We prepared digitized sounds of a trained female speaker who read those sentences at standard speed, and re-digitized them into three different speeds: one and a half times, twice and two and a half times as fast as standard speed. Pitch and loudness were the same among four conditions. We used experimental design of event-related fMRI (Siemence Symphony 1.5 T, TR = 5 s, 25 slices, 4 mm thickness, 1 mm gap). We excluded the data of MR images of incorrect trials. We performed data analysis using SPM99. In order to investigate the effect of the listening speed and training career, we used the speed rate and total number of training session as covariates (mean total number of training was 102.3 ± 199.6, range 0 to 720 times). We performed simple regression analysis to detect the activated regions correlated with the listening speed and training career. Results and Conclusion In subtraction analysis between the rapid speed and the baseline conditions, bilateral inferior and middle frontal gyri and bilateral cingulate gyri were activated. In these areas, the bilateral inferior frontal gyri, the left middle frontal gyrus and bilateral cingulate gyri had positive correlations with the speed of speech. In addition, the right prefrontal area had a positive correlation with speed. These results are in the same line with another study of ours (Sassa et al.), except that the left middle frontal and the cingulate areas were activated. On the other hand, we found that activations of these areas were positively correlated with the training career. These results suggest that the training of rapid listening may effect on these cortical areas.

e175

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 110 Hemispheric specialization of the visual word form area :Inter-individual variability Mathieu Vigneau 1 , Gaël Jobard 1 , Fabrice Crivello 1 , Guy Perchey 1 , Bernard Mazoyer 1,2 , Nathalie Tzourio-Mazoyer 1 1 GIN, UMR 6194, CNRS/CEA/Univ. Caen & Paris 5, France, 2 IRM CHU Caen, Institut Universitaire de France Introduction A previous meta-analysis 1 of word reading functional imaging studies allowed to demonstrate the reliable activation of the so-called visual word form area (VWFA), and to calculate its coordinates and SD in the stereotactic space. For some authors, the VWFA is specifically implicated in word and pseudoword reading 2 , while for others it is recruited as well during object perception 3 , naming and word listening 4 . In order to shed some light on the functional role of the VWFA, we calculated within the ROI defined a priori from the results of the meta-analysis as the VWFA (x = -44, y = -58, z = -15; SD = ± 5mm) both left and right as well as individual asymmetry indices signal variations during various linguistic tasks. Materials and methods A first group of 13 right-handed male were submitted 4 runs of fMRI (BOLD) acquisitions (1.5-T GE Signa). The runs alternated 30 sec duration periods of either reading words or non-words (string of vowels, string of consonants, finish words) with a cross fixation task, using a block design. The second group of 10 subjects did the same word reading run, plus a word listening task run, with the same design. Within the VWFA ROI, the BOLD signal variations were calculated in each individual contrast (SPM99), for each run. Both left, right and asymmetry indices of fractional BOLD signal variations were calculated in the VWFA ROI. Results and discussion All subjects (N = 23) activated the left VWFA during word reading and 19 subjects the right VWFA, thereby demonstrated the robustness of both its anatomical location and functional implication (Fig. 1). When comparing word and non-word reading in the first group as mean (N = 13), no difference was detected in either the left (p=0.57) or right (p=0.08) VWFA (Fig. 2) The investigation of the functional asymmetry index (FAI, left minus right) revealed the presence of a large leftward asymmetry, significantly higher for words than for nonwords (p=0.008, Fig. 2), due to a larger right-hemisphere activity during non-word reading. Individual analysis revealed a disparity in term of functional asymmetry: 17 subjects showed a leftward asymmetry while 6 presented a rightward lateralization or symmetry in the VWFA during word reading (Fig. 3). In the second group the left VWFA showed a significantly higher activation during visually than auditory word presentation (p=0.006, N = 10), while the right VWFA showed the opposite pattern, i.e. a deactivation during listening (p=0.058), what led to similar leftward FAI during the two tasks (p= 0.686; Fig. 4). Conclusion The present results question the existence of a functional specialization of the left hemisphere VWFA for words reading. They suggest that such a specialization could seat in larger FAI, favoring the left hemisphere, during words than non-words reading. In fact, such difference in FAI was due to larger right hemisphere involvement during non-words reading. This could attest of a difference in strategy during word and non-word reading: prevalence of verbal processing for word and increase in visual analysis for non-words. One should note that these strategies showed inter-individual variability that could be linked to the mode of learning taught during childhood. Furthermore, the leftward activation of the VWFA during listening shows that this region can be considered as part of a supramodal semantic language network. 1. Jobard, G. et al. Neuroimage. 2003. 2. Cohen, L. et al. Brain. 2002. 3. Mellet, E. et al. Neuroreport. 1998. 4. Price, C.J. et al. Neuroimage. 2003

e176

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e177

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e178

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 111 See What I mean? Sentences With Spatial Meaning Engage Brain Regions Involved in Non-linguistic Spatial Tasks 1 Center

Mikkel Wallentin 1,2 , Svend Østergaard 2 , Leif Østergaard 1 , Andreas Roepstorff 1 for Functionally Integrative Neuroscience, Aarhus University Hospital, Denmark, 2 Center for Semiotics, University of Aarhus, Denmark

Introduction The ability to convey complex mental scenarios is at the heart of human language function. Advances in cognitive linguistics suggest that language comprehension is mediated by an ability to activate cognitive systems for non-linguistic processing of spatial information [1]. In this fMRI study we investigated the neural systems underlying spatial semantics, as it unfolds in Danish intransitive sentences governed by motion verbs and spatial prepositions. Methods 40x4 sentences like the below were created. All nouns were held in the singular determinate form, and verbs were all held in present tense. The subject noun could take on either an animate (e.g. man) or an inanimate role (e.g. trail), and the prepositional complement could be either concrete (e.g. house) or abstract (e.g. sorrow) yielding quite different interpretive framing to the context (i.e. spatial vs. social). This method resulted in 2x2 structurally identical, yet semantically very different, types of sentences, e.g.: 1. Manden går gennem huset [The man goes through the house] 2. Manden går gennem sorgen [The man goes through the sorrow] 3. Sporet går gennem huset [The trail goes (leads) through the house] 4. *Sporet går gennem sorgen [The trail goes through the sorrow] 17 subjects underwent fMRI scanning in a 1.5 Tesla GE scanner, both in a reading and a listening paradigm. In both experiments sentences were presented with a 3.2 sec. interval. Sentences from the four different groups were presented in blocks of five, amounting to a total of 16 sec., and a stimulus onset asynchrony of 64 sec. Data were submitted to an RFX analysis using SPM2. Significance threshold was set to p<0.05, FDR-corrected. Results and Discussion Sentences with a concrete prepositional complement (type 1 & 2) activated a distinct bilateral posterior network of brain regions, including: a) the temporal-occipital-parietal junction (TOP), b) the fusiform/parahippocampal region, c) precuneus, and d) the retrosplenial area. Two smaller unilateral frontal activations (e & f) were also detected (figure 1 &2, table 1). This network has previously been shown to be activated by mental navigation [2], and spatial memory tasks [3]. Sentences with an abstract prepositional complement activated a largely left-lateralised network in anterior temporal cortex, and inferior and superior prefrontal cortex (figure 2A), which has previously been found activated in the comprehension of complex social semantics such as narratives [4]. Of these regions Type 4 sentences only produced activation in the right inferior frontal lobe, probably due to the lack of conventional meaning in this sentence type (figure 2B/C). Conclusion Concrete mental spaces produced with verbal cues make use of a network of brain regions also involved in processing topographical information by non-linguistic cognitive modalities. This supports a view of language processing where spatial semantics overlap with the more general cognitive processes of monitoring location in space. References 1. Talmy, L., Toward a Cognitive Semantics. 2000: MIT Press. 2. Ino, T., et al., Mental navigation in humans is processed in the anterior bank of the parieto-occipital sulcus. Neuroscience Letters, 2002. 322(3): p. 182-186.

e179

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

3. Burgess, N., et al., A Temporoparietal and Prefrontal Network for Retrieving the Spatial Context of Lifelike Events. NeuroImage, 2001. 14(2 SU -): p. 439-453. 4. Maguire, E.A., C.D. Frith, and R.G.M. Morris, The functional neuroanatomy of comprehension and memory: the importance of prior knowledge. Brain, 1999. 122(10): p. 1839-1850. Table 1 Brodman

MNI x

MNI y

MNI z

Tal x

Tal y

Tal z

Z score

L fusiform & parahippocampal gyrus

20.37

-32

-44

-20

-32

-43

-15

5.35

R fusiform & parahippocampal gyrus

20.37

-38

-24

-38

-18

4.61

L retrosplenial

-16

-64

-16

-62

3.01

R retrosplenial

-58

-56

4.85

L occipital-temporal-parietal junction

19/39

-80

-77

4.74

R occipital-temporal-parietal junction

19/39

-40

-82

-40

-79

4.21

5.31

-4

-40

-4

-37

4.83

L precuneus

-24

-74

-24

-74

4.29

L middle frontal gyrus

-32

3.35

R medial frontal gyrus

-12

-13

3.67

Cluster locations Effects of Concrete Prep. Complement

L & R cingulate gyrus, R postcentral gyrus

Peak coordinates for significantly activated clusters in the Concrete>Abstract contrast (p<0.05, FDR-corrected). A bilateral pattern of activations in the temporal and parietal lobes is detected (figure 1 & 2).

e180

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Sentences with Concrete prepositional complement yield bilateral activation in: a) the temporal-occipital-parietal junction (TOP), b) the fusiform/parahippocampal region, c) precuneus, and d) the retrosplenial area, and two smaller unilateral frontal activations (e & f).

Concrete > Abstract contrast depicted on a 3D rendering of a mean normalised structural image of the 17 subjects. Cut surfaces in MNI coordinates: X= -10, y= -42, z= -20.

e181

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Abstract > Concrete. As can be seen from a comparison between A and B, most of the left lateralised activation generated by the abstract sentences (A) seem to stem from type 2 sentences (B), whereas the right lateralised inferior frontal activity primarily is generated by the less understandable Inanimate-Abstract sentences (C).

e182

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 112 Neural correlates of processing function words and content words Isabell Wartenburger 1,2,3 , Nicole Stadie 1 , Hauke R Heekeren 2 , Astrid Schroeder 1 , Eva Eckholt 1 , Ria De Bleser 1 , Arno Villringer 2 1 Dept. of Patholinguistics, University Potsdam, Germany, 2 Berlin NeuroImaging Center, Dept. of Neurology, Charité University Medicine Berlin Campus Mitte, Germany, 3 Dept. of Neurology II, Otto-von-Guericke University Magdeburg, Germany Lesions in inferior frontal cortex often result in agrammatic aphasia. Some agrammatic aphasics show impairments in processing function words (FW; pronouns, prepositions, conjunctions conveying sentential structure) but not content words (CW; nouns, verbs, adjectives conveying meaning/content of a sentence) while some fluent aphasics show the reverse pattern [1,2]. This dissociation is also observed in dyslexia, where patients with deep/phonological dyslexia (fronto-temporal lesions) show selective impairments in reading FW, while surface dyslexics (inferior temporal lesions) are impaired in processing CW [3-9]. The aim of this study was to examine the role of left frontal and temporal regions in processing CW and FW. Based on the lesion data we hypothesized greater fronto-temporal activation during processing FW and greater inferior temporal activation during processing CW. Seven healthy, right-handed young females had to judge whether a CW or FW is a word or a non-word, or whether two identical symbols occur in a false-font control task while their brain activity was monitored using functional MRI (1.5 T Siemens, Erlangen). Imaging data were analyzed using standard SPM2 methods. Behavioral data: There were no differences (reaction times, accuracy) between FW and CW. FMRI data: Comparing both CW and FW to control resulted in greater activation of the left inferior/middle frontal gyrus, left middle temporal gyrus, right inferior frontal gyrus, and right cerebellum. There was a main effect of word class: Processing FW compared to CW resulted in greater activation of left inferior/middle frontal gyrus (Brodmann area [BA] 45/46) and left inferior parietal lobe (BA 40). Comparing CW to FW resulted in greater activation of left fusiform gyrus (BA 19). There was neither a main effect of lexicality (processing words vs. processing non-words), nor an interaction between word class and lexicality. Our data suggest that left inferior frontal and parietal brain regions, which are supposed to be part of the phonological system [6], play a greater role in processing FW compared to CW. These results thus corroborate studies in deep/phonologic dyslexic patients (see above). The greater activation of the left fusiform gyrus during processing CW as compared to FW might be due to greater imageability or greater need for semantic retrieval in CW as compared to FW [10]. Supported by grants from BMBF (BMBF-MOS Cooperation FKZ 01GA0202 and Berlin Neuroimaging Center) and International Leibniz Program. References: 1. Friederici, AD. Brain Lang 15, 1982. 2. Ellis, AW, Young, AW Human cognitive neuropsychology Lawrence Erlbaum, Hove, 1988. 3. Marshall, JC, Newcombe, F. J Psycholinguist. Res. 2, 1973. 4. Vanier, M, Caplan, D. in Surface Dyslexia (eds. Patterson, K, Marshall, JC, Coltheart, M) Lawrence Erlbaum, London, 1985. 5. Coslett, HB. Semin. Neurol. 20, 2000. 6. Price, CJ. J Anat. 197 Pt 3, 2000. 7. Coltheart, M, Patterson, K, Marshall, JC Deep dyslexia Routledge & Kegan Paul, London, 1980. 8. Patterson, K. in Case studies in the neuropsychology of reading (ed. Funnell, E) Psychology Press, Hove, 2000. 9. Small, SL, Flores, DK, Noll, DC. Brain Lang 62, 1998. 10. Wise, RJ et al. Neuropsychologia 38, 2000.

e183

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 113 Hemispheric dominance for language: development of an fMRI testing battery for young children Marko Wilke 1,2 , Martin Staudt 1,2 , Karen A. Lidzba 1,2 , Karin Buchenau 1 , Ingeborg Krägeloh-Mann 1 1 Pediatric Neurology and Developmental Medicine, Childrens Hospital, and, 2 Section Experimental MR of the CNS, Dept. of Neuroradiology, Radiological Hospital, University of Tübingen, Germany Introduction The question of the dominant hemisphere for language has been among the most-asked question since the advent of fMRI. The aim was and is to find an alternative and, ultimately, a replacement for the invasive Wada-maneuver [1]. However, many of the more sophisticated tasks cannot be used in children, due to compliance and performance issues [2]. We set out to compile and develop a test battery for use in young children (5y and up). Rationale & Approach We aimed at implementing tasks that should (i). yield strong activation in language areas of the dominant hemisphere; (ii). be easily doable by young children; (iii). allow for performance monitoring and (iv). ideally already be in use with children. We examined existing tasks and proceeded to develop new tasks, partly based on previous studies. Results Two existing tasks fit our criteria, albeit with different emphasis. The synonyms-tasks [3] yields strong activation in the dominant hemisphere and allows for performance monitoring. Still, the necessity of reading skills is a drawback. The verb generation paradigm [4] only allows for an indirect performance monitoring, but is otherwise easily applicable and well-tested in children. An error-story task based on previous works [5] was chosen since it seemed to accommodate the average childs liking for stories. Short stories for children developed by Holland et al. [4] were modified for this task. Two tasks were developed locally (vowel-identification and farm animal task). In the former, children were asked whether a pre-chosen sound (the vowel i [pronounced ee in German]) was present in the pronounced name of an object presented pictographically on the screen. In the latter, children had to decide whether an animal shown pictographically on a screen lives on a farm, a semantic decision similar to earlier tasks [6]. Illustrations of test runs for each task are provided in Figure 1. Discussion and Outlook The verb generation task has successfully been used in even young children. The synonyms task has the disadvantage of requiring reading skills, but yields a very reliable and strongly lateralized activation pattern in adults [3] and may thus be suitable for older children. The vowel identification and the error-story task show great promise in these early stages, with the latter tapping automated language processing components [7]. The animal task in its current form does not seem to provoke strong activation in language areas and will thus be modified to include additional verbal stimuli. The tasks have now all been implemented and testing in volunteers (healthy children aged 7-15 years) is in progress. Literature [1] Abou-Khalil B, Schlaggar BL: Neurology 2002; [2] Wilke M, et al.: Neuropediatrics 2003; [3] Fernandez G, et al.: NeuroImage 2001; [4] Holland SK, et al. NeuroImage 2001; [5] Ni W, et al.: J Cogn Neurosci 2000; [6] Szaflarski JP, et al.: Neurology 2002; [7] Friederici AD & Kotz SA: NeuroImage 2003

e184

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e185

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 114 Semantic Processing in Bilinguals: A Cross-modality Study Savio W. H. Wong 1 , John A. Spinks 1 , Ho-Ling Liu 2 , Jian-Chuan Chen 2 , Li-Hai Tan 3 1 Department of Psychology, The University of Hong Kong, 2 Lab for MR Research, MRI Center, Dept. Diagnostic Radiology, Chang Gung Memorial Hospital, Taiwan, 3 Joint Laboratories for Language and Cognitive Neuroscience, The University of Hong Kong A controversial issue in the neuroscience literature is how different languages are represented in the brain. Recent fMRI studies showed that languages with distinctive characteristics (e.g. Chinese and English) involved activation of dissimilar neural substrates for various processing. The present experiment studied the neural substrate responsible for semantic processing of Chinese and English across modality. This cross-modality study enables us to separate the semantic processing, which should be less sensitive to the orthography and the phonology of specific language, from the process of lexical encoding. Methods Six Chinese-English bilinguals participated in the experiment. They are undergraduate students with similar proficiency in both languages. The experiment consisted of two sessions: 1) the auditory session; and 2) the visual session. In the auditory session, a pair of Chinese or English words was presented to the participant through a headset in each trial. The participant then, had to judge whether the two words are semantic associated. Participants would listen to a pair of African words in the baseline condition of the auditory session and they would be asked to press the response key randomly. Similarly, participants had to do the same semantic judgment task in the visual session. The pair of words was presented visually through a LCD goggle. Participants would be asked to complete a font-size judgment task in the baseline condition. Participants were scanned with a 1.5T scanner and the acquired images were analyzed using SPM99. A random effect model was used to examine the difference between the semantic association condition and the baseline condition within modality and language. The contrast images of the two languages in different modalities were entered into two separate one-sample t-tests for locating the semantic processing of specific language. Then, the contrast images of different languages and modalities were put together into a one-sample t-test to locate the semantic processing which is independent of language and modality. Results The left inferior frontal area (BA 44) was activated in both Chinese semantic judgment condition (fig. 1) and English semantic judgment condition(fig. 2) across modality (p<0.05, corrected). Figure 3 presented the neural substrate responsible for the semantic processing across language and modality. It showed that the activation in area 44 was independent of language and modality and statistically significant (p<0.05, corrected). Conclusion The experiment showed that an unified area, the left inferior frontal region (BA 44), was responsible for the semantic processing in both Chinese and English and it is modularity independent. Thus, the area 44 can be regarded as a part of the central conceptual system in the cognitive model of word processing. Its function is independent from language and input modality. Therefore, the distinctive characteristic of languages may affect the recognition (lexical encoding) process rather than the semantic processing.

e186

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The neural system for Chinese semantic processing across modality

The neural system for English semantic processing across modality

e187

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

The neural system for semantic processing across language and modality

e188

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 115 Neural Systems Engaged in Reading Japanese Kanji and Kana Words: Effects of Phonetic and Non-phonetic Writing Systems on Brain Activation Benjamin Xu 1 , Kenji Ishii 2 , Takao Fushimi 3 , Keiichi Oda 2 , William Theodore 1 , Kiichi Ishiwata 2 , Itaru F. Tatsumi 3 1 CES/NINDS, National Institutes of Health, U.S.A., 2 PET Center, Tokyo Metropolitan Institute of Gerontology, 3 Language, Cognition and Brain Science Research Group, Tokyo Metropolitan Institute of Gerontology Reading is an acquired cognitive skill that relies upon essential components of the language system. Behavioral studies have shown that word reading automatically engages cognitive processes common to reading in languages with vary different writing systems such as English, a sound-based (or alphabetic/phonetic) system, and Chinese, a logographic (or non-phonetic) system. However, evidence from neuropschological studies of patients with brain lesions and neuropathology suggests that brain injury may selectively impair the ability in reading and writing of words in one type of the system more than the other even in the same language system like Japanese (Iwata, 1984; Sakurai et al., 1994). Iwata (1984) summarized findings in studies of Japanese patients with difficulties in reading and/or writing Japanese kanji (a non-phonetic writing system) and kana words (a phonetic writing system), and proposed that kanji and kana words are processed with two distinct neural networks. Recent studies using functional neuroimaging techniques with normal functioning readers have not presented consistent evidence of such distinct neural networks. In this study, we show that reading Japanese kanji and kana words engages neural systems common in other languages. However, comprehension of the meaning of kanji and kana words may be differentially mediated by neural systems, particularly in the right prefrontal cortex, important for more general cognitive functions. A HEADTOME-V SET 2400W PET scanner (Shimadzu, Kyoto, Japan) was used. A bolus of 150 MBq H215O was injected intravenously for each one-minute scan of ten scans. The PET data were processed and analyzed with SPM99 software. Twenty right-handed healthy adult volunteers participated in the study and viewed kanji or kana words/non-words (in separate scan runs). Stimuli were presented one at a time (stimulus duration = 500 ms; SOA = 1500 ms). Participants made either phonological or semantic judgment decisions. Experimental controls for kanji and kana words matched the experimental conditions for visual input, response characteristics, and memory load. Two scans were acquired for each of the five task conditions. Results of the study showed that reading kanji and kana words activated brain regions common to similar findings in other alphabetic languages (Rumsey, 1997; Xu et al., 2001). Specifically, using SPM conjunction analyses, both types of words activated left mid and inferior posterior-prefrontal cortex (BA: 44, 45, and 47) when the tasks required phonological similarity judgment. When semantic judgments were required, similar brain regions and the left mid-temporal area (BA: 21) were activated regardless of word type. In addition, semantic judgment tasks with either kanji or kana words consistently activated the right inferior-prefrontal cortex (BA: 45 and 47). However, phonological processing of kana words did not appear to significantly activate these brain regions in the right prefrontal cortex. These results suggest that neural systems important for word reading may be universal for all languages regardless of differences in the writing system. Nevertheless, reading words in a phonetic and a non-phonetic writing system such as Japanese kanji and kana may differentially engage neural systems (e.g., the inferior-prefrontal cortex in episodic and semantic memory [Buckner, 1996]) important to more general cognitive functions.

e189

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 116 The cerebral and functional difference of the regular and irregular inflection of Korean verb Eojeols Hyungwook Yim 1 , Changsu Park 1 , Heuiseok Lim 2 , Kichun Nam 1 1 Department of Psychology, Korea University, Seoul, 2 Department of Information and Communication, Cheonan University, Cheonan The ’English past tense debate’ has long been an issue not only in language processing but also in human mental computation itself. One side of the debate has been single-route approaches that propose all past tense are from the stored associate memory. On the other side of the debate are dual-route approaches that propose only irregular past tenses are stored in the memory while the regular past tenses are generated by rules. The paper presents both human behavioral(RT) experiments and brain imaging(fMRI) experiments and applies the aforementioned hypothesis to Korean verb *Eojeols. In behavioral experiments, LDT(lexical decision task) and naming task were used to investigate comprehension and production respectively. The LDT session had four conditions which were regular and irregular verb Eojeol conditions, a noun filler condition and a non-word condition while, the naming task session had three conditions that did not have a non-word condition compared to the LDT session. Analyses on the two tasks were performed with partial correlation between reaction time and subjective familiarity of both regular and irregular verb Eojeols. The subjective familiarity was measured by over 100 undergraduate students who did not participated in the experiment. Double dissociation between regular and irregular conditions was found both in LDT session and naming task session but was not statistically significant. In brain imaging experiments, block designed fMRI scans were used during LDT and inner speech tasks. The LDT session had four conditions which were regular and irregular verb Eojeol conditions, noun filler condition and a control condition while the inner speech task session had five conditions which had a extra control condition compared with the LDT session. Analysis using SPM99 did not show any difference between the regular and irregular verb Eojeols conditions. Both conditions showed statistically significant activation mainly in the temporal region and the medial temporal region which is related with lexical memory. The results did not show significant activation on the prefrontal regions which is related with generating rules. As shown in the results above, experiments with Korean verb Eojeols indicate that both irregular and regular verb Eojeols in Korean are not generated by rules, which supports the single-route approach. Moreover, in studies with English dealing with the past tense debate, computational models were the main supporters of the single-route approaches while the present study was supported by behavioral experiments and brain imaging experiments. The difference between the two experiments could be inferred by the characteristics of the two languages. The Indo-European language is defined as an inflected language while Korean is defined as an agglutinative language. An agglutinative language has richer morphemes and complex morpheme formation rules. Thus, in Korean it would be more efficient to find a fitting morpheme during the inflecting process by a semantically ordered index than a rule as in an inflected language. * In Korean, an "Eojeol" (word phrase) is a unit for spacing consisting of more than one combined morpheme.

e190

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 117 An event-related fMRI study of how active and passive sentences are comprehended in Japanese Satoru Yokoyama 1,2 , Wataru Nakamura 1 , Jobu Watanabe 2,3 , Yuko Sassa 2,3 , Kazuki Iwata 2,3 , Yuko Akitsuki 2,4 , Naoki Miura 2,5 , Hyeonjeong Jeong 1,2 , Naho Ikuta 1,2 , Jorge Riera 2,3 , Hideyuki Okamoto 2,6 , Nobuo Usui 7 , Masato Taira 7 , Shigeru Sato 1 , Kaoru Horie 1 , Ryuta Kawashima 2 1 Graduate School of International Cultural Studies(GSICS), Tohoku University, Sendai, Japan, 2 New Industry Creation Hatchery Center, Tohoku University, Sendai, Japan, 3 LBC Research Center, Tohoku University 21st Century Center of Excellence Program in Humanities, Sendai, Japan, 4 Department of Psychiatry, Tohoku University Graduate School of Medicine, Sendai, Japan, 5 Graduate School of Engineering, Tohoku University, Sendai, Japan, 6 Graduate school of Medicine,Tohoku University, Sendai, Japan, 7 School of Medicine, Nihon University, Tokyo, Japan The present study is one of the few initial steps toward providing a neurological foundation or verification of controversial linguistic analyses. In this study, we carried out an event-related functional magnetic resonance imaging study to investigate whether there is any difference in case of language comprehension between an active and passive sentence. Twenty native speakers of Japanese participated in this study. They were asked to read an active or passive sentence in Japanese and to identify who the agent or patient is in that sentence by pressing a button. The results showed that the reading of a passive sentence results in a significantly more activation of the inferior parietal lobule of the bilateral hemispheres than the reading of the corresponding active sentence. Linguistic studies (Perlmutter and Postal, 1977; Bruzio, 1986) claim that a passive sentence is more marked than an active sentence in terms of the morphological complexity and the way thematic roles are mapped onto grammatical relations. Through linguistic theory (Jackendoff, 1972; 1990) and psycholinguistic study (Ito et al., 1993), we may assume that the more processing load of comprehension of a passive sentence is related to the reanalyzing of the thematic roles of the sentence than that of an active sentence. Previous functional brain imaging studies (Inui et al., 1998) reported that the bilateral inferior parietal lobule is involved in understanding the relationship among characters which appear in a sentence. Previous ERP study (Osterhout et al., 1994) also suggested that the parietal region is involved in processing of the integration or reanalyzing of thematic roles of a sentence. Linguistics, psycholinguistics, functional brain imaging study, ERP study, and the present study converge to establish that comprehension of a passive sentence requires more processing effort than that of the corresponding active sentence and that the increased processing load is attributable to the reanalysis of thematic roles toward the end of the sentence. Furthermore, we found that the identifying of an agent of a transitive sentence yields a greater activation of the right middle temporal gyrus than that of a patient. Linguistics (e.g., Jackendoff, 1972; 1990; Grimshaw, 1990) and nuropsycholgical study (e.g., Caramazza and Miceli, 1991) claimed that an agent has a more prominent status than a patient in a transitive sentence. The results obtained in this study are compatible with claims by these previous studies.

e191

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 118 Properties of oxygen metabolic response in Wernicke’s area during the verbal tasks using functional NIRS-imaging Kayoko YOSHINO 1 , Ruriya WATANABE 1 , Toshihide KOIKE 1 , Takanori MAESAKO 2 , Toshinori KATO 3 1 Department of Human Development, Faculty of Education, Tokyo Gakugei University, 2 Communication and Media Lab., Graduate School of Human Sciences, Osaka University, 3 Ogawa Laboratories For Brain Function Research, Human Life Science Research Foundation Introduction How is positional information of space detected by light? Two basic principles using light absorption to give positional information are found. One is with light CT method by straight light by invention of Jobsis in 1977. Other is NIRS-Imaging by diffusion/scattering light discovered by Kato et al. in 1991. A basic principle of NIRS-Imaging decides positions of two probes and, by local temporal response of object between probes, samples photon functional pixel (PFP) which is the border unclearness that had time series". NIRS-Imaging uses the light that is ambiguity of positional information, and it is an idea of a reversal to locate it. NIRS-Imaging depends on local reaction nature of an object, and light CT method is a completely opposite principle. Jobsis-method sets distance between probes than 4.25cm to avoid the artifact from skull. Kato, et al. shortened it in 2.5cm boldly after the 14 years. Furthermore, by the present, we improved in the distance of 1cm and confirmed a fast-oxygen response in capillary event (FORCE) effect. Therefore, it is important how we arrange those probe in head in order to let precision of instrumentation improve. We established strip of paper which narrowed an interval of measurement locus of each channel to 1cm from conventional 3cm-shaped amount of selectivity alignment method in order to detect high reaction of precision more. Reproducibility was good and, by this method, detected reaction of a Broca’s area (Koike & Okawa, et al. 2003). we studied functional oxygen metabolic responses in and around the Wernicke’s area. Method Subjects were eight healthy adults. Recording device was NIRS-imaging system (Hitachi, ET-100). Thus a measurement locus interval seemed to become 1cm and, set strip of paper-shaped amount of selectivity alignment method (distance 2.5cm between probes), determination area at a range to rear 7cm from Broca’s area. The probes were placed on bilaterally covered Wernicke’s area. Sampling rate of data acquisition was 10Hz. An instruction of tasks with a phone condition or a non-phone condition of 2s was presented by mirophone. Subject hear their sounds. A video films all processes. We calculated the time course plots of that oxyHb, deoxyHb and totalHb from a start point in time of each sound as trigger to following 30 seconds. Result Transient alteration of totalHb which was different from a phone condition between non-phone conditions was observed in and around the Wernicke’s area. FORCE effect with increasing deoxy-Hb was observed in Wernicke’s area for a phone condition. Appearance of a Hb change is different from the left hemisphere in each PFP of the right hemisphere. These differences might reflect the distinct activities of the Wernickes which relate to the verbal function. Conclusion The our strip of paper-shaped amount of selectivity arrangement method was effective in activation detection of Wernicke’s area. References 1) Kato T, et al. Annual Report from Japanese Ministry of Health & Welfare Grantin 1992. 179-181. 2) Kato T, et al. 1993JCBFM 13: 516, . 3) Koike T, et al. 2003 NeuroImage 19:2:871 4) Ohkawa Y, et al. 2003 NeuroImage 19:2:1337

e192

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 119 An fMRI study of mental writing for Chinese John X. Zhang 1 , Xuchu Weng 2 , Chongyu Lin 2 , Zhuanwei Xiao 3 , Li Hai Tan 1 1 Joint Laboratories for Language and Cognitive Neuroscience, University of Hong Kong, Hong Kong, 2 Laboratory for Higher Brain Function, Institute of psychology, Chinese Academy of Sciences, Beijing 100101, China, 3 Molecule Imaging Research Center, Medical College of Shantou University, Shantou, China As a logographic script, Chinese has characters with square configurations. Each character is composed of a number of basic strokes arranged in a two-dimensional area. The phonology of each character cannot be deduced based on orthography. There is also a spelling system (pinyin) introduced about six decades ago so that the sound of each character can be spelled with the English alphabet. To dissociate the orthographical representation and phonological representations of Chinese characters, we presented Chinese subjects with sound recording of 2-character words and cued them to recall either the spatial configuration of the characters or their pinyin spelling. They then mentally wrote out either the configuration stroke-by-stroke (the character or orthography condition) or letter by letter (the phonology or pinyin condition). It is assumed that in writing out the strokes, participants accessed the orthographical representation more often and in writing out the pinyin spelling, they accessed the phonological representation more often. Ten native Chinese speakers participated. Twenty T2*-weighted EPI images were acquired while participants mentally dictate auditorily presented words. The words were presented for 1 seconds, followed by a 4-6 seconds blank interval. A cue then informed subjects to start writing. The cue would indicate whether they should write the phonology spelling (pinyin) of the words or their constituting characters stroke-by-stroke. Subjects pressed a key to indicate the end of writing. The design was an event-related design with the two types of writing trials randomly intermixed (12s ITI). SPM99 was used in imaging analysis. Areas showing significantly greater activation to pinyin relative to characters or the opposite were identified. Pinyin writing elicited greater activation than characters in left inferior frontal cortex (Fig. 1). Character writing elicited greater activation than pinyin in left middle frontal gyrus BA9 (Fig. 2). The left inferior prefrontal cortex activation for phonology is consistent with existing evidence of phonological processing of this area in English (Bookheimer, 2002). The middle frontal activation is consistent with existing evidence of finer spatial analysis in this region due to the unique configuration of Chinese characters (Tan & Siok, 2003). Our results provide evidence for the dissociation of orthographical and phonological representations in Chinese. References Tan, L. H. and W. T. Siok (2003). Handbook of Chinese Psycholinguistics. P. Li, L.H. Tan, E. Bates and O. Tzeng. Cambridge University Press. Bookheimer, S. (2002). Annu. Rev. Neurosci. 25: 15188.. Acknowledgement Supported by a Research Grants Council Central Allocation Vote (RGC CAV) group research grant (HKU 3/02C) awarded to L.H. Tan, a China National Natural Science Foundation grant (NNSFC 30128005) and a China Ministry of Science and Technology grant (G1999054000) awarded to X.C. Weng, and a Key International Collaboration Project Grant from National Natural Science Foundation of China (NSF F2003-79) awarded to Z.W. Xiao.

e193

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1

Figure 2

e194

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 120 The ventral occipitotemporal regions activated by viewing Chinese words and English words Wutian Zhang 1 , Yiyuan Tang 2 , Hsuan Chen 3 1 Institute of psychology Chinese academy of sciences, 2 Institute of Informatics Dalian Technology University, 3 The Chinese University of Hong Kong Introduction Visual object processing, including processing of letters and words, is mainly subserved by ventral areas comprising the occipitial and inferior temporal cortex. Recently some researches on the recognition of alphabetic words have proposed that middle fusiform gyrus of left hemisphere is responsible for the processing of prelexical visual word form[1]. Other researchers thought that fusiform area may activated by words meaning[2]. Chinese character is ideogram and it has two-dimensional form. Do Chinese words activate common location in the fusiform gyrus as alphabetic words? Methods High frequency, single Chinese character words and English word were chosen, pseudowords (no pronunciable and no meaning for Chinese) and nonwords also used as stimuli. Ten young right-handed, native Chinese speakers from university served as subjects. They were asked to view stimuli silently while a series of functional MRI images were collected. A crosshair presented in the center of the screen was served as baseline. A block design was used. BOLD effects were measured by a fMRI scan performance on a 1.5T Signa (GE) clinical scanner. 18 axial anatomic image were collected with echo sequence (TR/TE=2000/40ms, FOV=24°¡24cm2, thinkness=7mm£¨ 1mm inter-slice spacing, matrix=256°¡192). 100 functional images per slice were then acquired using single shot gradient-echo echo planar imaging (EPI) plus sequence (TR/TE = 3000ms/40ms, flip angle = 90_£¨matrix=64°¡64, thickness=6mm, 1mm inter-slice spacing). The sagittal anatomical image of a 60 slices was acquired using FSPGR sequence, with TR = 1100 ms TE=2100msec, ?=30°, slice thickness was 2.5 mm, acquisition matrix was 256×256. Data analysis was performed individually for each subject with SPM99 software. The statistic significance of task in comparison with fixation period was accessed using t-statistics. Correction for multiple comparisons was performed. Common activation across item types was examined using conjunction analysis in SPM99. Result and discussion The common activation areas during Chinese word and English word viewing relative to the baseline fixation are left middle frontal and left precentral gyrus, left superior parietal lobules, left middle occipital, bilateral inferior occipital and left fusiform gyrus. A few brain areas activated by English versus Chinese involving right middle temporal and right precingulate gyrus. The main difference of activation between Chinese and English psudowords are left and right fusiform gyrus activated by English psudowords and Left parietal is activated by Chinese psudowords. Left and right fusiform gyrus are activated by English words versus psudowords but not by Chinese word versus psudowords. The results of English words and psudowords are consistent with the idea that middle part of fusifom gyrus is the location of prelexical processing. But the results of Chinese words and psudowords suggest that the function of fusiform gyrus may connect to the processing of Chinese word meaning not to word form. Why the function of fusiform gyruse for processing Chinese words is different to processing English words. Its need study further. Reference Dehaene S. et al., (2002) NeuroReport 13 :321-325 Mechelli A. et al., (2003) Journal of Cognitive Neuroscience 15: 260-271 Thanks to the support of Chinese science foundation (30170321)

e195

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 121 Combined DTI and fMRI Application in Presurgical Evaluation of Patients With Brain Tumor Manzar Ashtari 1 , Mark Mittler 2 , Kenneth Perrine 3 , Alan Diamond 1 , Tana Clarke 4 1 North Shore Long Island Jewish Health System, LIJ Radiology Department, 2 Long Island Neurosurgical Associates, P.C., 3 Nroth Shore LIJ Health System, LIJ Neurology Department, 4 North Shore LIJ Health System, Zucker Hillside Hospital Objective To use fMRI in conjunction with DTI to evaluate human language centers for presurgical evaluation of patients with brain tumor. Materials and Method A strongly right-handed fifteen-year old male patient presenting with seizure was diagnosed with left posterior inferior tempro-parietal cavernous angioma. Based on tumor location and its vicinity to language centers of the brain, we performed fMRI tasks engaging these areas such as naming and reading. In addition to the two fMRI tasks we obtained a 25 direction DTI sequence to evaluate the fiber tracks around and within the tumor volume. SPM96 was used for analysis of fMRI and the FA images were automatically reconstructed on the console (GE Medical Systems, Milwaukee).The reading task was composed of four blocks of unfinished sentences interleaved with four blocks of letter strings and eight blocks of rest as control. Subject was asked to finish the sentences and find the appropriate letter for the letter strings. The naming task was composed of active blocks where the subject was shown simple objects to name. The control blocks were composed of nonsense objects and rest. All paradigms were programmed in Eprime (Psychology Tools, Pittsburgh) and presented to the subject using a MR compatible goggles (Resonance Technology, California). Results Reading minus letter (R-L) show significant activation of the left posterior temporal cortex extending into inferior parietal lobule. However, the activation was spread more posteriorly than was obtained in normal control subjects. The FA images showed intact fiber bundles not disturbed by the tumor. Fig.1 shows a sagittal view of the FSE and FA image (Fig.2) with R-L activation superimposed. The activation areas of the naming are normally found to be in the left inferior junction of the temporal and parietal lobule. The main activation area for Naming Object (N-O) in this subject was in the right tempro parietal junction. Fig. 3 shows a FSE and corresponding FA axial slice (Fig. 4) depicting the activation areas of N-O contrast and track formation in and around the tumor area. Conclusions Our results showed no track formation in or around the tumor volume in the area responsible for naming. FMRI results independently confirmed the finding by showing activation for N-O contrast on the opposite side. On the other hand the fMRI and DTI both showed an intact area for reading on the same side as the tumor. Based on these findings we predicted that the patient would have no language deficits following a surgical resection of the tumor. The recent patients seven month follow-up exam continued to find the patients language functions intact as was predicted by our study. In conclusion, combining the DTI and fMRI provides powerful information that aids neurosurgeons in their approach for a successful surgery.

e196

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Reading minus Letter contrast showing an activation area in the posterior aspect of left temporal lobe.

Reading minus Letter activation superimposed on the FA image depicts the intact fiber bundels for reading area.

Naming minus Object contrast showing a greater extent of activation in the right inferior tempro-parietal lobule.

e197

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Naming minus Object contrast superimposed on diffusion tensor image showing no activation in the tumor area and no track formation.

e198

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 122 Prefrontal cortex involvement on strategies used in the phonemic verbal fluency task Demis Basso 1 , Patrizia S. Bisiacchi 2 , Lavinia Vitale 3 , Martin Lotze 4 , Marta Olivetti Belardinell 1,3 , Niels Birbaumer 4,5 1 ECONA, Interuniversity Center for Research on Cognitive Processing in Natural and Artificial Systems, Italy, 2 Department of General Psychology, University of Padua, Italy, 3 Department of Psychology I, University of Rome "La Sapienza", Italy, 4 Institute of Medical Psychology and Behavioral Neurobiology, University of Tuebingen, Germany, 5 Institute of Cognitive Neuroscience, University of Trento, Italy Introduction and Hypothesis: The phonemic verbal fluency test is used as an index of frontal functionality because it demands to produce words that respond to a criterion, spontaneously, without repetitions and, above all, it primes mechanisms of planning in the search of words. Therefore, beside the so-called fluency component, this test involves: working memory, long term memory (LTM), response inhibition, selective attention and shifting, planning and control processes. Impaired verbal fluency has been shown to be associated with frontal lobe damage. Brodman area (BA) 9 is thought to be responsible for the planning of the word search in LTM storage. We hypothesized that repetitive transcranial magnetic stimulation (rTMS) of BA 9 would induce differences in the generation of words which is based predominantly on search strategies. Method: Subjects were asked to produce as many words within a minute as possible, beginning with a given letter. Three letters were provided: C, P, and S. The performance of 250 healthy, 30 frontal-TBI, and 30 healthy subjects under repetitive transcranial magnetic stimulation was compared. In order to assess the role of BA 9 in the process, rTMS at the intensity equal to the motor threshold of the thumb muscles was performed over Fz at a frequency of 1 Hz. Results: TBI subjects performed poorly in respect to both healthy and rTMS subjects: significant differences were observed in the number of produced words (F(2,31)= 56.94) and the number of strategies used (F(2,31)= 20.74). Moreover, the 30 healthy subjects that performed the task with 1Hz rTMS stimulation showed a lower variability in the use of strategies with respect to healthy subjects. Furthermore, the number of strategies used was increased (mean: 5.93 respect 4.22, in particular the simple phonological and the grammatical strategies) but the number of word produced was not significantly different. Discussion: It seems that frontal TBIs’ problem in this test is not due to a limited vocabulary but to a difficult retrieval of the words, due to lack in research strategies. It was proven that the BA 9, stimulated by rTMS was related to strategy utilization, but the selected frequency resulted in a better performance.

e199

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 123 Functional imaging of familial dyslexia Simona M Brambati 1 , Cristiano Termine 2 , Milena Ruffino 1 , Massimo Danna 5 , Giacomo Stella 3 , Stefano F Cappa 1, 4, 5 , Daniela Perani 1, 4, 5 1 Vita-Salute San Raffaele University, Milan, Italy, 2 Child Neuropsychiatry Unit, University of Insubria, Varese, Italy, 3 Psychology Institute, University of Urbino, Urbino, Italy, 4 IBFM, CNR, Segrate Milan, Italy, 5 Scientific Institute H.San Raffaele, Milan, Italy Developmental dyslexia might present as an hereditable neurobiological syndrome (familial dyslexia). Phenotypic profiles are very heterogeneous among dyslexics at different developmental stages. For example, compensated dyslexics obtain normal scores on reading tasks, but they might complain of difficulties and discomfort in reading, and frequently show associated impairments in meta-phonological skills and in short-term memory. We used fMRI to investigate brain activation in familial dyslexics during reading tasks. Subjects: Dyslexic subjects were recruited from families with a proband with diagnosis of developmental dyslexia according to ICD-10 and at least a first-grade relative with clinically evident or compensated dyslexia. The control group was composed by subjects with no history of reading disabilities. An extensive neuropsychological battery assessing reading, phonological awareness, short-term memory, language comprehension, visual abilities and intelligence level was administered to all participants. Tasks: The experimental design consisted in two activation conditions (covert reading of words and pseudowords) and a baseline (false font string observation). To be sure that the subjects effectively performed the tasks in the scanner, they were invited to press a button every time a pre-established distracter appeared on the screen: distracters were pseudowords in the word reading condition, words in the pseudowords reading condition, and strings containing an ascendent letter in the false font string obsevation condition (Fig. 1). The stimuli were presented for 4000 msec, with an inter stimulus interval of 500 msec. Scanning Procedures: 154 echo planar volumes for each session were acquired using an echo planar imaging (EPI) gradient echo sequence (TR = 3000 msec., number of slices = 24, thickness = 6 mm). All the image-processing procedures and statistical analysis were performed using SPM2 (Wellcome Department of Cognitive Neurology, London, UK). Results: Most of dyslexic subjects had impairments in phonological awareness and short-term memory tests. FMRI behavioral results showed that dyslexics were significantly slower than controls in detecting the distracters during the reading conditions but not during false font string observation. In controls, a prevalent left hemispheric system was activated during reading task (words and pseudowords) as shown in fig. 2a: activation foci were mainly located in middle and inferior frontal gyrus, posterior insula, inferior parietal lobule, superior and middle temporal gyrus, fusiform gyrus and cerebellum. hom*ologous areas were also found to be active in the right hemisphere at the level of inferior and middle frontal gyrus, and middle temporal gyrus. Familial dyslexics showed a bilateral pattern (fig. 2b) with activation foci mainly located in the inferior frontal gyrus, inferior parietal lobule and cerebellum in the left hemisphere, middle inferior frontal gyrus, supramarginal gyrus and superior temporal gyrus in the right hemisphere. Conclusions: The main difference in the pattern of brain activation during reading, between familial dyslexics and controls, was the lack of activation in the former in crucial areas of the reading network located in the temporal lobe. These findings indicate that familial dyslexia is associated with a disruption of the same neurobiological system which has been found to be abnormal in sporadic and compensated dyslexia during phonological tasks.

e200

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig. 1 Example of false font strings containing (a) and non-containing (b) ascendant letters.

Fig. 2 Comparison between reading conditions (words and pseudowords) and baseline in controls (a) and dyslexics (b). (Second-level analysis, p<0.01, minimum cluster size = 10 voxels)

e201

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 124 FUNCTIONAL MRI OF LANGUAGE AND EEG DISCHARGES IN EPILEPSY PATIENTS WITH MALFORMATIONS Regula S Briellmann , Graeme D Jackson , Anthony B Waites , John S Archer , Paolo Federico , Angelo Labate , Richard Masterton , David F Abbott Brain Research Institute, Melbourne, Australia Introduction: Refractory epilepsy can be caused by malformations of cortical development (MCD). Dysplastic tissue can generate seizures (pathological function), but at least with some MCDs it can also be involved in physiological function. Here we assess with functional MR both pathological function (activation associated with epileptiform EEG discharges) and physiological function (activation with language tasks) in patients with different types of MCD. Subjects: 25 patients with MCD were assessed with fMRI language, seven of these had frequent interictal EEG discharges and had fMRI/EEG. Nine patients had focal cortical dysplasia (one with fMRI/EEG), eight had a dysembryoplastic neuroepithelial tumour (DNET, 2 with fMRI/EEG), five had polymicrogyria (two with fMRI/EEG), two had band heterotopia (both with fMRI/EEG) and one had tubero-sclerosis (TS). MR methods: MR imaging was performed on a GE Signa 3T scanner. Functional sequences used Gradient-Echo Echo-Planar Imaging (EPI) with whole brain coverage (22 axial slices, 4mm thick, 1mm gap), 128 x 128 matrix, 24cm x 24cm FOV, 60º flip angle, TE 40ms, and TR 3000ms. Analysis of the imaging data was performed in SPM99 and iBrain®. For language activation, a noun-verb generation (NVG) and an orthographic lexical retrieval (OLR) task were used, both in a block design with visual stimulus presentation and covert response. Language lateralisation was determined on the number of activated pixels in language-related areas (laterality index, LI), and compared with 30 controls. For fMRI/EEG, EEG was recorded from eighteen non-metallic scalp electrodes with carbon fibre leads, using the conventional 10-20 EEG format. Event-related analysis was performed. Resultant statistical parametric maps are displayed at a threshold of P<0.05, corrected for multiple comparisons. Results: Of the 25 subjects, only three showed right-lateralised language. Average LI was for NVG 0.53 ±0.3, and for OLR 0.48 ±0.3 (controls: 0.6 ±0.3 for both tasks). Five of the 6 patients with bilateral MCD had left lateralised language. Language lateralisation in left temporal lobe MCD (OLR, n=7, mean LI 0.49 ±0.3) was not different from patients with left-sided hippocampal sclerosis (n=10, mean LI 0.45 ±0.4). Only patients with DNET or TS did not activate dysplastic tissue, whereas the other MCDs generally showed language associated activation in dysplastic tissue, if this was in eloquent areas (figure). Four of the seven subjects with fMRI/EEG had discharges during scanning, of these two had band heterotopia, one had polymicrogyria and one had focal cortical dysplasia. Discharge related BOLD signal change in all four studies was found in dysplastic cortex, but this did not overlap with language related activation. Conclusion: Left-lateralised language activation is generally preserved in MCD. Dysplastic cortex can be involved in complex physiological functions (language), but also generate seizure discharges.

e202

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Top panel Language fMRI (SPM analysis) showing activation in warm colors, and deactivation in cold colors. The pattern of cortical activation is similar to a healthy control except for additional activation in band heterotopia (*). Bottom panel fMRI/EEG (SPM analysis) of the same subject, based on 12 discharges, captured in 30 minutes recording. Activation is present in the mesial frontal / anterior callosal area and in band heterotopia (*).

e203

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 125 Silent word reading in dyslexic children: an fMRI study Muriel Brun 1 , Manuel Bouvard 2 , Jean-François Chateil 1 , Gaëlle Benichou 2 , Martine Bordessoules 1 , Michèle Allard 1, 3 1 Fédération dImagerie Médicale. CHU Pellegrin, Bordeaux, France., 2 Psychiatrie des Enfants, CHU Pellegrin, Bordeaux, France, 3 Laboratoire dimagerie moléculaire et fonctionnelle ERT CNRS. Bordeaux. France Purpose: A previous study highlighted that dyslexic children exhibited different activation patterns than control subjects during phonological decoding tasks. Phonological analysis has also been reported to be involved in silent reading of real words by skilled readers. Our purpose in this present fMRI study is to explore dyslexic children during two reading tasks. Subjects and methods: Dyslexic and control children were included. All subjects performed two reading tasks adapted to fMRI conditions. Subjects had to decide if the written word corresponded to the image shown on a screen. The discrepancy could be orthographic (task 1) or semantic (task 2). 24 slices with a thickness of 3.5 mm were acquired on a 1.5 T Philips scanner using GE-EPI sequence (FOV 230 mm, 64 x 64 mm, TE= 55 ms, TR = 2400 ms). Data were analyzed with Spm 99. Activation maps were color-coded according to statistical significant difference activation to rest. Individual and group analyses were performed, group results were presented and performances were correlated to statistical activation maps. Results: fMRI was reliably obtained in 17 subjects (10 dyslexics and 7 controls). The high response rate, for all subjects, allowed us to consider a high participation level and to take into account obtained activation maps. Performances were worse in dyslexics than in controls since dyslexics presented more errors and a longer mean delay response. In task 1, not only did dyslexics presented larger bilateral activation than controls but they also recruited additional areas such as BA 9-46, anterior cingulum, inferior frontal gyrus (IFG), anterior insula and cerebellum (figure 1). In task 2, controls presented the same activation pattern than in task 1. Dyslexics exhibited larger activation than in task 1 with recruitment of SMA but without activation in the anterior cingulum (figure 2). Conclusion: In task 1, the discrepancy is identified at a visual level and rapidly treated by visual ventral pathways for controls. On the opposite, this discrepancy is not so easily identified in dyslexics, as revealed by the activation of the anterior cingulum reported to be involved in conflict monitoring,. A recruitment of the articulatory loop (IFG and anterior insula) is also needed to pick the error out. Task 2 also appeared quite easy for controls since the temporo-occipital cortex is mainly recruited. Anterior language areas are activated to search for the true semantic concordance in dyslexics. As the discrepancy occurs in the same categorization field, they do not have to monitor a conflict as in task 1, as revealed by the absence of activation in the anterior cingulum. References: Picard et al Imaging the premotor area Current Opinion in Neurobiology 2001, 11:663-672 Gaillard et al Cortical localization of reading in normal children Neurology 2001, 57: 47-54 Simos et al Cerebral mechanisms involved in word reading in dyslexic children: a magnetic source imaging approach Cerebral cortex 2000, 10:809-816

e204

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Fig 1: location of statistical activation in controls (top) and dyslexics (bottom) during task 1

Fig 2: Task 2 controls (top) and dyslexics (bottom)

e205

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 126 Intraopaerative Cortical Mapping of Bilingual Representations by ESM, iOIS, and iNIRS. Andrew F. Cannestra 1,2,3 , Michael Guiou 2,3 , Linda Liau 1 , Susan Bookheimer 3 , Arthur Toga 2,3 1 UCLA Division of Neurosurgery, 2 UCLA Laboratory of NeuroImaging (LONI), 3 UCLA Brain Mapping Studies using ESM, PET and lesion mapping of bilingual brains have revealed conflicting maps of distinct versus overlapping cortical language representations. Often the age of acquisition and fluency are considered factors determining the extent of overlapping language sites. In this study we mapped 5 patients undergoing surgical resection for neurosurgical pathology. Informed consent was obtained prior to mapping. Subjects underwent neurolinguistic testing and fMRI prior to surgery. Once the cortex was exposed, electrocortical stimulation mapping (ESM), intraoperative optical imaging of intrinsic signals (iOIS), and intraoperative near infra-red spectroscopy (iNIRS) were used to map cortical language representations. All subjects were fluent in their second languages. This was defined as daily use of their second language with little to no effort in comprehension (spoken or written), conversation, and media (radio, TV etc). Age of second language acquisition/fluency varied from birth to age 18. Intraoperatively, subjects were awakened after craniotomy and reflection of the dura. Frontal (Brocas) and /or temporal-parietal (Wernickes) language regions were exposed for testing. Language mapping was performed using a visual object naming task with figures from the Boston Naming Test. Electrocortical stimulation mapping used standard clinical techniques in first and second languages combined with continuous electrocorticography (ECog). iOIS and iNIRS were then performed using the same testing protocols in a block paradigm design. In all subjects ESM, iOIS and iNIRS provided cortical maps of language. Cortical language maps varied from complete overlap (a single cortical representation containing both languages) to complete separation of individual language representations. In subjects where ESM maps provided distinct regions corresponding to individual languages, iOIS and iNIRS also provided distinct language activation maps (i.e. no-overlap). Alternating language paradgims confirmed these regions to be distinct. However, in cases where ESM only identified one cortical language area for both languages (total overlap), iOIS and iNIRS were still able to resolve language specific differences within this cortical region. When bilingual language maps were compared to age of acquisition, no correlation was observed between age of language acquisition, fluency and the degree of separation/overlap of language maps. In this study we defined the bilingual cortical representations (activation and disruption maps) of 5 subjects undergoing neurosurgical procedures. Bilingual maps varied from overlap of first and second languages to complete separation of cortical representations. Unlike previous studies, the degree of overlap and/or separation in our multi-modality study was independent of age of acquisition of language or fluency.

e206

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 127 fMRI evaluation of hemispheric language dominance using various methods in laterality index calculation Pavel Chlebus 1,2 , Michal Mikl 1 , Milan Brázdil 1 , Marta Paourková 2 , Petr Krupa 2 1 1st Department of Neurology, Masaryk University, St. Annes University Hospital, Brno 656 91, Czech Republic, 2 Diagnostic Imaging Clinic, Masaryk University, St. Annes University Hospital , Brno 656 91, Czech Republic Introduction Several functional MR imaging studies evaluating the lateralization of linguistic function in patients who underwent Wada testing have been reported. In these studies, a calculation of the language lateralization index (LI) was defined, and the results were compared with those of the Wada test. There was usually a significant correlation between the results from the two modalities. However, it remains unclear whether LI is a reliable predictor of hemispheric language dominance in individual subjects. There is extensive variance in the LI calculation across the studies, and the optimal calculation method remains unclear. We have attempted to calculate LI in different ways in the same subjects, in order to find the LI calculation method with the highest correlation to the Wada test. Methods Wada tests and functional MRI were performed on 12 patients (4 males, 8 females) suffering from intractable temporal lobe epilepsy. We chose a silent form of verbal fluency task (VFT) as a language paradigm. ROIs, each with a known association with language function (Brocas area, the lateral prefrontal cortex, the medial prefrontal cortex, the cingulate gyrus, the cerebellum, and the anterior two-thirds of the hemisphere), were defined. For each ROI, the numbers of suprathreshold voxels (with t=2.5, 3, 3.28, 4 and 4.5) were counted in the left (L) and right (R) hemispheres. LI was calculated by a previously reported method: LI =(L-R)/(L+R)*100. The values of L and R were filled in for each individual ROI and also for their various combinations. At first, LIs were calculated from a simple suprathreshold voxel count. Next, we used t-weighted voxels (i.e., each voxel was weighted with a corresponding t-value or with a t-value function). In order to increase the possibility of correlation, the linguistic functions (comprehension, reading, object and picture naming, repetition, and spontaneous speech) were scored separately. The Wada LI (WLI) was obtained by the formula: WLI=(L-P)/6*100. We used Spearman Rank Order Correlations to compare fMRI-derived LIs and corresponding Wada test results (WLI). Results LIs calculated from the majority of the ROIs, individually and in combination, significantly correlated with Wada test results (WLI). The most significant correlation was proven in the LIs that were evaluated from the lateral prefrontal cortex and from the anterior two-thirds of the hemisphere (R=0.92, p=0.00002). The most appropriate statistical threshold to obtain the data for LI calculation appeared to be t=3.28 (p=0.001). The only exception was in the cerebellum, where t=4.5 was proven as the most profitable threshold (p=0.000024). Conclusions Standard LI calculation from the number of activated voxels within the anterior two-thirds of the hemispheres and alternatively within the both-sided lateral prefrontal cortices correlated better with hemispheric language dominance (as revaled by Wada test) than the other methods used. New LI calculation methods (combination of ROIs, statistical weighing of voxels) used in the present study seem to be inferior to the standard calculation. Further research on this issue is nevertheless suggested.

e207

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 128 Remote effects of aphasic stroke on the ventral temporal cortices. Jenny T Crinion , Cathy J Price Fil, The Wellcome Department of Imaging Neuroscience, 12 Queens Square, London WC1N 3Bg, UK. Previous functional imaging and cortical stimulation studies of epileptic patients have proposed a role for the left ventral temporal cortices in semantic processing 123 . Using fMRI we investigated the remote effects of left middle cerebral artery (MCA) infarction on the structurally intact ventral temporal cortices in a group of aphasic patients with intact simple speech comprehension. Twenty two age-matched healthy control subjects and 8 chronic aphasic patients, following left MCA infarction were investigated. All patients had documented evidence of impaired speech comprehension acutely post stroke but, at the time of testing, intact speech comprehension. Four of the patients had left dorsal temporal lobe infarction sparing the frontal lobe. Four patients had frontal lobe infarction sparing the temporal lobe. A Siemans 1.5T scanner was used to acquire 360 T2*- weighted echoplanar images, with a TR of 3.15 seconds. We presented each subject with blocks of stories (22 seconds each), alternating with blocks of reversed stories. The stories, spoken by a female speaker, were comprised of high imageability and high frequency words within syntactically simple sentences. Subjects were instructed to listen carefully. Data were analyzed with SPM2 using standard procedures. We analyzed each subject independently to identify activation for stories relative to baseline. A second level ANOVA then modeled the effects of (i) controls; (ii) patients with temporal lobe lesions; and (iii) patients with frontal lobe lesions. In the control group, simple stories were associated with symmetrical activation of both the anterior-lateral superior temporal and ventral temporal cortices. There was asymmetry of function in left posterior superior temporal and the left dorsal frontal cortices. Both structure and function in the superior temporal cortices, ipsilateral and contralateral to the lesions was preserved in all patient groups. In contrast, bilateral ventral temporal lobes were activated significantly less in both patient groups than in the control group (x= -34 y= -30 z= -24; Z-score = 4.5; x=30 y=-36 z=-23; Z-score= 3.64). These abnormal effects survived a correction for multiple comparisons at the cluster level yet they occurred in regions that were structurally preserved in all patients. There was no region of increased activation in patients relative to the controls; and no significant group differences were observed in the temporal and frontal patients pattern of activation. Additional to the area of destruction, there was reduced activation bilaterally for the patients in the ventral temporal cortices irrespective of the lesion site. Thus, speech comprehension of simple narratives following aphasic stroke is more likely to be attributable to persevered anterior superior temporal responses than to the function of the ventral temporal lobe. This study highlights the remote bilateral functional effects of left MCA infarction on the structurally intact ventral temporal cortices. The implication is that function of the ventral temporal cortices is not necessary for the mapping of simple speech onto meaning. Further investigation into the role of the superior left and right temporal cortices will inform the neuronal system involved in speech comprehension. References 1 Wise et al. Neuropsychologia 38 (2000) 985- 994. 2 Nobre et al. J Neuroscience 1995; 15:1090-8. 3 Schaffer et al. Epilepsia 1994; 35; 525-8.

e208

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 129 Neural Correlates of Syntax, Semantics, and Discourse Monitoring: An fMRI Study in Children with Autism. 1 Department

Mirella Dapretto 1 , Audrey T. Wang 2 , Susan S. Lee 1 , Rochelle Caplan 1 of Psychiatry and Biobehavioral Sciences, 2 Department of Psychology, UCLA Brain Mapping Center, Los Angeles, CA

High-functioning individuals with autism spectrum disorders (ASD) typically show a disparity between their formal linguistic skills (e.g., phonology and syntax) on one hand and sociopragmatic impairments (e.g., conversational discourse) on the other. The goal of this study was to examine the neural representation of these relative strengths and weaknesses in the autistic brain. Data from eight high-functioning children with ASD were compared to those previously collected from a sample of eight typically developing children. Both groups performed a sentence judgment task [1] tapping syntactic and semantic processing, as well as a discourse monitoring task tapping the ability to detect discourse coherence [2]. In the sentence judgment task, subjects listened to pairs of sentences and had to decide whether their meaning was the same or different. In two different activation conditions, this judgment rested on either processing the underlying syntactic structure of the sentences or their word meanings. In the discourse coherence task, subjects listened to short question-answer dialogues between two people and had to determine whether the respondent’s answers made sense. In two different activation conditions, this judgment rested on detecting either a break in topic maintenance or in the underlying logic of the answers. All subjects were scanned using a GE 3T magnet with an ANMR upgrade for echo planar imaging. Functional images were acquired over 19 axial slices (TR=3000ms, TE=23ms, flip angle=90 o , matrix size=64x64, FOV=20cm, 4mm slices/1mm gap) along with a set of coplanar, high-resolution structural images (TR=4000ms, TE=65ms, flip angle=90 o , matrix size=128x128, FOV=20cm). Each subject’s data were realigned to correct for head motion, spatially normalized, and smoothed with AIR [3]. Group analyses were then conducted with SPM99 using a 6-sec delayed box-car reference function. Response times and accuracy data, acquired during the functional scans revealed no behavioral differences between groups. Across tasks, both groups recruited canonical language areas in frontal and temporal regions. However, for the sentence judgment task, the normal group showed greater activity than the ASD group in the cerebellum during the syntax condition (Fig. 1), in Wernicke’s area in the semantic condition (Fig. 2), as well as in Broca’s area and the anterior cingulate during both sentence judgment tasks. With regard to the discourse coherence task, the ASD group showed less distinct lateralization profiles in response to the demands of reasoning vs. topic maintenance than the normal group (see Fig. 3 & 4). Region of interest analyses were conducted to assess the reliability of the observed group differences. These preliminary results suggest that children with ASD in part recruit different neural networks than typically developing children during both more basic levels of sentence processing and more complex linguistic functions such as discourse monitoring. Supported by NIDCD grant #DC5159-02, the Cure Autism Now Foundation, and the UC Davis MIND Institute. References 1. Dapretto & Bookheimer (1999). Neuron, 24, 427-432. 2. Caplan & Dapretto (2001). Neuroreport, 12, 3625-3632. 3. Woods et al. (1999). Human Brain Mapping, 8, 73-79.

e209

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e210

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e211

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 130 Implication of the left dominant premotor cortex in language: an electrostimulation study Hugues DUFFAU 1 , Peggy GATIGNOL 2 , Laurent CAPELLE 1 1 Department of Neurosurgery and Inserm U494, Hôpital Salpêtrière, 47-83 Bd de l’hôpital, 75013 Paris, France. Email: [emailprotected], 2 Department of Neurology, Hôpital Salpêtrière, 47-83 Bd de l’hôpital, 75013 Paris, France Objective Although the role of the premotor cortex (PMC) was widely studied in motor function, very few data are currently available about the participation of this structure in language (1,2). The present work reports a series of 28 patients harboring a brain tumor near or within the left dominant PMC, operated on under local anesthesia with intraoperative sensorimotor and language mapping using electrical stimulations. On the basis of the functional findings collected during surgery, the role of the left PMC in language is discussed. Patients and method Twenty-eight right-handed patients (13 males, 15 females, mean age: 37 years) with a normal clinical examination were operated on for a glioma revealed by seizures and located near or within the left PMC. All surgical procedures were conducted in awake patient, in order to perform an intraoperative real-time functional mapping, using the method of direct cortico-subcortical electrical stimulations (biphasic current, 1 msec/phase, 60 Hz, 2 to 8 mA, OCS1 Radionics*), previously described by the authors (3, 4). Language tasks performed by the patient all along the resection, consisted in counting and picture naming (preceded by reading a short sentence) (5). Results Stimulations of the left PMC induced transient speech disturbances in all patients, namely with disruption of both counting and reading/naming during stimulation of the ventral PMC due to the elicitation of a complete anarthria while generating a pure anomia during stimulation of the dorsal PMC. Moreover, corresponding subcortical pathways generated the same language disorders than at the cortical level when stimulated all along the glioma resection. Eloquent structures were systematically preserved, allowing to avoid postoperative sequelae. Conclusions These findings suggest the following points: - the left dominant PMC seems to play a major role in language, since inducing language disruption when stimulated ; - this structure seems to have a well-ordered functional organization. Indeed, the ventral PMC is likely involved in planification of articulation, explaining the anarthria elicited during cortical and subcortical stimulations. The dorsal PMC could be more implicated in naming network, since counting (and even reading of a short sentence) was still possible despite the electrical stimulations. These symptoms should be distinguished from reduction of spontaneous speech observed following stimulations and/or resection of the dominant supplementary motor area (SMA)(5). A multimodal study of this complex area (combining the data obtained using pre- and post-operative functional neuroimaging with those obtained by electrical brain mapping) is needed, to better understand its integration in language networks, in particular its relationship with the other structures involved in speech production such as the left dominant insula (6), primary motor area and SMA. References 1. Grafton ST, et al. Neuroimage 1997;6:231-236 2. Wise RSJ, et al. Lancet 1999;353:1057-1061 3. Duffau H, et al. Ann Neurol 2000;47:132-135 4. Duffau H, et al. Brain 2002;125:199-214 5. Duffau H, et al. J Neurol Neurosurg Psychiatry 2003;74:901-907 6. Duffau H, et al. NeuroReport 2001;12:2159-2163

e212

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 131 Text comprehension after brain injury: An fMRI study of inference processes in patients with anterior temporal or fronto-medial lesions 1 Max-Planck-Institute

Evelyn C. Ferstl 1,2 , D. Yves von Cramon 1,2 of Human Cognition and Brain Sciences, Leipzig, 2 Outpatient Clinic for Cognitive Neurology, University of Leipzig

Text comprehension requires the continuous integration of utterances with both general world knowledge and the prior discourse context. These inference or coherence building processes have been shown consistently activate the left fronto-medial cortex and anterior temporal regions bilaterally (Mazoyer et al., 1993; Ferstl & von Cramon, 2001, 2002). However, in behavioural tests, there was evidence for an inferencing deficit after left frontal lesions only, but not after anterior temporal lesions (Ferstl, Guthke & von Cramon, 2002). The goal of the present study was to characterize the network necessary for higher level language processing by describing the changes caused by lesions to one of the key components. We used event-related whole head fMRI at 3 Tesla to directly study brain damaged patients’ cerebral blood flow during an inferencing task. The paradigm closely resembled that of the aforementioned previous studies. 80 sentence pairs were auditorily presented over head phones. The participants were instructed to judge using a YES/NO response whether there was a pragmatic, content-based connection between the sentences. Pseudo-word sentences were used as the baseline. The brain injured participants were selected from a pool of former patients of the outpatient clinic for cognitive neurology at the University of Leipzig. Exclusion criteria were cognitive or medical deficits preventing the patient from participating in the scanning session or from being able to listen to the sentences. For the fMRI analysis participants with more than 25% errors were excluded. 16 patients’ data have been analysed so far: 10 patients with anterior temporal lesions (n=6 left temporal, n=4 right temporal), and 6 patients with a likely involvement of the FMC. These latter patients were closed-head injury patients whose anatomical MRI scan showed bifrontal contusions (n=3) or evidence for microbleeds along the superior frontal gyrus (n=3). The behavioral data showed no systematic differences between the groups in reaction times or the distribution of errors across conditions. The response times varied between 220 ms and 1500 ms after the offset of the target sentence, indicating that the task was appropriate for the patient population (Price & Friston, 2002). The fMRI data were first analysed for the whole group. Comparing all language trials to the control task, we found very similar activation patterns as in the previous studies with healthy control participants. As expected, the FMC group had bilateral involvement of the anterior temporal lobes, whereas for the temporal groups this regions was active in the contra-lesional hemisphere only. Furthermore, the patient group showed large regions of activation in the left-sided prefrontal cortex, different from the previous control studies. In particular, there was activation in the pars triangularis of the inferior frontal gyrus and near the junction of the inferior frontal sulcus and the precentral sulcus. The fronto-medial activation depended on the patient’s performance level. Further analyses will focus on the single cases. The individual’s activation patterns are characterized with respect to the particular patient’s lesion location, performance pattern and neuropsychological profile.

e213

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 132 The role of the left dominant striatum in language: a study using intraoperative electrical stimulations Santiago GIL ROBLES 1 , Peggy GATIGNOL 2 , Laurent CAPELLE 3 , Hugues DUFFAU 3 of Neurosurgery, Hospital Clinico San Carlos, Universidad Complutense De Madrid, C Profesor Martin Lagos, 28040 Madrid, Spain, 2 Department of Neurology, Hôpital Salpêtrière, 47-83 Bd de l’hôpital, 75013 Paris, France, 3 Department of Neurosurgery and Inserm U494, Hôpital Salpêtrière, 47-83 Bd de l’hôpital, 75013 Paris, France. Email: [emailprotected] 1 Department

Objective While the role of the striatum is currently better understood in motor function, very few data are currently available about the implication of this structure in language (1,2). In the present work, the authors used intraoperative electrical stimulations during surgery of tumors involving the caudate nucleus and/or putamen in the dominant hemisphere. The goal was both to study the potential role of these structures in language and to avoid postoperative definitive aphasia. Patients and method Seven patients (4 males, 3 females, mean age: 37 years; 6 right-handed, 1 left-handed) with a normal clinical examination except a slight dysarthria in one case, were operated on for a glioma revealed by seizures and located near or within the dominant striatum: 4 gliomas invaded the dominant insula and putamen, 3 gliomas involved the dominant fronto-mesial precentral structures and the head of the caudate. All surgical procedures were conducted in awake patients, in order to perform an intraoperative real-time functional mapping, using the method of direct cortico-subcortical electrical stimulations (biphasic current, 1 msec/phase, 60 Hz, 2 to 8 mA, OCS1 Radionics*), previously described by the authors (3, 4). Language tasks were performed by the patient all along the resection, consisting in counting and picture naming (preceded by reading a short sentence) (5). Results In the 4 cases of glioma involving the dominant putamen, stimulations systematically induced a transient anarthria, while in the 3 cases of glioma involving the dominant caudate nucleus, each stimulation elicited reproducible perseveration. Consequently, the striatum was systematically preserved. Postoperatively, all patients except one presented a transient dysphasia, which always resolved within 1 to 3 months. Conclusions On the basis of these findings, we suggest the existence of two separated basal ganglia systems in language: - one mediated by the putamen (e.g. the « sensorimotor » part of the striatum, connected to the sensorimotor cortex), which may have a « motor role » in language, explaining why its direct stimulation inhibits articulatory sequences thus elicites anarthria ; - one mediated by the head of the caudate nucleus (e.g. the « associative » part of the striatum, connected to the prefrontal cortex), which may have a role of « cognitive control », explaining why its direct stimulation induces a failure to inhibit previously learned responses thus generating perseveration. Such a knowledge could have important implications in the surgical strategy of lesions involving the dominant striatum. References 1. Moro A, et al. Neuroimage 2001; 13: 110-8 2. Ullman MT. Nat Rev Neurosci 2001;2: 717-26 3. Duffau H, et al. Ann Neurol 2000;47:132-135 4. Duffau H, et al. Brain 2002;125:199-214 5. Duffau H, et al. J Neurol Neurosurg Psychiatry 2003;74:901-907

e214

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 133 Natural vs. grammatical gender an fMRI study on lexical processing in normal language and aphasia Marion Grande 1,2 , Francesca Longoni 1,2 , Verena Hendrich 1,2 , Frank Kastrau 2,3 , Walter Huber 1 1 Neurolinguistics at the Department of Neurology, University Hospital Aachen, 2 Interdisciplinary Centre for Clinical Research, 3 Department of Neurology, University Hospital Aachen Psycholinguistic models of lexical processing assume different levels of processing for concepts and word forms. Levelt et al. (1999) postulate an additional level of processing, the so-called lemma-level, for grammatical features like grammatical gender or word category. Neuroimaging studies on word form processing reported activation in Broca’s area and in the supramarginal gyrus whereas studies focussing on processing of lemma information (mainly grammatical gender) identified Broca’s area (e.g. Heim, Opitz, & Friederici, 2003). The aim of the present study was to determine the neural substrates of natural (concept information) and grammatical (lemma information) gender and of word form processing. 14 young, right-handed, native speakers of German were included in an fMRI study consisting of two different decision tasks one focussing specifically on natural gender and the other on grammatical gender. Discrimination between words played backward and complex sounds served as control task. Additionally we examined a male, 47-year-old patient, who had suffered a left MCA infarction and presented with moderate Broca’s aphasia with very good comprehension 19 months post onset. For the normal subjects we found the following specific activations for each experimental condition vs. the control task: - natural gender task: anterior part of BrocaŸs area (BA 45) (-44, 20, 21) and frontal operculum (BA 47) (-51, 39, -2), - grammatical gender task: posterior part of BrocaŸarea (BA 44) (-44, 12, 14), The patient showed bilateral frontal activation for the natural gender task, but only right frontal activation (hom*ologue of BrocaŸs area) regarding the grammatical gender task. These results suggest that grammatical gender is processed in different brain areas than natural gender, showing a functional fractionation of the left inferior frontal cortex for level of lexical processing. With regards to the functional distinction of the anterior and posterior parts of BrocaŸs area for conceptual and grammatical processing respectively, these finding are consistent with previous functional brain imaging data (e.g. Friederici, Opitz, and von Cramon 2000). References Friederici, A.D., Opitz, B., & von Cramon, D.Y. (2000). Segregating semantic and syntactic aspects of processing in the human brain: a fMRI investigation of different word types. Cerebral Cortex, 10, 698-705. Heim, St., Opitz, B., & Friederici, A.D. (2003). Distributed cortical networks for syntax processing: BrocaŸs area as the common denominator. Brain and Language, 85, 402-408. Levelt, W.J.M., Roelofs, A., & Meyer, A.S. (1999). A theory of lexical access in speech production. Behavioral and Brain Sciences, 22, 1-75. Acknowledgment: This research project was supported by the "Interdisciplinary Center for Clinical Research on Biomaterials and Tissue-Material-Interaction in Implants (BMBF grant No. 01KS9503/9)".

e215

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 134 FUNCTIONAL MRI FOR ASSESSING CORTICAL LOCALIZATION OF LANGUAGE IN REFRACTORY EPILEPTIC PATIENTS UNDER INVASIVE MONITORING. Cecile, B Grandin , Kenou van Rijckevorsel , Marianne de Tourtchaninoff , Christian Raftopoulos Cliniques Universitaires St Luc, Universite Catholique de Louvain Introduction fMRI has been proposed as a valid and non invasive method for assessing language lateralization in epileptic patients before neurosurgery. It gives more precise anatomical information than the Wada test, and the results can be combined with those of other modalities. We presented herein our experience in fMRI of language combined with invasive EEG in refractory epileptic patients. Methods From our series of 52 epileptic patients who underwent a complete presurgical evaluation (EEG-video monitoring, with subdural grids in 11 of them, structural MRI, interictal FDG-PET, cognitive tests, fMRI of language), we selected 7 patients in whom the results of fMRI were combined with those of invasive EEG monitoring. This group (3M and 4F, 24 to 49 years old) included six patients with a cryptogenic epilepsy, one patient with a dual pathology (right parieto-temporal post-traumatic lesion and right mesiotemporal sclerosis), and one patient with a left occipital non specific cortical scar. All patients with an invasive EEG underwent a 3D GRE T1-weighted MR sequence with the subdural grids in place. The fMRI protocol included a structural T2-weighted sequence and whole brain GRE-EPI sequences for functional tasks (TR/TE 3000/50 ms, voxel size 3.6 mm isometric). The language paradigms were all covert and consisted in a verbal fluency task (word generation to a letter or a category), a reading task (response naming from a short sentence), and an auditory task (response naming from a short sentence). Each paradigm alternated rest and task (12 epochs) for a total duration of 4 min 48. Statistical t-maps of brain activation were generated with the SPM software and superimposed on the T2-weighted images (slices) and on the 3D T1-weighted images with the grids (surface rendering), after coregistration and with the knowledge of the results of the EEG monitoring. Results Two patients were discarded from resection and were considered for vagual nerve stimulation because the left frontal epileptic focus was near and on the Broca area in one patient, and because of an extensive left frontal focus in a patient who had her language exclusively lateralized to the left. One patient had a left frontal focus near the activated left dorso-lateral prefrontal cortex, and the therapeutic decision is still pending. One patient had a bilateral localisation of the Wernicke area but the activated area was more anterior than the right posterior temporal focus. One patient with a right parieto-temporal focus and one patient with a right frontal focus had their language exclusively lateralized to the left. One patient had a left occipital focus located at a distance from the language areas lateralized mostly to the left. Conclusion fMRI was useful to map the cortical language areas in refractory epileptic patients considered for surgery. Images combining fMRI and the exact position of subdural grids were an invaluable tool for therapeutic decision, allowing us to assess the risk and/or the possibility to operate these patients, and to exclude from resection only those candidates who have their epileptic focus on (or too close to) a functionally important area.

e216

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 135 Unaveraged giant MEG activity from language processing areas evoked during speech tasks in patients with lesions adjacent to Brocas and Wernicke area Peter Grummich 1,2 , Christopher Nimsky 2,3 , Rudolf Fahlbusch 1,2,3 , Oliver Ganslandt 1,2,3 Group, University Erlangen, 2 Neurocenter, University Erlangen, 3 Department of Neurosurgery, University Erlangen

1 Biomagnetism

Magnetoencephalography (MEG) can be used for the precise localization of language processing brain areas for presurgical functional imaging, but it has additionally the advantage of a high temporal resolution and is able to depict the temporal interrelationship of language related brain areas during stimulation tasks. In 6 patients with lesions very close to language areas in the dominant side of the brain during language stimulation we found very large MEG activations but with varying intensity in the non-averaged data. The activation amplitudes ranged up to several pico Tesla. The MEG intensity was even larger than the alpha or theta waves. These high intensities are remarkable, because in general language activitiy ranges only up to 200 femto Tesla in intensity. The activations following another stimulus vary in size and latency. Localizations ( by a single dipole algorithm) performed on the data yielded results in the typical language areas with correlations up to 98%. It is remarkable, that the activity is not only found in the area close to the lesion but in other language related areas as well. Furthermore sometimes the activity is even found on the contralateral side of the brain in the areas corresponding to the language areas on the dominant side. The duration of each of the waves evoked from language centers is about 100ms. We found these bursts in most patients more clearly on the dominant hemisphere, where the lesion was located. But in two patients the bursts were found on both hemispheres. During one trial the bursts were often only on one side of the brain, but sometimes they are found in both hemispheres. But in these instances not all waves are simultaneous. There are several possibilities to explain these activities. They might be distortions from excitable tissue or real language processing activity or some kind of after-discharge as the alpha rhythm. The fact that we found these giant waves also contralaterally to the lesion in 2 cases seems to support the hypothesis, that it is real language activity. The possibility to investigate single responses to language stimuli is really interesting. The findings imply a consequence for measuring and interpreting epileptic data: To be careful and try to avoid language processing, to avoid being misled by this giant language waves. These results show also that activity from the nondominant hemisphere may arise from language corresponding areas, which should be taken in consideration when determining the dominant hemisphere by means of MEG or fMRI.

e217

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Activation bursts during picture naming. Upper channels dominant hemisphere. Lower channels nondominant hemisphere. Cavernoma on dominant side. Pitch line intervall 1pT.

Localization sequence in non dominant hemisphere during picture naming. Correlation higher than 0.89.

e218

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 136 Comparison of hemispheric language dominance in epilepsy patients between MEG and Wada procedure: a prospective MEG study Keiko Hara 1 , Deirdre M. Foxe 1 , Susana Camposano 1 , Susanne Knake 1 , Valerie A. Carr 1 , Hideaki Shiraishi 1 , P. Ellen Grant 1 , Elizabeth Donner 2 , Donald L. Schomer 3 , Edward B. Bromfield 4 , Blaise F. Brougeois 5 , A.J. Cole 5 , Eric Halgren 1 , Steve M. Stufflebeam 1 1 1Athinoula A. Martinos Center for Biomedical Imaging, Massachusetts General Hospital/Harvard Medical School,, 2 Beth Israel Deaconess Medical Center, Dept. of Neurology, 3 Childrens Hospital, Division of Epilepsy & Clinical Neurophysiology, 4 Brigham and Women’s Hospital, Dep. Of Neurology, 5 Massachusetts General Hospital, Dep. Of Neurology Rationale: Wada procedure is important in the pre-surgical evaluation of epilepsy patients. It is reliable, but invasive with some serious risks. Magnetoencephalography (MEG) may be used as a non-invasive tool for localization of language cortex (Castill et al, 2003) Previous studies have shown that language dominant hemisphere (HLD) can be determined with fMRI (functional magnetic resonance imaging) or MEG. In this study we prospectively investigated the difference of HLD between MEG and Wada procedure in 20 patients with medically intractable focal epilepsies who were pre-surgical candidates Methods: 18 patients with intractable partial epilepsy aged 11-54 years who underwent MEG language mapping during their pre-surgical evaluation using whole-head 306-channel MEG and simultaneous 64 or 70-channel EEG (Neuromag Vectorviewä Finland). All patients were right-handed. A visually presented verbal memory task presenting 240 stimuli, previously shown to activate language areas, was used. 4 of them were underwent Wada procedure to determine language lateralization. Equivalent Current Dipoles (ECD) based on a spherical head model were fitted using sequential single dipole fitting with a time range of 150ms 600ms and time step of 1ms. Only dipoles with a goodness of fit (GOF) of 70% were considered for analysis. The Laterality Index (LI) was calculated for each pat using the formula: LI = # Dipoles Left Hemisphere - # Dipoles Right Hemisphere A positive LI represented left-sided language dominance and a negative LI corresponds to right-sided language dominance. Results: 16 (89%) out of 18 patients showed the left HLD, 2 patients (11%) showed right HLD. Three patients were underwent Wada tests successfully. One out of three showed the same HLD result. One patient showed bilateral for Wada, and left HLD for MEG, one showed the contralateral HLD result between Wada and MEG. (See TABLE) Abbreviations:Wada: Wada procedure, HLD: hemispheric language dominance, LI: laterality index, F: female, M: male, R: right, L: left, Bil: bilateral Conclusion: MEG shows clear language dominance in partial epilepsy patients. This result doesnt contradict with prior studies. There are a few reports comparing MEG and Wada procedure. (Breier et al, 1999) Many of our patients are planning to have Wada procedure. In order to know MEG is enough to be replacement of Wada procedure, we need more investigation for a comparison of hemispheric language dominance in epilepsy patients between MEG and Wada test with a larger population. Data should be confirmed in a larger population. Funding: The MIND Institute

e219

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Table: Result Pt.#

Age

Onset

Sex

Wada

HLD

LI (GOF:70)

100

planed

100

Bil

100

planed

100

planed

-52

planed

-57

e220

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 137 Converging evidences for basic language processing deficits in Landau-Kleffner Syndrome shown by neuropsychological assessment, source-localization and event-related brain potentials: A case study Ferenc Honbolygó 1 , Valéria Csépe 1 , Rozália Kálmánchey 2 , Gergely Sárközy 2 1 Research Institute for Psychology, HAS, Department of Psychophysiology, Research Group of Developmental Psychophysiology, Budapest, Hungary, 2 Semmelweis Medical University, 2nd Department of Pediatrics, Department of Neurology, Budapest, Hungary In the present study we assessed the linguistic abilities of a 6 years old boy (ZZ) suffering from Landau-Kleffner Syndrome (LKS) using the method of event-related brain potentials (ERPs). LKS is a rare childhood neurological disorder characterized by acquired childhood aphasia accompanied by an abnormal EEG. The peaking onset of LKS is in the ages of 3 and 7 and typically affects the left centro-temporal part of the brain. The medical history of ZZ commenced at the age of 4.4, when he had a 1-2 minutes long grand mal seizure after awaking. Since then he has been hospitalized several times and obtained first a carbamazepine therapy, then after the diagnosis of LKS he received ACTH. His medical treatment was changed repeatedly according to his actual conditions. ZZs neuropsychological assessments showed an apparent dissociation between verbal and nonverbal (visuospatial) functions, inasmuch as his performance in the visuospatial tasks was in the normal age range, while his performance shown in the verbal tasks was well bellow the normal range. To infer the possible generator of epileptiform activity, we used the BESA 2000 source-localization program. We measured awake EEG, and averaged it time-locked to the epileptic spikes. The analysis showed two possible generators in the left centro-temporal lobe, but the exact localization was not feasible with EEG because of their subcortical origin. An MR diffusion PET measurement accomplished after our study supported this finding, as the PET measurement showed an elevated metabolism in the left temporal lobe, in the superior temporal gyrus. In order to follow processing changes in the auditory domain, a passive oddball paradigm was introduced, in which the automatic discrimination of initial phonemes of words, as well as that of altered stress patterns were measured. As we showed in our previous study (Honbolygó, Csépe, Ragó, submitted) in both conditions a Mismatch Negativity (MMN) component is evoked by the deviants in normal subjects. ZZ showed an MMN to the phoneme contrasts only on the right side, and no MMN at all to the stress deviants was observed. Our results confirm that ZZ’s epileptic activity affecting seriously his linguistic abilities might cause basic processing deficits as shown by the abnormal MMNs.

e221

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 138 Characteristics of Brain Activation During Semantic Processing in Schizophrenia Jae-Jin Kim 1,2 , Jeong Ho Seok 1 , Jun Soo Kwon 3 , Dong Soo Lee 4 , Myung Chul Lee 4 1 Department of Psychiatry, Yonsei University College of Medicine, Seoul, Korea, 2 Institute of Behavioral Science in Medicine, Seoul, Korea, 3 Department of Psychiatry, Seoul National University College of Medicine, Seoul, Korea, 4 Department of Nuclear Medicine, Seoul National University College of Medicine, Seoul, Korea Objectives : Many investigators have reported that patients with schizophrenia have deficits in semantic processing. However, it is unclear which brain area is associated with semantic processing dysfunction in schizophrenia. This study was designed to explore the activated brain areas associated with semantic processing in schizophrenic patients when compared with controls. Methods : Twelve patients with schizophrenia and twelve healthy controls were studied under two different visual task conditions. Subjects were required to respond to a specific semantic category in a specific figure among word-figure stimuli during the first task, and respond to a specific figure among figure-only stimuli during the second task. Brain activation during each task was measured using [15O]H2O PET. Activated brain areas were analyzed by subtraction methods using SPM99 in each group. Results : In healthy control group, the left superior temporal gyrus, left premotor area and left cerebellum were activated during semantic processing along with activation of the left inferior temporal gyrus which is a main semantic processing area. But activation of the main semantic processing area in patient group was observed more posteriorly than controls. In contrast with control group, lateralized activation pattern to the left and cerebellar activation were not observed in patient group. Conclusion : Our results suggest that patient’s deficit in elaboration due to early semantic processing, decreased efficacy due to loss of lateralization and decreased modulatory ability due to loss of cerebellar activation may be intervened in the characteristics of brain activation patterns in schizophrenia. This distorted semantic processing in schizophrenia may play a role as one of the basic determinants in thought disorder.

e222

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 139 Brain activity during sentence reading in dyslexic readers: a fMRI study Martin Kronbichler 1,2 , Heinz Wimmer 1 , Florian Hutzler 3 , Alois Mair 2 , Wolfgang Staffen 2 , Gunther Ladurner 1 1 Department of Psychology, University of Salzburg, 2 Christian Doppler Clinic, PMU, Salzburg, 3 Bereich Allgemeine Psychologie, Freie Universität Berlin

Introduction The present fMRI study explored abnormalities in brain activity of German-speaking dyslexic readers during a natural sentence reading task requiring semantic judgments. To our knowledge, few functional neuroimaging studies have used reading of sentences to explore brain activation in dyslexic readers. Due to the regularity of German, our dyslexic readers suffer mainly from slow effortful reading and less from reading errors. Methods Thirteen dyslexic and 15 normally reading boys (all right-handed), in the age range from 14 to 16 years, matched for age and nonverbal intelligence, participated. During functional imaging short sentences had to be read and judged as correct or incorrect. In the visual control task, the words of the sentences were replaced by strings of pseudoletters which had to be judged according to whether or not just one type of pseudoletter was present. During 7 epochs of reading and 7 epochs of visual judgment 178 functional images were acquired. Image processing including realignment, unwarping, normalization and smoothing (9 mm FWHM) and data analysis was performed using SPM 2 (http://www.fil.ion.ucl.ac.uk). Separately for each group, enhanced activity of reading compared to visual processing was assessed by a RFX analysis (FDR threshold: p >.05). Evaluation of group differences (FDR threshold: p > .05) was restricted to regions showing higher activity for reading vs. visual processing in at least one group. Results Figure 1 shows that normal readers (first panel) relative to visual processing exhibited more extended activation during reading in the left temporal cortex, whereas dyslexic readers (second panel) showed more extended activation in the left frontal cortex. These group differences (third panel) were reliable for small regions of the left posterior middle temporal gyrus and the left supramarginal gyrus, where normal readers showed higher activity. In regions of the left frontal cortex (inferior frontal, precentral and postcentral) dyslexic readers exhibited higher activity. In addition, dyslexic readers exhibited higher activity in medial temporal and in several subcortical areas, including the thalamus. Discussion In summary, the present fMRI findings showed that dyslexic readers exhibited lower activation in regions of the left posterior middle temporal and the left supramarginal gyrus, comparable to previous reports (e.g. 1,2), which may reflect a dysfunction of posterior visual word processing networks (3) leading to slow effortful reading performance. The need to compensate for the dysfunction of the left posterior visual word processing network may lead to higher activation of left frontal regions, reflecting reliance on phonological articulatory word processing strategies. The high effort required by this compensation may have led to enhanced activation in subcortical regions. 1.Paulesu et al., Science, 2001, 291, 2165-7. 2.Rumsey et al., Arch Neuro, 1997, 54, 562-73. 3. Shaywitz et al., Biol Psychiatry, 2002, 101-10. 4.Pugh et al., Ment Retard Dev Disabil Res Rev, 2002, 6, 207-13.

e223

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

e224

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 140 Impaired development of fast, specialized print-processing in dyslexic children during learning to read Urs Maurer 1,2 , Silvia Brem 1 , Felicitas Kranz 1 , Kerstin Bucher 1,3 , Rosmarie Benz 1 , Pascal Halder 1 , Hans-Christoph Steinhausen 1 , Daniel Brandeis 1 1 Department of Child and Adolescent Psychiatry, University of Zurich, 2 Sackler Institute for Developmental Psychobiology, Weill Medical College of Cornell University, New York, 3 Neuroradiology & Magnetic Resonance, Department of Diagnostic Imaging, University Children’s Hospital, Zurich Developmental dyslexia is occurring in 5-10% of all school children, is running in families (30-50%), and is defined as a disorder in learning to read. This definition focuses on the process of reading acquisition, which is best studied in longitudinal designs. An important achievement during learning to read is the development of fast, specialized print processing (N200 word-symbol effect, see poster of Brandeis et al.), which may be impaired in children with dyslexia. In Switzerland children do not get reading training in kindergarten, but only later in school. We thus investigated kindergartners (24 with and 24 without familial dyslexia-risk) who could not yet read words, and followed them up 1.75 years later in 2nd grade after they had learned to read. We recorded 43-channel event-related potential (ERP) maps, while the children detected immediate repetitions of words, pseudowords, symbol-strings, and pictures. ERP segments (P100, N200, P350, P300-like components) were analyzed regarding map strength (Global Field Power, GFP) focusing on word and symbol-string conditions in overall and segmentwise analyses. According to an early reading test in grade 2 the subjects were divided into children with dyslexia (n=12), control children (n=19, without risk), and not affected children at risk (n=11). Dyslexic children did not differ behaviorally from controls in kindergarten, but they were less accurate in 2nd grade in detecting repeated targets, especially for word and pseudoword targets. Over the whole ERP range, words elicited more GFP than symbols in the N200 segment and less GFP in the last two segments. This was more pronounced in 2nd grade after learning to read than before in kindergarten, but such developmental effects were observed to a lesser degree in children with dyslexia (dyslexia x age x wordlike x segment, p<0.01). In the N200 segment, control childrens word-symbol difference increased from near zero in kindergarten to a large posterior negative difference in 2nd grade, whereas the increase was only moderate in children with dyslexia (dyslexia x age x wordlike, p<0.01). This was partly due to a slightly reduced word GFP and an increased symbol GFP compared to controls in 2nd grade, but also due to an already more pronounced word-symbol difference at the kindergarten level in dyslexic children. Children with larger N200 word-symbol difference in 2nd grade were also reading faster (r=-0.51). The impaired development of the N200 word-symbol effect in children with dyslexia suggests that fast print processing is becoming less specialized in these children, although the diagnoses of dyslexia remain to be confirmed. The group differences after learning to read were more prominent for the N200 word-symbol effect, rather than for the word N200 alone, which suggests impaired discriminative processing. This reduced specialization was paralleled by behavioral deficits in the same task and was strongly correlated with results in a reading test, which suggests that fast, specialized print processing plays a key role for successful reading acquisition as well as for its failure in dyslexia. Supported by grants SNSF 32-59276, Stiftung wissenschaftliche Forschung, Universität Zürich, and NCCR Neural Plasticity and Repair

e225

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 141 Secondary, fully compensatory bilateral representation of speech in a right-handed female with a left fronto-temporal tumor Janpeter Nickel 1,2 , Raimund Kleiser 1 , Silke Joergens 1 , Heiko Neeb 2 , Tony Stoecker 2 , Nadim J Shah 2 , Peter Indefrey 3 , Ruediger J Seitz 1 1 Department of Neurology, University Hospital Duesseldorf, Germany, 2 Institute for Medicine, Research-Center Juelich, Germany, 3 Max Planck Institute for Psycholinguistics, Nijmegen, Netherlands INTRODUCTION: We describe the case of a right-handed 33 year old female patient, who presented with a speech-arrest followed by a generalized tonic-clonic seizure as her first ever symptoms of an anaplastic astrocytoma in the left fronto-temporal region. Before invasive diagnostics and surgical treatment, she underwent functional MRI-investigation for the localization of speech areas. METHODS: Several neuropsychological tests on verbal skills were performed and a block-designed fMRI-speech paradigm was carried out, in which the subjects were requested to insert visually presented verbs into visually presented sentences (see Sach et al.). The control condition consists of a resting state with a fixation cross being presented. Scanning was realized on a 1.5 T MRI Scanner with an EPI sequence (TR 4000ms, 30 slices with a 64x64 matrix, whole brain coverage). The functional data was processed using "BrainVoyager 2000" (Version 4.9.6.0, © 1996-2001 by Rainer Goebel and the Max-Planck-Society). The calculated volume-time-code was overlaid on a 3D-MRI of the patient (160 slices with a 256x256 matrix) and the speech activation (insertion+production vs. rest) was analyzed using a general linear model approach. RESULTS: The patient was 100% right-handed according to the Edinburgh Handedness Scale (Oldfield, 1971) and had normal percent-ranks (16-84%) in the neuropsychological tests: Rivermead Behavioural Memory Test (RBMT immediate recall 48,5%, delayed recall 63,6%), Wechsler Digit-Span (forward 95%, backward 26%), Wechsler Block-Span (forward 43%, backward 19%), Verbal fluency (category forenames 85%). The Token Test from the Aachener Aphasie Test was performed without any error. Structural MRI showed a large tumor in the left frontotemporal region with surrounding edema and an older central hemorrhage. Overlay with the functional data revealed an inferior frontal and temporoparietal activations in the left hemisphere, which were displaced from the natural position of classical speech areas as found in 12 healthy righthanded volunteers (see Joergens et al.). In addition, the patient deviated from the strictly unilateral left-hemispheric localization in the healthy controls by an activation of hom*ologous regions in the right hemisphere (Figure 1 and 2). CONCLUSIONS: The bilateral activation and the displacement of the speech areas seen in this strictly right-handed patient during a language task is an example for plasticity in the adult brain facing the progressive and invasive growth of a tumor at a moderate speed. As seen in this patient, it is possible to sustain normal function in a certain modality, such as language, by relocation of eloquent cortical regions within and across the hemispheres similar to post-stroke recovery. Thus, a preinvasive functional mapping of these functions is not only helpful but essential to help avoiding a disturbance or loss of function by tumor surgery, since normal speech abilities do not exclude residual function close-by the tumor. References: - Sach M. et al., Neuroreport, 2004 (in press) - Joergens S. et al., see poster contribution to this meeting (HBM 2004)

e226

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 1: Activation pattern of the Speech-fMRI as shown in the functional dataset (the green arrows mark the activation of hom*ologous right-hemispheric regions to the main left-hemispheric activity, statistical threshold at p<0.04)

Figure 2: Activation pattern of the Speech-fMRI as shown in the overlay of function and high-resolution 3D-anatomy (the green arrows mark the activation of hom*ologous right-hemispheric regions to the main left-hemispheric activity)

e227

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 142 Behavioral neuropsychological assessment and brain activation in an acquired dyslexia : an fMRI study SungBum Pyun 1 , JaeBum Jung 2 , HanYoung Jung 3 , SeongDon Yoo 4 , HyoJoung Son 2 , KiChun Nam 2 , HoKyu Lee 5 1 Department of rehabilitation medicine, Asan medical center, University of Ulsan, College of medicine, South Korea, 2 Department of psychology, Korea university, South Korea, 3 Department of rehabilitation medicine, Inha University College of Medicine, 4 Department of rehabilitation medicine, Myoungji hospital, Kwandong university, College of medicine, 5 Department of radiology, Asan medical center, University of Ulsan, College of medicine Background Dyslexia has long been interested in many researchers because it has unique characteristics in the processing of written words. There are dual routes in the written word processing, the first one is via semantic system and the other route is through grapheme phoneme conversion (GPC) without pass by semantic system. The acquired dyslexia was classified as several types according to the coexistence of agraphia or aphasia, by characterstics of neuropsychological language procesing, such as deep dyslexia, surface dyslexia, or phonologic dyslexia. Recent days the functional mapping study especially fMRI study has provided much insight to the cortical areas related to the specific language processing and more elaborate task may differentiate the unique site in each stage of the processing. In this study we investigate the characteristics of an acquired dyslexic patient using hierarchical neuropsychological assessment model and fMRI study with four tasks according to language processing. Method A sixty two-year old right handed dyslexic patient after left posterior cerebral artery infarct participate in this study. The test was designed to investigate the neuropsychological characteristics and brain activation pattern using a fMRI. Neuropsychological study was performed according to 4 stages (general cognitive, visual perceptual, pre-lexical, and lexical). The general cognitive function was assessed by MMSE, NCSE, Computerized neuropsychological Test battery including attention, memory, visuomotor and higher cognitive executive function and language was evaluated by K-WAB and K-BNT. Perceptual stage was evaluated using VEPs, MVPT, visual neglect test, visual searching tasks and pre-lexical assessment using picture-word matching, semantic categorization tasks. Lexical stage was assessed using lexical decision task, picture naming and reading of word and non-word. Funtional image was performed using 4 tasks (picture naming, picture-word matching, semantic categorization, and reading) of block design (12 sec dummy, 4 activation blocks, 5 control blocks per session). The data acquisition was performed using an ISOL 3.0 Tesla forte MR scanner with EPI sequence (TR/TE = 300/35ms, 4 mm no gap, 20 slices, 64x64 matrix, FOV=220mm, Flip angle 80°). All data processing and statistical analysis were performed using SPM 99 and statical significance level was p<0.01. Result The K-WAB results were dyslexia without agraphia (AQ 71, reading 28%, writing 75%) with mild anomia. In cognitive function test, MMSE and NCSE was of borderline value. Attention, visual learning, visuomotor and executive funtion was appropriate, however auditory learning test was of 5 percentile level. In perception stage, VEPs, MVPT, visual neglect test were normal and visual searching of same letter test was 90% correct response rate, and the next prelexical stage evaluation represents 80.6% (picture word matching) and 78.6% (semantic categorization) correct response rate. In the lexical stage, correct responses were picture naming (40%), lexical decision task (high frequency 80%, low frequency 60%, concrete word 93%, abstract word 80%), reading (word 88.8%, non-word 38.8%). In fMRI study, right broca area activation without left brain activation was typical in picture naming task, and basal temporal lobe, both prefrontal lobe, right temporoparietal area (picture-word matching), both inferior parietal, prefrontal area (semantic categrization), both anterior inferior frontal, prefrontal (reading) was represented. e228

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Conclustion This patient had a characteristics of deep dyslexia according to neuropsychological assessment and the brain activation pattern was different from that of normal person. There are evidences of expressive language area reorganization into right Broca’s hom*ologue area. During the reading tasks the processing was not appropriate and the activation was noteworthy in bilateral anterior inferior frontal area and prefrontal cortices.

e229

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 143 Lexical speech therapy in aphasic patients : a PET study G. Raboyeau , X. De Boissezon , C. Bezy , G Viallard , J-F. Démonet , D. Cardebat INSERM UNITE 455, Toulouse, France Introduction Neural correlates of recovery from aphasia after speech therapy are still largely unknown. We carried out a PET investigation in anomic patients before and after a one-month word therapy in order to assess the respective compensatory role of the left and right hemispheres in word re-learning. Methods Six right-handed aphasic patients (mean age 53 +/- 15) examined at a chronic stage (> one year after stroke) were included. All lesions were peri-sylvian, sparing the left superior temporal gyrus. The protocol involved 2 PET sessions, one before (T1) and one after (T2) a computer-assisted lexical training performed by the patients 5 days a week (about 15 min per day) during 4 weeks. Each PET session included 2 "Rest" and 4 activation scans including overt confrontation naming task (100 items). Scanning was performed using the O15-water method and a HR+ Siemens PET camera. Neuroimaging data were analysed using SPM 2. Threshold was set at P < .05 (cluster extent of 40 voxels) for subtraction analysis and at P<0.01 for correlation analysis. Results Behavioral results: mean accuracy scores were 46.5/100 (range from 2 to 78) at T1, 63.2/100 (range from 6 to 95) at T2. We observed a significant improvement after the training period in all subjects (t= -5.97, p=0.001). PET results: The neural pattern associated with naming improvement (Naming T2-Naming T1; fig1) implicated essentially the right hemisphere including the cerebellum, the superior occipital cortex (BA 18, 19), the inferior temporal cortex (BA 20), the insular cortex and the hippocampal formation. Left hemispheric activations were only found in the associative visual areas (BA 18, 19). We tested the correlations between neural changes (PET2-PET1) and performance improvement (perfT2-perfT1). The correlation (high performance/high activation) showed only right hemisphere clusters, including the cerebellum, the inferior occipital cortex (BA 18), the posterior and anterior cingulate areas (BA 31, 24), the lateral inferior and superior prefrontal cortex (BA 8, 10), and the insula. Discussion Whatever the initial performance, all patients showed a significant lexical improvement. PET T2-T1 analysis showed the crucial implication of the right hemisphere with a network including medial and inferior temporal regions and the insula. This pattern, although right-sided located, may be discussed in terms of lexical semantic memory [1] and articulatory processes [2]. The implication of the right insula was confirmed by the correlation analysis. Our results suggest that in anomic patients, compensatory processes assumed by the right hemiphere seem to be efficient at the word-level, even at a chronic stage. [1]Price, C., The anatomy of language : contributions from functionnal neuroimaging. J. Anat, 2000. 197: p. 335-359. [2]Wise, R.J., et al., Brain regions involved in articulation. Lancet, 1999. 353(9158): p. 1057-61.

e230

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Naming T2-NamingT1 p=0.05 k>40

e231

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 144 Abstract Grammatical Processing in Brocas Area: Evidence from fMRI and Intracranial Electrophysiology Ned T. Sahin 1,2 , Eric Halgren 2 , Istvan Ulbert 2 , Anders Dale 2 , Don Schomer 3 , Julian Wu 3 , Steven Pinker 1 Psychology Department, Cambridge, MA, 2 MGH Martinos Neuroimaging Center, Charlestown, MA, 3 Neurology, Beth Israel Hospital, Boston, MA

1 Harvard

We show that covert grammatical inflection of English nouns & verbs is a sufficient & consistent activator of Brocas area, via convergent evidence from fMRI & Depth Electrophysiology. Introduction Broca’s Area, identified almost 150 years ago, 1 is the original and most widely known example of brain-behavior correspondence, yet its specialized role in language remains heavily debated. 2 Broca’s role has shifted from expressive (vs. receptive) language, 3,4 to general grammatical processing 5 - the combinatorial process that unlocks Language’s infinite expressive power. More recently, various accounts have imputed Broca’s functional activations to non-language-specific processes such as working memory, selection demands, storing moved sentential elements, and integrating those elements. These processes necessarily come with manipulations of syntax, the construction of sentences from words, so isolating the language-specific role of Broca’s area may require a system with the combinatorics of language grammar but without the complexity of syntax. Such a model system is inflectional morphology, the construction of complex words (e.g. "walked") from meaningful constituents or morphemes (e.g. "walk" + "-ed"). There are few neuroimaging studies of isolated morphological production, and no intracranial electrophysiology. Methods Subjects viewed simple sentence-like frames cueing them to produce silently the ensuing target word in Overt-Inflect(plural/past), Zero-Inflect(singular/present) or simply Repeated form (Task Figure). 18 healthy, right-hand, English-speaking, undergraduates were scanned in 3T event-related fMRI. Two epilepsy patients clinically implanted with depth electrodes near language areas consented to research recording of evoked potentials during the same task. Results Whole-brain random-effects fMRI showed p<0.01 activations only in BA44/45, 47, SMA, insula: for Overt-Inflected > Repeated words(Figure 1b). Zero-Inflected > Repeat (identical phonology but manipulation of abstract grammatical features) selected Broca’s (Figure 2b); while phonological manipulation in Overt-Inflect > Zero-Inflect added anterior insula(Figure 2a) and SMA. This pattern converges with a demonstrated 100% correlation between insula lesions and motor speech deficits (apraxia). 6 Noun and verb-only contrasts yielded highly similar patterns (Figure 1c) suggesting inflectional morphology consistent across classes. Intra-cranial Electrophysiology: One of the patients completed the idential fMRI experiment, pre-operatively, and showed BA44 activation similar to group results from the healthy volunteers. This activation cluster lay along the trajectory of one of the depth electrodes eventually implanted. Recordings from this electrode confirmed task-selective activity for the Overt-Inflect task, and showed an inversion of signal across the three contacts closest to the IFG region implicated. This indicates unambiguous localization of the electrical signal source. See Figure 3. Conclusions 1.) Broca’s area serves central role in grammatical processing, independent of working memory. 2.) Broca’s and BA47 sensitive to abstract grammatical features; SMA and anterior insula sensitive to phonological production (even covert). 3.) Neural pattern is stable across noun pluralization and verb conjugation, suggesting common circuitry for inflectional morphology.

e232

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

3.) Morphosyntax validated as model system for studying grammar. 4.) Depth electrophysiology confirms within-subject fMRI activations, and across-subject fMRI-based cognitive mapping. References 1. Broca, P., Bull. Société Anatomique 6:330-357 (1861) 2. Kaan, E., & Swaab, T.Y., TICS 6:350-356 (2002) 3. Wernicke, C., Der Aphasische Symptomenkompleks. (1874) 4. Geschwind, N., Science 170:940-944 (1970) 5. Caramazza, A. & Zurif, E.B., BrainLang 3:572-582 (1976) 6. Dronkers, N., Nature 384:6605 159-161 (1996)

Support: NIH HD 18381 & NS18741

Task: Frames, target examples, and timings for the three inflectional conditions and two grammatical categories.

Figure 1b shows that our task did activate Broca’s area, and also insula, BA47 (SMA not shown). Figure 1c shows that this pattern is stable across nouns and verbs.

e233

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 2: Sensitivity of Broca’s Area to abstract grammatical features, and of the Anterior Insula to phonological material related to overt inflectional morphemes.

Figure 3: Intracranial electrophysiological data. Shows correspondence between group normal fMRI, individual patient fMRI, and electrophysiological components for this task condition.

e234

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 145 Early activation and time course of recovery in aphasia: a longitudinal f-MRI study Dorothee Saur , Rüdiger Lange , Annette Baumgaertner , Christina Musso , Michel Rijntjes , Cornelius Weiller Department of Neurology, University Hospital Hamburg-Eppendorf Introduction Recovery from aphasia in the early stage is determined by a range of different factors including the pathophysiology of infarction and lateralisation of the pre-existing language network. This entails a highly dynamic and variable recovery process whose mechanisms are as yet largely unknown. In order to track the dynamics of this process, we repeatedly examined patients beginning at the very early stage after symptom onset. Methods Five patients with aphasia were examined at 1-3, 12-15 and again at 6 months after symptom onset. In an auditory task, patients listened to correct sentences and sentences with a semantic violation e.g. "The pilot flies the plane" (correct) and "The pilot eats the plane" (semantic violation). The same sentences in reverse served as a control condition. A total of 184 sentences were presented in an event-related design. Patients were asked to press a button whenever they detected a mistake. Results During this time period, all 5 patients recovered clinically as demonstrated by their performance on a set of aphasia tests (Aachen Aphasia Bedside Test, Aachen Aphasia Test including the Token Test, functional outcome score) and by task performance in the scanner. Case 1 with an infarction in the frontal middle cerebral artery (MCA) territory including the posterior part of Broca’s area revealed early left perilesional inferior frontal activation (day2). Two weeks later, there was activation of the right Broca hom*ologue with concurrent clinical improvement. This pattern was unchanged in the follow up examination after 6 months, while there had been almost complete recovery. Case 2 with an infarction in the frontal and middle MCA-territory including Broca’s area showed no language specific early activation (day2). After 12 days, there was activation of posterior parts of the left middle temporal gyrus without substantial concurrent clinical improvement. After 6 months, a remarkable clinical improvement was accompanied by a recruitment of the right Broca-hom*ologue. Case 3 suffered an infarction of the posterior MCA-territory sparing Wernicke’s area but destroying the arcuate fascicle leading to a disconnection of the language areas. Three days post stroke the patient showed activation of Broca’s area, which significantly increased after another 10 days. During this time, scanner task performance had significantly improved, possibly by an increasing reliance on the ventral pathway via the uncinate fasciculus. Case 4 and 5 both showed a striatocapsular infarction due to temporary MCA occlusion. Both patients showed no language specific early activation (day1), but had reactivated Broca’s and Wernickes area 7 and 14 days later, paralleling a good clinical recovery. Conclusions Knowing early activation patterns helps to interpret subsequent changes in activation and sheds more light on the highly dynamic and variable processes of language recovery. These five cases suggest potential recovery mechanisms, including perilesional recruitment (case 1), substitution of a destroyed Broca’s area by its hom*ologue (case 2), increasing reliance on the ventral pathway via the uncinate fasciculus (case 3), and restoration of a preexisting left lateralized language system (cases 4 and 5).

e235

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 146 Agraphia following awake surgery for low-grade gliomas: The anatomo-functional network of writing revisited Pietro SCARONE 1 , Peggy GATIGNOL 2 , Laurent CAPELLE 3 , Hugues DUFFAU 3 of Neurosurgery, Ospedale San Raffaele, Università Vita-Salute, C Professor Massimo Giovanelli Barilari, Via Olgettina 60, 20132 Milan, Italy, 2 Department of Neurology, Hôpital Salpêtrière, 47-83 Boulevard de lHôpital, 75634 Paris Cedex 13, France, 3 Department of Neurosurgery and Inserm U494, Hôpital Salpêtrière, 47-83 Boulevard de lHôpital, 75634 Paris Cedex 13, France. Email: [emailprotected] 1 Department

Objective Agraphia is commonly observed in clinical practice (1), usually after vascular events or in the context of cerebral neoplasms, and has been included in various neurological syndromes, such as Gerstmanns. In spite of a great number of neuropsychological (2) and recent neurofunctional studies (3) on writing, many controversies currently exist about the organization of the cerebral network implied in writing and the exact localization of its areas, the contributions of each of the latter to the writing process, and their relations with the areas implied in oral language. Patients and method In the present study, we have analyzed the postoperative agraphia profiles of 15 patients (6 males, 9 females, mean age: 37 years), with preoperative normal writing, who underwent surgery for cerebral low-grade gliomas in functional language areas using electrical mapping under local anesthesia (biphasic current, 1 msec/phase, 60 Hz, 2 to 8 mA, OCS1 Radionics*) (4,5) then we tried to correlate these profiles to the sites of the lesions. Results Our findings showed (i) that language and writing functions could be dissociated, (ii) that writing relies, at least partially, on a network of five areas which were located in the dominant hemisphere for language: the superior parietal region, the supramarginalis gyrus, the second and third frontal convolutions, the supplementary motor area, and the insula, (iii) that each zone participated in different linguistic and/or motor aspects of writing. Among them, however, only patients with lesions of the supplementary motor area did not always recover from agraphia in the postoperative period (in 50% of cases). Conclusions On the basis of these results, and in the light of recent neuropsychological and neuroimaging data, we propose a preliminary cortical network model underlying the elaboration and production of written language, constituted by the five areas described above. The posterior regions involved in multimodal integration, i.e. the superior parietal lobule (spatial processing) and the inferior parietal lobule (access), might interact in a dynamic fashion with the anterior regions, i.e. the Exner / Brocas areas (rules processing) and SMA (initiation), by the mediation of the phonological loop (allowing transient storage and conversion) and the insula (planification). A better knowledge of such an anatomo-functional writing circuit might have important implications in the peri-operative evaluation of cerebral neoplasms. References 1. Roeltgen DP, et al. Neurology 1983; 33:755-765 2. Shallice T. Brain 1981;104:413-429 3. Longcamp M, et al. Neuroimage 2003;19:1492-1500 4. Duffau H, et al. Brain 2002;125:199-214 5. Duffau H, et al. J Neurol Neurosurg Psychiatry 2003;74:901-907

e236

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 147 fMRI of Language in Neurosurgical Planning for Temporal Lobe Epilepsy: Dependence on Input Modality and Language Variable Robin J. Schafer , Brian Roberts , R. Todd Constable Yale University An important clinical application of fMRI is that of language mapping to identify regions that must be spared in neurosurgery. While the use of fMRI in this setting is growing, questions remain as to the best means to maximize accuracy and reliability of the method. This work addresses how task parameters affect the accuracy of localization. In an earlier block design fMRI studies by our group, we presented healthy normal subjects and surgery candidates with a sentence comprehension task (Carpentier et al. 2001, Constable 2004). Parallel materials were presented through either the visual or auditory pathways, and varied in structural complexity. Blocks varied according to whether sentences contained subject relative clauses or object relative clauses. The current work involves a similar design in which sentence materials are presented in two modalities, but the language variable is the degree of anomaly in sentence blocks. Recent work on anomaly and complexity suggests that complexity predominantly involves activation in the inferior frontal gyrus pars opercularis, whereas anomaly predominantly involves activation in the superior and medial temporal gyrus. This suggests the possibility that language tasks based on anomalies will result in better localization of activation in the temporal lobes. Data analysis compares images from 10 control and 10 clinical subjects. Imaging was performed using a 1.5T Siemens Sonata with a standard quadrature head coil: 16 axialoblique slices (6 mm thick, 0 gap) with gradient echo planar imaging (64x64 matrix, TE=45ms, a= 80, TR=1800ms, FOV=24cm). This work will also further explore the language lateralization patterns within clinical subjects with temporal lobe epilepsy. In the complexity task, normal controls show left lateralized differential activations in the visual modality and bilaterality in the auditory modality. Surgery candidates show recruitment of contralateral hom*ologous language areas significantly above that in the control group in both modalities. In the anomaly task, normal controls show bilaterality only in the visual modality. The question we explore is whether the lateralization patterns of surgery candidates vary as well.

e237

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 148 Cortical Distinctions in Visual and Auditory Naming Sandra Serafini , Michael Haglund Department of Neurosurgery, Duke University Introduction Visual confrontation naming tasks are used extensively in intraoperative cortical stimulation mapping (CSM) to determine language-sparing resection margins. Many patients, however, experience word finding difficulties severe enough to interfere with daily speech but show little if any deficits on visual confrontation naming tasks. It has been further reported that these patients show significant post-operative deficits in auditory relative to visual naming (Hamberger & Tamny, 1999). Using an auditory naming paradigm we developed, we demonstrate that there are cortical distinctions in visual and auditory naming in LTE patients during CSM. Methods Patients (N=7, 1 female, ages 17-56) who were scheduled to undergo an awake left temporal lobe resection with language mapping for the treatment of epilepsy were recruited for the study. Black-and-white line drawings were shown to the patient every 4 seconds for the visual naming task. Auditory naming consisted of the patient listening to a short (5-7 words) spoken definition describing common natural and man-made objects. Patients responded in each paradigm with a carrier phrase and the name of the object. Using the Fishers Exact test (p < 0.05), we compared the error rate for each task at a specific site with the error rate of all non-stimulated trials. Results Seven of seven patients showed cortical distinctions between visual and auditory naming sites. In many cases, overlapping sites significant for both visual and auditory naming were found. Significant sites are indicated below (Figure) in yellow for auditory naming, blue for visual, and green for overlapping sites, i.e. where both visual and auditory errors were made. The pattern most commonly found shows visual naming sites appearing in relatively superior/posterior areas such as Brocas area or the supramarginal gyrus. Auditory naming sites, however, appear in relatively inferior/anterior areas such as the medial superior and middle temporal gyri. Overlapping sites tend to be located in more posterior temporal areas. Conclusions There are distinct cortical areas found in patients during CSM between auditory and visual naming tasks. Including an auditory task in the language mapping procedure provides a more comprehensive language map and sparing these sites may prove significant in reducing post-operative language deficits in patients.

e238

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Visual and Auditory naming sites.

e239

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 149 Partial left prefrontal lobectomy sparing working memory center manifested language deficit: A case report Isamu Shibamoto 1, 3 , Tokutaro Tanaka 2 , Naoto Sakai 2 , Atsushi Mizutani 2 , Yasue Kanemoto 1 , Norimasa Katagiri 1 1 Dept. of Rehabilitation Medicine, Seirei Hamamatsu General Hospital, 2 Dept. of Neurosurgery, Seirei Hamamatsu General Hospital, 3 Tokyo Medical and Dental University Background: Working memory is always used for language comprehension and language expression. We had a patient who suffered from language deficit although working memory center was not damaged during partial left prefrontal lobectomy. Case: 49 year-old right-handed female was admitted to our hospital after having epileptic seizure and losing consciousness. Left frontal intra-axial brain tumor was revealed by neuroimaging. After discussing the relevant case information and potential treatments we opted to perform a partial left prefrontal lobectomy. The patient had language and cognitive examinations pre-operatively and there were no abnormal findings. We examined and identified her working memory center using functional magnetic resonance imaging (fMRI). We used a 1.5 Tesla Magnetic Resonance Imager (Signa 1.5) under one shot echoplanar imaging (EPI) sequence with TR 3000 msec, TE 60 msec, FOV 24 cm, 8 mm thickness, and 64x64 matrices. We obtained 12 axial images with 2-mm slice gaps every 3 seconds. This sequence was repeated for total of 984 images. While undergoing fMRI, the patient performed a word play exercise called Shiritori. In the experiment, 30 seconds of rest was followed by 30 seconds of word play task. This sequence was repeated four times. The images were analyzed using SPM99. Using parameters based on our language experiment, SPM99 software highlighted the areas where signal intensity changed during the alternating periods of rest and word play. We used intra-operative neuronavigator to avoid damaging the working memory center. For post-operative assessment, we used the same language and cognitive examinations as pre-operation. After the operation, the patientÅfs auditory comprehension, spontaneous speech, reading comprehension, writing, and dictation were impaired. Pathological examination suggested left frontal lobe anaplastic oligodendroglioma. These findings suggest that working memory has a deep relationship with the prefrontal lobe area, located more anterior to working memory center, and that both influence linguistic ability. Conclusion: Although working memory center is spared, linguistic ability can be disturbed by partial left prefrontal area lobectomy.

e240

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 150 Functional disconnectivity in subjects at high genetic risk of schizophrenia. Enrico Simonotto 1,2 , Heather C Whalley 1 , Martin Meyer 3 , Marie-Claire Whyte 1 , Ian Marshall 4 , Klaus P Ebmeier 1 , David GC Owens 1 , Nigel H Goddard 5 , Eve C Johnstone 1 , Stephen M Lawrie 1 1 Division of Psychiatry, University of Edinburgh, UK, 2 MRI Devices, Waukesha, WI, USA, 3 Department of Neuropsychology, University of Zurich, Switzerland, 4 Division of Medical Physics, University of Edinburgh, UK, 5 School of Informatics, University of Edinburgh, UK Introduction: Schizophrenia is a highly heritable psychotic disorder. It has been suggested that the associated deficits could arise from abnormal interactions between brain regions, particularly involving connections with the prefrontal cortex. Studies in established schizophrenia have reported altered prefrontal connectivity, with temporal 1,2 , cerebellar 3 , and parietal regions 4 , and between lateral and medial prefrontal regions 5 . We sought to examine the hypothesis that if deficits were genetically mediated, then altered patterns of prefrontal connectivity seen in the established state would be observed in subjects at high genetic risk of the disorder, and may be related to the degree of inherited risk. Conversely, if these abnormalities were associated with the manifestation of symptoms, then these may be evident in those displaying some of the characteristic features of the disorder. Methods: Functional connectivity analysis was performed on fMRI scans from 21 normal controls and 69 subjects at high genetic risk of schizophrenia (27 of whom reported isolated transient psychotic symptoms at interview), performing the Hayling sentence completion task. Three groups of seeds were selected based on (i) regions associated with language comprehension and word production, (ii) regions reported in the schizophrenia literature, and finally (iii) a group of empirically derived seeds. Maps of cross correlation coefficients were computed by measuring the correlation between the time course of the seed and the time courses for all the other brain voxels. In order to reduce the amount of cross correlation purely induced by task related effects, task conditions were modelled and removed from the data. We also sought to confirm functional connectivity results in two additional tasks performed by the subjects during the same scanning session, a verbal encoding and a retrieval task. This was performed to control for the large number of multiple comparisons carried out in the main analysis. Results: We found disruptions in connectivity across the different tasks between lateral-medial prefrontal, prefrontal-temporal, prefrontal-cerebellar, and prefrontal-parietal regions in high risk subjects compared with controls. These results are not confounded by anti-psychotic medication, by the effects of illness, or by performance differences between groups. Connectivity between the right inferior frontal gyrus and bilateral medial frontal gyrus also showed a strong tendency to be related to the degree of heritability within the high risk group. Abnormal patterns of connectivity were also found specifically in those with psychotic symptoms, however these were not replicated in the additional tasks. Conclusion: Connectivity abnormalities in high risk subjects across the different paradigms may represent the inherited neurophysiological basis of the diverse deficits seen in the established state and in individuals at high risk of schizophrenia. This work was funded by an MRC programme grant. Scanning was conducted at the Brain Imaging Research Centre (BIRC) for Scotland, funded by SHEFC. 1. Friston KJ et al., Cereb. Cortex 6, 156 (1996). 2. Lawrie SM et al., Biol. Psychiatry 51, 1008 (2002). 3. Schlosser R et al., Neuroimage 19, 751 (2003). 4. Kim JJ et al., Am. J. Psychiatry 160, 919 (2003). 5. Spence SA et al., Br. J. Psychiatry 176, 52 (2000). e241

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 151 The stuttering brain: Less lateralized and less asymmetric. A combined fMRI and VBM study in stuttering adults Karsten Specht 1 , Joachim Schläfer 2 , C. Paul Stracke 3 , Jürgen Reul 4 , Rolf Biniek 2 of Medicine, Research Center Jülich, Germany, 2 Department of Neurology, Rheinische Kliniken, Bonn, Germany, 3 Department of Radiology, University Hospital Aachen, Germany, 4 Department of Neuroradiology, Robert-Janker Klinik, Bonn, Germany

1 Institute

Introduction In this study, we investigated a group of stuttering adults and compared them to a control group. Since stuttering while fMRI-scanning could cause several artefacts, we performed an attentive listening task. In a previous study, we demonstrated, that this task activates areas, relevant for phonological and semantical processing [1]. In addition to that, we combined the functional study with a voxel-based morphometry (VBM) study to test for differences in grey-matter density between the stuttering and the control subjects. Recent neuroimaging studies are demonstrating that stuttering involves multiple neural components. Therefore, we focused our analysis mainly onto the role of frontal areas (including Broca’s area), the left temporale lobe, and the cingulate gyrus. Methods The 12 right-handed healthy subjects and the 12 right-handed stuttering subjects were instructed to listen attentively to the auditory stimuli. To investigate the lateralisation effects, we used three types of auditory stimuli. First, pure tones with a frequency range of 400-1600 Hz, second, sounds of animals and instruments, and third, words with one or two syllables. The order of stimuli was pseudo randomised and arranged as a single session, event-related paradigm. In addition to that, we acquired high-resolution T1-weighted images for a voxel-based morphometry analysis. All data were processed with SPM99. Results Comparing the activations during word perception for the two groups, we detected a reduced activation in the stuttering subjects in the left prefrontal cortex and the cingulate gyrus. In the opposite contrast, the stuttering subjects didn’t show any higher activations. Furthermore, we detected a reduced left lateralisation in the pre-frontal cortex in the stuttering subjects. Moreover, the stuttering subjects showed a significantly reduced activation in the left superior temporal sulcus. Focusing onto the temporal plane, we detected in the VBM analysis a reduced asymmetry between the left and right temporal plane, compared to the control group. Beside this, we detected a reduced grey matter density within the anterior cingulate gyrus for the stuttering subjects as well. On the other hand, the right BA44 of the stuttering adults showed a higher density in grey matter, compared to the controls. Discussion This study demonstrated converging results from a functional and a morphometrical imaging study. In the literature, several areas are discussed in the circ*mstance with persistent developmental stuttering. Our study supports the view of a reduced left-right lateralisation for phonological processing. The planum temporale, as well as the superior temporal sulcus, which are relevant for the processing of phonological cues, showed reduced activations. Further, in pre-frontal areas, we uncovered a reduced lateralisation in the functional and a reduced asymmetry in the morphometrical analysis. Finally, the anterior cingulate gyrus, which is involved also in the coordination of speech-motor actions, showed also a reduced density in the grey-matter. In summary, we uncovered that the phonological processing in the stuttering adults is less lateralized, and the reduced grey-matter density in anterior cingulate gyrus could be hint for a reduced motor control during articulation. [1] Specht and Reul 2003, Neuroimage(20), p1944

e242

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 152 Using lesion-behaviour correlations to study language function. Emmanuel A. Stamatakis 1 , Lorraine K. Tyler 1,2 1 Department of Experimental Psychology, University of Cambridge, Cambridge, UK, 2 Wolfson Brain Imaging Unit, University of Cambridge, Cambridge, UK The use of lesion data to make inferences about the neural basis of cognitive functions is a useful and widespread practise in cognitive neuropsychology (Damasio & Damasio, 1989 to Bates et al., 2003). We propose a novel approach for the determination of lesion/behaviour relationships, which does not require explicit lesion quantification but instead utilises signal intensity level on a voxel by voxel basis. The method involves pre-processing T1-weighted images by spatially normalising them to the MNI template in SPM99 (www.fil.ion.ucl.ac.uk). The images are then skull stripped and smoothed with a 10mm Gaussian kernel. Covariate analysis is then carried out in the context of the general linear model (Friston et al., 1995). This type of analysis allows you to enter a value/score for each scan and effectively examine correlations of the signal at each image voxel with a covariate of interest. Confounding covariates can also be used in order to remove effects due to these covariates from the final result. To assess this approach we carried out a study in which we correlated the signal intensity across the scans of 19 brain-damaged patients with their performance on a speeded lexical decision task. Patients were included in this study if they had had a T1 weighted 3D MRI scan and could perform the lexical decision task; they were not selected on the basis of either their lesion location or their pattern of behavioural scores [see figure for lesion frequency distribution]. All patients were native speakers of English, ranging in age from 25 to 67 (mean=52, SD=14). The experiment examined the patients ability to process spoken words and non-words. Patients heard a mixture of 80 words and 80 pseudo-words and made a speeded lexical decision to each stimulus. Patient age was included in each analysis as a confounding covariate. Results are reported at cluster level corrected for multiple comparisons. Reaction times (RTs) to real words negatively correlated with an area in the left temporal pole, including the inferior and middle temporal gyrus. Longer RTs reflect increasing difficulty of processing and therefore these results suggest that damage to these left temporal regions is associated with increased difficulty in processing the meaning of words. RTs to non-words were negatively correlated with a similar region of left temporal cortex. RTs to both real words and non-words did not correlate with signal intensity in the right temporal cortex despite the presence of lesions in this area. The fact that signal intensity in the same left temporal regions correlates with real and non-word RTs is consistent with neuroimaging data showing that real words and non-words activate the same regions of temporal cortex (Binder et al., 2000). This shows that listeners attempt to map speech sounds onto meaning irrespective of whether they are hearing real words or non-words, consistent with current theories of spoken word recognition (Marslen-Wilson, 1987), and with the view that lexical decision typically requires access to word meaning.

Lesion frequency distribution in 19 patients included in the analysis. The lesions are superimposed on the mean patient T1 weighted image. Talairach z values in mm are shown on the right hand side of each panel. Purple indicates the least frequently occurring lesion sites and red the most frequent. e243

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 153 An fMRI study of Language Processing in Normal Siblings and Siblings at High Genetic Risk for Schizophrenia a Pilot Study. 1 Center

Kamila U Szulc 1,2 , Neha Desai 1 , Raj Sangoi 1 , Lynn E DeLisi 1,3 for Advanced Brain Imaging, Nathan S. Kline Institute, Orangeburg, USA, 2 Fordham University, New York, USA, 3 NYU School of Medicine, New York, USA

Introduction This study is aimed at investigating the hypothesis that aberrant language processing is an underlying feature of schizophrenia. Siblings at high genetic risk for schizophrenia and normal siblings are being examined to asses the functional and neuronal correlates of language processing in both groups. Methods The stimuli consist of 50 words and 50 pseudowords. Words and pseudowords were matched for length (number of letters). Words were obtained from the MRC Psycholinguistic Database (http:www.itd.clrc.ac.uk) according to the following criteria: 1) number of letters 5, and 2) imagability rating - from 550 to 700. The results of the search were entered into Excel, randomized, and first 50 neutral pseudowords were chosen. Pseudowords come from the study of Pexman et al. (2002). Stimuli are presented in blocks, which are alternated with motor control blocks and rest blocks. The subjects are instructed to press the left mouse button if they think that they saw a word, and to press the right button otherwise. During the control task subjects are presented with strings of Rs or strings of Ls. They press the right button if they see a string of Rs, and the left button if they see a string of Ls. There is a total of 11 blocks in one run. The duration of one block is 30s. 10 stimuli are presented during one block. Each stimulus is presented for 1000ms and an interstimulus interval is 3000ms. The number of words and pseudowords presented randomly during one run is counterbalanced across the run and across the subjects. There are a total of 4 runs for each subject. Echo-planar images (EPI) are acquired using a 1.5 Siemens Vision (TR=2s, TE=40ms, slice thickness = 5mm, 24 slices, pixel size=3.5 x 3.5mm, matrix=64x64). The data analysis is being performed using SPM99. Results Preliminary analysis of normal controls for the word recognition task versus rest shows significant activation in the left superior temporal gyrus, Brocas area, right cerebellum, and bilaterally in parietal regions. Conclusion The results indicate that the experimental design used in this study is highly efficient in evoking activity in brain regions previously associated with language processing in individual subjects. Analyses are ongoing to examine patients with schizophrenia and their unaffected siblings who may be at high genetic risk for developing illness. The heritability of variation in language processing will be examined and its relationship to the inheritance of schizophrenia. References Cohen, L., et al. (2002) Brain, 125, 1054-69 DeLisi (2001). Schizophrenia Bulletin, 27, 481-96 Pexman, P., et al. (2002). Journal of Experimental Psychology: Learning, Memory, & Cognition, 28, 572-84

e244

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 154 Event-related fMRI reveals modality-independent repetition priming in object and spoken word naming after a one day delay. Miranda van Turennout , Esther Aarts F.C. Donders Centre for Cognitive Neuroimaging, Nijmegen, The Netherlands Introduction Naming objects and words involves distinct stages of perceptual and higher-order processing. Previous studies have shown that activity in brain regions subserving these processes is affected by repetition, even after a long delay (e.g., van Turennout et al., 2000; Wagner et al., 2000). Here, we use identical and cross-modal repetition in object and spoken word naming to investigate the neural dissociation between perceptual, and higher-order priming in both the visual and auditory modality. Method Sixteen subjects participated in the experiment. The experiment included a study session, and a scanning session after 24 hours. In both sessions, pictures of objects, and spoken object names were presented randomly intermixed. Subjects were instructed to name the pictures and spoken words using a noun phrase (such as, little table). In Dutch, as in, for example, German and French, nouns are characterized by syntactic gender. This is a purely syntactic property that is retrieved from the mental lexicon during noun phrase production to determine the definite article of the noun, or the adjectival suffix. Thus, subjects had to access the mental lexicon to retrieve the correct gender during both object and word naming. Whole brain images (36, 3.5 mm axial slices, TR=2268 msec, FOV=224mm, TE=30) were collected on a Siemens Trio scanner while subjects named novel and repeated objects and words, 24 hours after the study session. Half of the study items were repeated in their identical form, the other half in the other modality. For example, subjects saw a picture of an anchor on day 1, and on day 2 they either saw the same picture again (identical repetition), or they heard the word "anchor" (cross-modal repetition). Visual random-dot patterns, and pink-noise sounds served as low level control stimuli. Results and Discussion Brain imaging data showed that, compared to spoken words, objects elicited more activity in posterior visual brain areas. Spoken words elicited more activity in brain regions known to be involved in auditory word processing (bilateral superior temporal gyrus). Consistent with earlier data, identical object repetition resulted in decreased activity in occipitotemporal and left inferior frontal regions. A region showing decreased activity for both identical and cross-modal repetition of objects was the left fusiform gyrus (Figure 1). The left fusiform gyrus was involved in object but not spoken word naming, supporting the idea that this is a unimodal visual brain area. Cross-modal priming observed in this region possibly reflects top-down modulation by higher-order brain regions. The left inferior frontal cortex showed decreased activity following repeated noun phrase production in object as well as in spoken word naming (Figure 2). In addition, increased activity for repeated as compared to novel noun phrase production was observed in the left angular gyrus. This increase occurred independent of modality and repetition type, indicating an amodal effect (Figure 2). In conclusion, we observed long-term identical, and cross-modal repetition priming in object as well as spoken word naming. These results suggest a hierarchical lexico-semantic network, involving downstream unimodal, and higher-order heteromodal brain regions.

Figure 1. Identical and cross-modal repetition priming in object naming (decreases in red, increases in blue). e245

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

Figure 2. Repetition-related changes in activity during both object and word naming.

e246

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 155 Diagnosis of language dominant hemisphere by optical topography using primary component analysis. Eiju Watanabe 1 , Shingo Kawasaki 2 , Ayako Isoo 1 , Naoki Tanaka 2 1 Department of neurosurgery, Tokyo metropolitan police hospital, 2 Advanced Research Laboratory, Hitachi, Ltd. Intracarotid amytal test (IAT) has been a gold standard in diagnosis of hemispheric language dominance. As a non-invasive alternative to IAT, we have been examining the feasibility of near infrared spectroscopic topography (NIRS). It uses near infrared light projected from the scalp surface, receiving at scalp surface 30 mm away from the projection probe. The reflected portion of the light carries the blood volume signal of the brain tissue underneath the optical probes. [Method] We measured blood volume change at 48 points covering bilateral frontal lobes. Word generation task with duration of 15 seconds is used as stimulus, separated 30 seconds of rest period. Five trials were averaged for one session. After averaging, language dominance was decided by two methods; i.e. (1) source wave analysis (SWA) (2) primary component analysis (PCA). SWA: Cross-correlation between averaged oxy-hemoglobin ([Hb-Oxy]) waveform and expected reference response waveform was calculated. Frontolateral channels with cross-correlation coefficiency larger than 0.65 were selected and sum of the peek height of these channels were used for laterality index (LI): LI=(Σ(L)-Σ(R))/(Σ(L)+Σ(R)). PCA: Forty-eight wave-components were extracted from a set of 48 [Hb-Oxy] time course by PCA, which were executed on MATLAB. A component, which showed highest similarity with expected time course, was objectively selected using cross-correlation analysis with reference response waveform. The rate of contribution of the component was evaluated for each channel data. The hemispheric dominance was determined using the LI, which was calculated from the frontolateral contribution data. [Cases] Twenty six pre-surgical patients who underwent both IAT and NIRS were included in this study. 20 cases were candidates for epilepsy surgery, 6 cases were those for tumor surgery. [Result] With SWA of NIRS data, 16 cases (62%) showed concordance with IAT, 3 cases (11%) showed no definite response, 7 cases (27%) showed activation on the opposite to the IAT dominant side. On the centrally, LI with PCA method was concordant with IAT in 22 cases (85%), 4 cases (15%) showed the activation on the opposite-side to IAT. [Conclusion] Results of present study show that PCA method increased the reliability of NIRS in the diagnosis of language dominance. These results suggest that, in addition to serving as a tool for investigating higher cognitive brain functions, NIRS has significant clinical potential as a non-invasive method of language lateralization.

e247

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 156 Functional imaging predictors of schizophrenia. Heather C Whalley 1 , Enrico Simonotto 1, 2 , Ian Marshall 3 , Klaus P Ebmeier 1 , David GC Owens 1 , Nigel H Goddard 4 , Eve C Johnstone 1 , Stephen M Lawrie 1 1 Division of Psychiatry, University of Edinburgh, UK, 2 MRI Devices, Waukesha, WI, USA, 3 Division of Medical Physics, University of Edinburgh, UK, 4 School of Informatics, University of Edinburgh, UK Introduction: Schizophrenia is an incapacitating psychiatric disorder characterized by hallucinations and delusions. The established condition has been associated with structural and functional brain abnormalities, but the timing of these abnormalities in relation to development of the illness is unclear. Prospective studies of young individuals at high genetic risk of the disorder allow the investigation of whether such abnormalities predate development of the illness. Methods: We are engaged in an ongoing study of individuals at high risk of developing schizophrenia as they have at least two relatives with the disorder. Previously, fMRI in these subjects using a covert verbal initiation paradigm (the Hayling sentence completion task) found localised abnormalities in those at high risk with isolated psychotic symptoms in the left inferior parietal cortex. At the time of the scan none of the participants met criteria for any psychiatric disorder; however four subjects have subsequently developed schizophrenia. This report concerns the baseline functional imaging findings in these subjects compared to normal controls (n=21), those at high risk with (n=24), and without psychotic symptoms (n=41). We examined two contrasts, sentence completion (all four levels of task difficulty) versus rest, and a parametric contrast identifying regions were there was increased activation with increasing task difficulty. Results and discussion: There were no significant differences between the groups in terms of gender, IQ, handedness, or genetic liability. Those who subsequently became ill were however significantly younger and demonstrated more within scanner movement than the other groups, hence these measures were entered as nuisance covariates in the model. Behavioural measures indicated that the subjects were performing the task appropriately in the scanner. There were no significant differences between the groups in terms of performance measures. Those who subsequently became ill demonstrated activation differences mainly in anterior cingulate, temporolimbic and cerebellar regions. They also demonstrated increased left parietal activation although this was not significant. Abnormal activation in these regions is highly consistent with studies in established schizophrenia, and with structural and functional imaging studies of high risk subjects. Although the current findings should be considered cautiously as there are only as yet four subjects who have subsequently become ill, they suggest that there are functional abnormalities present before the development of schizophrenia, which are distinguishable from normal controls and those at high risk who have not developed the disorder. This work was funded by an MRC programme grant. Scanning was conducted at the Brain Imaging Research Centre (BIRC) for Scotland funded by SHEFC.

e248

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

MO 157 The neuronal integrity of the temporal lobe and language lateralisation in temporal lobe epilepsy Kris K.-S. Yuen , Regula S Briellmann , R. Mark Wellard , David F. Abbott , Graeme D. Jackson Brain Research Institute Introduction: Patients with temporal lobe epilepsy (TLE) are known to show increased incidence of atypical language dominance. However, the factors contributing to this are not well known. Language transfer may be influenced by neuronal integrity of the temporal lobe. N-acetyl-aspartate (NA) is a marker for neuronal integrity that can be measured with proton MRS. Here we assess the relationship between language lateralisation and temporal lobe NA concentration in TLE patients. Methods: MR imaging was performed at 3T. Functional sequences used Gradient-Echo Echo-Planar Imaging (EPI) with whole brain coverage (22 axial slices, TE 40ms, TR 3000ms). Analysis was performed with SPM99 and iBrain®. For language activation, a noun-verb generation task was used. Language lateralisation was determined from the number of activated pixels in language-related areas using a laterality index (LI). Language lateralisation was considered atypical if the LI was <= 0.2. For Proton MRS bilateral temporal lobe single voxel spectra were acquired (TR=3.0 s; TE=30ms). Concentrations of N-acetylaspartate (NA), creatine (Cr), phosphocholine (Cho), myoinositol (mI), and glutamine/glutamate were obtained using LCModel. Thirty-two TLE patients were included, 13 had right TLE, and 19 had left TLE. All patients underwent both fMRI language and Proton MRS examinations. Language lateralisation in patients was compared to a series of 33 controls. Proton MRS in patients was compared to 48 controls. The relationship between fMRI and MRS was compared to 19 controls that had both examinations. Results: The language LI was not different in patients (0.36 ±0.3) compared to controls (0.51 ±0.3, table). There was also no difference in the LI between left TLE and right TLE. However, atypical language lateralisation was more frequent in patients (33%), than in controls (9%, p=0.03). The frequency of atypical lateralisation was the same in left TLE (32%) and right TLE (36%). Left-handedness was present in 25% of the patients, but only in 3% of the controls. All but one of the left-handed patients showed atypical language lateralisation. The NA concentration was reduced in patients in both the right (4.9 ±0.9) and left (5.1 ±1.0) temporal lobe, when compared to controls (table, p<=0.05). When split for the side of the focus, left TLE patients showed a decrease in the left temporal lobe (p=0.002), whereas right TLE patients were not statistically different from controls. Patients with atypical language lateralisation showed higher right temporal NA concentrations (5.3 ±0.8) than patients with typical lateralisation (4.7 ±0.9, p<=0.05), whereas left temporal NA showed no difference in relation to language lateralisation. This effect was particularly strong in right TLE patients. In controls, there was no difference in the NA concentration between typical or atypical language lateralisation. Conclusion: Atypical language lateralization was associated with relative preservation of neuronal integrity of the right temporal lobe, even when this was the site of the seizure focus. One interpretation may be that in these cases atypical lateralisation does not reflect language transfer as a consequence of epilepsy, but indicates pre-existing probably genetically determined language processing in the right hemisphere.

e249

Abstracts presented at the 10th International Conference on Functional Mapping of the Human Brain, June 13-17, 2004, Budapest, Hungary

table

e250