The Inferior Frontal Gyrus and Phonological Processing: an Investigation Using Rtms

Home , Inferior frontal gyrus, Inferior frontal sulcus, Middle frontal gyrus, Operculum (brain)

The Inferior Frontal Gyrus and Phonological Processing: An Investigation using rTMS

Philip Nixon1, Jenia Lazarova2, Iona Hodinott-Hill1, Patricia Gough1, and Richard Passingham1 Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021

Abstract & Repetitive transcranial magnetic stimulation (rTMS) offers a task. Delivered at a time when subjects were required to powerful new technique for investigating the distinct contri- remember the sound of a visually presented word, rTMS butions of the cortical language areas. We have used this impaired the accuracy with which they subsequently per- method to examine the role of the left inferior frontal gyrus formed the task. However, when delivered later in the trial, as ( IFG) in phonological processing and verbal working memory. the subjects compared the remembered word with a given Functional neuroimaging studies have implicated the posteri- pseudoword, rTMS did not impair accuracy. Performance by or part of the left IFG in both phonological decision making the same subjects on a control task that required the and subvocal rehearsal mechanisms, but imaging is a correla- processing of nonverbal visual stimuli was unaffected by the tional method and it is therefore necessary to determine rTMS. Similarly, performance on both tasks was unaffected by whether this region is essential for such processes. In this rTMS delivered over a more anterior site (pars triangularis). paper we present the results of two experiments in which We conclude that the opercular region of the IFG is necessary rTMS was applied over the frontal operculum while subjects for the normal operation of phonologically based working performed a delayed phonological matching task. We com- memory mechanisms. Furthermore, this study shows that pared the effects of disrupting this area either during the delay rTMS can shed further light on the precise role of cortical (memory) phase or at the response (decision) phase of the language areas in humans. &

INTRODUCTION active does not establish that they are necessary for Transcranial magnetic stimulation (TMS) provides an phonological processing. effective new method for investigating cognitive func- TMS has been used to examine the roles of the tions in human subjects (see Jahanshahi & Rothwell, temporal lobe language areas in picture-naming tasks 2000, for a review). TMS can be used to test the (Stewart, Meyer, Frith, & Rothwell, 2001; Topper, Motta- functional necessity of the cortical activations seen in ghy, Brugmann, Noth, & Huber, 1998). Studies of frontal functional neuroimaging studies on human subjects. lobe language functions have remained largely restricted Neuroimaging techniques have demonstrated wide- to frontal regions outside Broca’s area; these have dem- spread left-sided ventral frontal activations in relation to onstrated that TMS can be used to disrupt verbal working many forms of linguistic processing (Price, Indefrey, & memory (WM) (Mottaghy et al., 2000; Mottaghy, Doring, Turennout, 1999). Hemodynamic changes that have been Muller-Gartner, Topper, & Krause, 2002) and verb gen- attributed to the phonological encoding of words into eration (Shapiro, Pascual-Leone, Mottaghy, Gangitano, & articulatory plans have been reported during repetition of Caramazza, 2001). TMS studies attempting to reproduce heard words (Price et al., 1996), silent repetition of non- some of the production deficits associated with Broca’s words (Warburton et al., 1996), phonemic fluency tasks aphasia have had limited success (Stewart, Walsh, Frith, (Paulesu et al., 1997), and when making phonological & Rothwell, 2001; Flitman et al., 1998; Epstein et al., 1996, judgements (Poldrack et al., 1999; Demonet et al., 1992; 1999). Speech disruption can be obtained using repeti- Zattore, Evans, Meyer, & Gjedde, 1992). All these activa- tive trains of TMS pulses at levels well above the thresh- tions were localized to a region that included the poste- old for eliciting movements of the digits when applied rior (opercular) division of Broca’s area (BA 44) and tissue over motor cortex. However, the sites at which such in the descending limb of the precentral sulcus (BA 6). disruption generally occurs tend not to lie below the However, the fact that these areas are shown to be more inferior frontal sulcus where Broca’s area is located (Stewart, Meyer, et al., 2001; Bartres-Faz et al., 1999). Furthermore, the relatively severe speech production deficits associated with Broca’s aphasia may to some 1University of Oxford, 2Whittier College degree result from damage lying outside this area, espe-

D 2004 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 16:2, pp. 289–300

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 cially to the white matter lying below the cortical surface cause movements of the temporal muscle/jaw are com- (Willmes & Poeck, 1993; Signoret, Castaigne, Lhermitte, monly elicited by TMS over the ventral part of the frontal Abelanet, & Lavorel, 1984). TMS typically affects a region lobes, the subjects were instructed not to use overt of cortex no larger than 1–2 cm in diameter (Walsh & speech during performance of the homophone task. Cowey, 2000) and might not therefore be expected to Covert speech has been shown to elicit strong hemody- reproduce the clinical effects of lesions resulting from namic responses in the posterior division of Broca’s area major arterial strokes or the removal of large tumors. (BA 44), near its border with premotor cortex (BA 6) In the study presented here, we examine whether or (Grezes & Decety, 2001; Warburton et al., 1996; Paulesu, not it is possible to interfere with phonological process- Frith, & Frackowiak, 1993). This region was the target Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 ing by applying short, high-frequency trains of TMS pulses for TMS stimulation in this experiment (see Figure 5C). over Broca’s area. We conducted two experiments in which TMS was applied over this region while normal Results and Discussion subjects performed a delayed phonological matching task. The design of the first experiment allowed us to On non-TMS trials, subjects generally performed both contrast the effects of the TMS on verbal WM and on tasks with a high degree of accuracy. For the homo- decision-making processes. The second experiment com- phone task the mean error rate (ER) was 0.83 trials and pared the effects of applying TMS separately to the for the control task it was 0.67 trials. The mean RTs on anterior (BA 45) and posterior (BA 44) divisions of Broca’s non-TMS trials were 831 msec (homophone task) and area. In order to control for any nonspecific effects of TMS 584 msec (control task). that might affect performance on this task, the results When TMS was applied during the delay it significantly were compared with the effects of TMS on a nonverbal impaired accuracy on the homophone task, t(5) = 2.70, visual matching task. Because the TMS in these experi- p < .05, but not on the control task, t(5) = .10, ns ments caused reaction time (RT) decreases, a third (Figure 1A). The differential effect on the two tasks was experiment was conducted using sham TMS to test also significant, t(5) = 3.16, p < .05. TMS applied over whether or not these decreases were caused by effects the same site at the time of the response was also other than the direct magnetic stimulation of the brain. accompanied by an increase in errors but this did not reach significance because of the greater between-subject variability (Figure 1B). RESULTS On both tasks, there were significant decreases in RT (Figure 1), both when TMS was applied in the delay, Experiment 1: Contrasting Working Memory and homophones: t(5) = À6.29, p < .01; control: t(5) = Decision Processes À6.74, p < .01, and at response, homophones: t(5) = In the first experiment we compared the effects of À3.36, p < .05; control: t(5) = À4.65, p < .01. However, applying TMS either while subjects remembered the there were no significant differences between the tasks sound of a word they had seen, or when they were on this measure, delay: t(5) = À2.15, ns; response: required to match this word with the sound of a t(5)= À1.09, ns. subsequently presented pseudoword (homophone The results clearly show that TMS had a greater effect task). The control task differed from this task only in on a task that required processing of verbal material the nature of the stimuli used (Korean letters). Subjects than on a similar task that required only visual matching. were thus required to remember and subsequently The stimulation was more effective when applied during match visual symbols that lacked (to non-Korean speak- the delay period between presentation of the verbal ers) any phonological representation. The experiment stimuli. This is consistent with the evidence that the was based on a 2 Â 3 block design with Task (phono- posterior inferior frontal gyrus (IFG) is activated during logical, visual) and TMS (none, delay, response) as the maintenance of phonological representations (Pau- manipulated factors. We were thus able to compare lesu et al., 1993). It also showed that this activity is the effects of applying TMS during the delay period necessary for accurate performance. (1 sec), while subjects held cue information in memory, Facilitatory effects on RT were seen for all TMS blocks with TMS applied later in the trial as subjects compared and are of interest for two reasons. First, TMS is generally the memorized cue with a second cue and implemented considered to disrupt processing in higher cortical areas their response. (Walsh & Cowey, 2000; Pascual-Leone et al., 1999), and The two tasks were presented consecutively with task this is indexed by measurable increases in RT on cogni- order, but not TMS onset, counterbalanced across sub- tive tasks. However, this outcome can be counterbal- jects. Short trains of TMS were applied over the pars anced by a facilitation of responses resulting from opercularis region of the left hemisphere at approxi- intersensory facilitation effects (Hershenson, 1963). We mately 120% of the subject’s active motor threshold. The measured the magnitude of this effect in Experiment 3 by site was estimated with respect to the physiologically testing with sham stimulation. The second concern is determined hand representation of motor cortex. Be- that given the RT decreases, it could be argued that they

290 Journal of Cognitive Neuroscience Volume 16, Number 2

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 Figure 1. Results of Experiment 1. Changes in percent error rate (%ER) and RT for TMS versus non-TMS trials. (A) When rTMS was applied over the posterior (opercular) site during the delay period the mean increase in %ER was much larger for homophone task trials ( black bars) than for control task trials (gray bars). There was a similar Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 decrease in RT for both tasks at this site (lower graph). (B) When TMS was applied at the decision phase of each task, there was some evidence of an increase in %ER for the homophone task but the variance of the mean was very high. As for stimulation applied during the delay, there was a strong facilitation of RT for both tasks (lower graph). The standard error of the means is shown by the thin bars. *p < .05, **p < .01.

may have had an adverse impact on the accuracy with control site for the uncomfortable facial muscle which subjects performed the task (due to a speed–error twitches associated with ventral frontal repetitive TMS trade-off ). However, it can be seen (Figure 1A) that while (rTMS). The design of this experiment thus differs from there were decreases in RT on both tasks, accuracy was the previous one in that the manipulated factor TMS impaired only on the verbal task. now refers to site (i.e., none, anterior, posterior) rather than onset, with TMS being applied exclusively during the delay period. A further modification was made by Experiment 2: Contrasting Pars Opercularis and increasing the within-trial delay period to 2000 msec to Pars Triangularis increase the WM load, with the TMS onset starting at In this experiment we sought to replicate the main 1500 msec and lasting 500 msec. finding of the previous experiment, that is, the detri- mental effect of TMS on verbal WM, and to compare Results and Discussion this with the effects of TMS over the pars triangularis region of Broca’s area. The design of the current The results for stimulation over the posterior site were experiment improves on the previous one because of similar to those of Experiment 1. TMS applied during the the introduction of a frameless stereotaxy imaging delay period impaired accuracy for the phonological system into the laboratory. This permits a more precise task, t(5) = 2.74, p < .05, but not for the control task neuroanatomical localization of the stimulation target (Figure 2A). The magnitude of the impairment was sites based on structural MRIs of the subjects partici- similar to that observed in the first experiment despite pating in the experiment. The system was used to the increase in delay length in the current design. The enable a better localization of the pars opercularis differential effects of TMS at this site on the two tasks region of the IFG than had been achieved in Experi- were also significant, t(5) = 3.61, p < .05, but there was ment 1 and to allow a comparison of stimulation no difference in its effects over the anterior site, t(3) = applied to this area with stimulation applied to the 1.00, ns (Figure 2B). In addition, a 2 Â 3 ANOVA with adjacent, pars triangularis, region. There is some de- Task (phonological, visual) and TMS (none, posterior, bate as to whether this latter region has a dissociable anterior) as independent factors confirmed a significant role in semantic, rather than phonological, processing interaction, F(2,6) = 5.4, p < .05). (McDermott, Petersen, Watson, & Ojemann, 2003; Pol- Significant decreases in RT were seen for the homo- drack et al., 1999; Price et al., 1999; Demonet et al., phone task when TMS was applied over the opercular 1992). We hypothesized that TMS applied here would region, t(5) = À3.50, p < .05, and similarly for the be less likely to impair verbal WM for phonological control task when TMS was applied over the triangular representations of words and would provide a suitable region, t(3) = À5.51, p < .05, but there were no

Nixon et al. 291

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 Figure 2. Results of Experiment 2. Changes in percent error rate (%ER) and RT for TMS versus non-TMS trials. (A) The effects of rTMS applied over pars opercularis during the delay period were reversed for homophone task trials ( black bars) as compared to control task trials (gray bars).

There was a marked decrease Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 in RT for homophones, as seen in Experiment 1. (B) However, when TMS was applied over pars triangularis, there were no significant changes in %ER seen on either task. As before, there was a marked facilitation of RT on both tasks (lower graph). The standard error of the means is shown by the thin bars. *p < .05.

significant differences between tasks at either site effect on performance, but this does not prove that (Figure 5) and no interactions, F(2,6) = 1.39, ns. the effect was directly caused by the TMS pulses inter- The results of this experiment confirm the findings of acting with the underlying brain tissue. It might be that Experiment 1 for TMS applied over the posterior (pars nonspecific effects associated with the stimulation (e.g., opercularis) division of the IFG during the delay period. auditory and somatic sensations and the preceding The absence of any increase in ER for TMS over the anxiety) could summate to distract attention during anterior (pars triangularis) region suggests this region is critical stages in the performance of the task. Further- not necessary for maintaining phonological representa- more, all subjects demonstrated a strong TMS-related tions of words over time. Yet neuroimaging studies have decrease in RT on the phonological task. For stimulation generally reported that both regions are active during applied in the delay period the mean RT gain was in the phonological and during semantic processing (Devlin, order of 100 msec (Figures 1A and 2), and for stimula- Matthews, & Rushworth, 2003; Paulesu et al., 1997; tion applied at response the gain was approximately Demonet et al., 1992). It is possible that both regions 150 msec (Figure 1B). Facilitation effects have been are coactivated because semantic processing invokes observed on other language tasks (Topper et al., implicit phonological representations of verbal stimuli, 1998). Such decreases may also have contributed to and vice versa. (Devlin et al., 2003; Poldrack et al., 1999; the increase in ER by promoting premature responses Price et al., 1999). However, the results of Experiment 2 from the subjects (due to a speed–error trade-off ). show that disruption of activity within anterior Broca’s Finally, the decreases in RT might reflect learning-related area does not interfere with phonological processing, changes, because the TMS blocks always followed the supporting claims that areas BA 44 and BA 45 are indeed control blocks. functionally dissociable. Correspondingly, Devlin et al. The purpose of this final experiment was to examine (2003) have used TMS to demonstrate disruption of these possible confounding effects in a third group of semantic processing when applied over pars triangularis subjects performing the same tasks but under condi- (BA 45) where it borders with pars orbitalis (BA 47). tions of sham TMS only. The design of the experiment was largely identical to the previous one except that the TMS coil was angled in such a way that the Experiment 3: Effects of Sham-TMS on magnetic field was not directed at the scalp but at Performance right angles to it. This had the effect of exposing the In the previous experiments, there was an increase in ER subject to the ‘‘clicking’’ sound associated with rTMS when TMS was applied over the opercular region of the but there were no peripheral somatosensory effects or IFG. Clearly, the stimulation was having a disruptive interactions with neuronal tissue.

292 Journal of Cognitive Neuroscience Volume 16, Number 2

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 Results and Discussion group all showed significant decreases in RT during the sham TMS trials, their accuracy was not impaired. Subjects generally performed faster and more accurately It also shows that the noise of the TMS pulses did not during sham TMS trials than during nonsham trials distract the subjects during the delay period enough (Figure 3). The ER decreased slightly for both tasks to impair performance. during sham TMS trials; however, none of the changes Lastly, the control data obtained in Experiment 3 in ER were significant and there was no significant allowed a between-subjects comparison of the effects of difference between the two tasks for either of the sham TMS in the previous experiments. This post hoc analysis TMS blocks: Block 1: t(5) = .67, ns; Block 2: t(5) = 1.00, confirmed the main finding of a significant decrease in ns. For the phonological task, there were significant Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 response accuracy for the phonological task when TMS decreases in RT for both sham TMS blocks, Block 1: was applied over the posterior part of the IFG, one-way t(5) = À3.65, p < .05; Block 2: t(5) = À2.89, p < .05, ANOVA F(2,15) = 5.53, p = .02, corrected, but revealed and the difference between tasks reached significance no significant differences between groups for RT. for Block 1, t(5) = À3.99, p = .01. For the control task, RT decreases were smaller but significant for Block 2, t(5) = À5.91, p <.01. Finally, there were no significant DISCUSSION differences in RT between successive sham TMS blocks Overall, the results of the three experiments provide that might implicate learning effects as a major source of strong evidence for a disruptive effect of rTMS on the RT decreases in these experiments: homophones: phonological memory when applied over the posterior t(5) = À1.50, ns; control: t(5) = .12, ns. part of Broca’s area. Identical stimulation over the same Given these results, it is likely that the decrease in cortical region did not result in an increase in errors on a RT observed for both tasks can be explained in large nonphonologically based memory task. This finding sup- part by a nonspecific intersensory facilitation effect ports the view that the activations shown in this region (Hershenson, 1963). This effect has been observed by neuroimaging studies reflect its contribution to verbal before with TMS (e.g., Rushworth, Ellison & Walsh, WM strategies such as subvocal articulatory rehearsal 2001; Marzi et al., 1998) and can have the conse- (Davachi, Maril, & Wagner, 2001; Jonides et al., 1998; quence of masking any potential increases in RT Cohen et al., 1997; Fiez et al., 1996; Paulesu et al., 1993). directly related to the disruptive effects of TMS on neural function. The results of this experiment also Cognitive Processes provide further evidence that the significant effects of TMSonresponseaccuracyseenintheprevious While the design of this study was intended to minimize experiments were not the consequence of a speed– the contribution of other cognitive processes, activations error trade-off. Although the subjects in the present in the IFG have also been attributed to learning/encoding

Figure 3. Results of Experiment 3. Changes in percent error rate (%ER) and RT for sham TMS versus non-TMS trials. Neither the first (A) nor second (B) sham TMS blocks revealed increases in %ER for either task. However, significant decreases in RT were associated with the clicking sound of the TMS coil (A + B, lower graphs), suggesting the action of intersensory facilitation effects. The standard error of the means is shown by the thin bars. *p < .05, **p < .01.

Nixon et al. 293

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 (Davachi et al., 2001; Raichle et al., 1994), phonological item is that in the phonological task the subject had to decision making (Rumsey et al., 1997), semantic and read the pseudoword and sound it out. phonological attention (McDermott et al., 2003), and The next question is why the changes in ERs seen in generative processes (Paulesu et al., 1997). The subjects Experiments 1 and 2 were relatively small, although in the present study were not required to generate statistically significant. There are several possible reasons paired associates as occurs in phonological verbal fluency why this may be so. First, the phonological task used in tests. Neither were they required to learn anything from this study was somewhat undemanding, using uncompli- one trial to the next, although the small decreases in RT cated stimuli and short delay lengths. Yet strong activa- observed over consecutive blocks in Experiment 3 sug- tions have been observed in BA 44 when subjects simply Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 gests that the subjects were becoming more efficient at remembered the sound of a letter for a subsequent executing the task with time. rhyming judgment (Paulesu et al., 1993) or when reading In Experiment 1, TMS was applied at the decision short words was compared with observing a line of (response) stage for one block of trials and this caused asterisks (Georgiewa et al., 2002). However, two inde- some subjects to make more errors compared with non- pendent studies have shown that activity in this region TMS trials. However, the variance of the data was large can be influenced either by stimulus difficulty—increas- and no firm conclusions can be drawn from this part of ing with the number of syllables or with pseudowords the study. Nonetheless, the possibility that BA 44 is (Chein & Fiez, 2001)—or by working memory load— concerned with aspects of phonological decision making increasing with the number of nontarget items in an warrants further investigation with TMS. In a PET study n-back sequential letter task (Cohen et al., 1997 ). This by Rumsey et al. (1997), subjects were required to suggests that increasing the delay length in the present decide which of two simultaneously presented non- study, or introducing distractor stimuli during the delay, words either (a) sounded like a real word or (b) visually might have placed more demands on the region such matched a real word. The authors found that compared that the effect of the TMS may also have been greater. with visual fixation, both tasks produced increases in TMS was applied for brief (500 msec or less) periods activity in BA 44, but in a phonological versus ortho- during the delay between stimuli and this may have graphic analysis the significant left IFG activations were been insufficient to disrupt ongoing working memory more widespread, encompassing BA 44, 45, and 46. This mechanisms for long enough to have a more profound indicates that the posterior IFG region tested in the effect. In a more complex design it would be possible to present TMS study is likely to be part of a wider network compare the effects of stimulating at different times of inferior frontal regions that are recruited in phono- during the delay, but increasing the duration of the logical decision making. In another PET study by Fiez pulse train at the frequency used here would not be et al. (1996), it was observed that while covert rehearsal possible without exceeding current safety guidelines of verbal material was associated with left frontal oper- (Wassermann, 1998). Mottaghy et al. (2002) applied cular regions, the active maintenance of novel verbal TMS constantly at 4 Hz during the 30-sec performance stimuli (as opposed to silent counting) also recruited block of a 2-back verbal WM task. As in the present study more dorsolateral prefrontal areas. These and other accuracy, but not RT, was the principal indicator of results suggest that the frontal opercular area (BA 44) behavioral impairment related to prefrontal stimulation. is concerned with relatively low-level aspects of phono- The effects were also somewhat greater than in our logical processing, perhaps including orthographic- study and this may have been related to the chronic to-phonological transformation (Fiez & Petersen, 1998). nature of the TMS or to the longer memory period required (3 sec), or indeed, to the more superior location of the stimulation site used in their study Task Demands (middle frontal gyrus). It could be argued that the reason why TMS interfered A related issue concerns the uniformity with which with performance of the phonological task but not the the stimulation can be applied and the consistency of its control task is that the phonological task was more effects in different subjects (Maeda, Keenan, Tormos, difficult. On non-TMS trials, the subjects were between Topka, & Pascual-Leone, 2000; Pascual-Leone et al., 247 msec (Experiment 1) and 181 msec (Experiment 2) 1998). In the present study, the level of stimulation slower responding on the phonological task than on the was determined by reference to each subject’s motor control task. However, it was TMS applied during the threshold, but there is a question as to whether this is an delay before presentation of the second item that appropriate baseline for applying TMS over nonmotor caused a significant increase in errors. There is no areas of neocortex (Stewart, Walsh, & Rothwell, 2001). reason to suspect that remembering a word is more Thedistancefromthescalptothesurfaceofthe difficult than remembering a Korean symbol; indeed, if prefrontal cortex tends to be greater than the distance anything the opposite is probably the case. The reason from the scalp to the motor cortex and this may be why it took longer to make the phonological judgment relevant (McConnell et al., 2000). The consequence of than the visual judgment on presentation of the second this added variability is that it is difficult to equate the

294 Journal of Cognitive Neuroscience Volume 16, Number 2

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 level of stimulation being applied across subjects with- (delayed homoform test) was identical to the verbal task out considering this variable depth factor. Ideally, a brief except that the stimuli consisted of Korean symbols and cognitive task should be employed to test for an effect of could only be compared by visual form (none of the TMS over the prefrontal cortex, such as is commonly subjects could read or speak Korean). Sample trials from used in TMS studies of the posterior parietal cortex both tasks are shown schematically in Figure 4A. (Ashbridge, Walsh, & Cowey, 1997). For each task the trials were grouped into blocks of Finally, despite the significant methodological refine- 30 experimental trials preceded by a block of 20 ments described in this paper for ensuring the accurate practice trials with error feedback provided on screen. anatomical placement of TMS, loci determined in this Subjects were seated in an adjustable chair approxi- Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 way may not correlate precisely with functional regions mately 60 cm from a 21-in. VDU on which the stimuli of interest. For even greater precision it is necessary to were presented. The first and second fingers of the obtain functional images, using PET or fMRI, of subjects subject’s left hand were placed over the leftmost two performing the same tasks that they will subsequently keys of a response box (PST, Pittsburgh) labeled ‘‘yes’’ perform with TMS (Mottaghy et al., 2000; Paus et al., and ‘‘no,’’ respectively. Visual stimuli were presented in 1997). Used in this way, TMS has the potential to the center of the screen, black text on a white back- provide a powerful method for testing the functional necessity of the cortical activations seen in neuroimaging studies of language, working memory, and other cognitive processes associated with the frontal lobes.

METHODS Participants Six volunteers (3 women) took part in Experiment 1. They were all right-handed and aged between 18 and 33 (mean 21.7). For Experiment 2 another group of 6 volunteers (3 women) were recruited, none of whom had participated in the previous experiment. The subjects were aged between 21 and 46 (mean 28.5). Five were right-handed and 1 was left-handed but this subject had earlier participated in an ERP study that strongly suggested left hemisphere dominance for language. A further 6 subjects (3 women, all right handed) were recruited for Experiment 3. Their mean age was 23.8 years (range 20 to 33). All the participants in this study were native English speakers and were paid for their participation. The study was approved by the local institutional ethics board and all subjects gave their informed written consent beforehand. Screening forms were given before testing to assess handedness and to ensure that none of the subjects had any contraindications to TMS. All subjects were informed that they were under no obligation to complete the experiment and could leave at any time.

Task Design Subjects were presented with two tasks designed to test either verbal or visual WM. In the verbal WM task (delayed homophone test) subjects were required to Figure 4. The behavioral tasks used in Experiments 1–3. (A) The silently read a word (e.g., ‘‘loops’’) and to remember the upper half of this figure shows a typical trial for the delayed sound of the word during a delay period. They were homophone test (phonological matching task) and the lower half then presented with an orthographically and phonolog- shows a typical trial for the delayed homoform test (visual matching ically legal pseudoword (e.g., ‘‘lupes’’ or ‘‘lirps’’) and task). In both cases, subjects reported their decision by pressing either had to decide whether this sounded identical to or a ‘‘yes’’ or ‘‘no’’ response key. (B) Time-line showing the onset and offset of the TMS pulse train (solid rectangles) in Experiments 1 and 2. different from the preceding stimulus. The stimuli were s1 and s2 denote the appearance of the first and second cues. For matched for the number of syllables and the word lists Experiment 1, TMS was applied either during the delay between the were balanced for word frequency. The visual WM task cues or during the response (decision) phase of the trials.

Nixon et al. 295

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 ground in an 18-point font size. After an intertrial provided the stimulation. Previous studies have esti- interval lasting 6000 msec, a fixation crosshair appeared mated that such coil designs can produce functionally in the center of the screen for a further 1000 msec. dissociable effects within a scalp area of 1 cm2 or less This was followed by the appearance of the primary (Brasil-Neto, Pascual-Leone, Valls-Sole, Cohen, & Hal- cue (s1 in Figure 4B), in the same position as the lett, 1992). Two such coils (maximum field strength crosshair, for 1000 msec. This screen was then blanked 2.8 T ) were used interchangeably to offset the effects for a delay period (1000 msec in Experiment 1 and of overheating the coil windings. The TMS generator 2000 msec in Experiments 2 and 3), followed immedi- was connected to a PC computer running custom ately by the onset of the secondary cue (s2 in software to control the timing of the pulse trains Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 Figure 4B). This stimulus persisted until the subject independently of the computer running the behavioral responded by pressing a key to indicate their choice of tasks. The task computer simply sent a brief (1 msec) either a match or nonmatch between s1 and s2. signal to the TMS computer to initiate a pulse train at At the beginning of each task, an instruction page was the required time. shown on the screen for the subjects to read. Subjects were then asked if they understood what was required Parameters and Localization of them. For the homophone test, it was emphasized that they would need to remember the sound of the In an earlier pilot study with 2 TMS experienced subjects primary stimulus rather than its form. At the start of each (1 male, 1 female) we ascertained that TMS-induced block of each task, subjects were reminded to respond speech disruption did not provide reliable localization as quickly and accurately as possible. A block design was of Broca’s area. From posttest examination of structural found to be more agreeable to subjects than a random MRI scans of the two subjects, the region of maximal design because of the increased anxiety inherent in not disruption lay above the left inferior frontal sulcus, in being able to predict stimulation trials in the latter case. agreement with Stewart, Meyer, et al. (2001). The site of After completion of the practice block, subjects pro- stimulation for Experiment 1 was therefore estimated gressed to interleaved blocks of control and rTMS trials. with reference to the hand area of the left motor cortex In Experiment 1, stimulation was applied either 300 msec for each individual subject. The level of stimulation used after the start of the delay period or 300 msec after the was determined with reference to each subject’s active onset of s2. These parameters were determined from motor threshold (aMT). data by Breier, Simos, Zouridakis, and Papanicolaou A Lycra swimming cap with chin ties to prevent (1999), who showed peak activations in the left inferior slipping was fitted over the subject’s head so that the frontal region between 300 and 600 msec after presen- stimulation sites could be clearly marked with 5-mm- tation of verbal stimuli in a rhyme-matching task. In diameter stick-on Velcro spots of different colors. An Experiment 2, stimulation began 1500 msec after the estimate of the hand area was obtained by moving 5 cm appearance of s1and lasted until the onset of s2 lateral to the midline at the vertex and 2 cm rostral. With (Figure 4B). This change was made so that the disrup- the center of the coil (also marked by a Velcro spot) tive effects of the TMS were directed toward the latter placed against the scalp and with its axis lying in the part of the delay period in both experiments. We were rostrocaudal direction, single pulses of TMS were deliv- concerned that if a significant period of time elapsed ered at various locations until a site was found where it between the end of the pulse train and the appearance was possible to elicit small movements of the first or of s2, the subjects might have been able to ‘‘refresh’’ middle fingers of the right hand with the muscles in that their phonological memory trace for s1 so as to over- hand when voluntarily tensed. The minimum level of come the interference arising from the TMS. stimulation required to elicit such movements was then Three of the subjects from each experiment performed determined by gradually reducing the output power the visual WM task first followed by the phonological WM until movements cold no longer be observed. task, and the remaining subjects vice versa. A minimum The target site for stimulation in Experiment 1 was 2-min rest was enforced between blocks and a further determined by measuring 4 cm ventral and 3 cm 10-min rest period was interposed between the two tasks. rostral to the motor cortex site. This was estimated None of the subjects were informed either before or to lie within the opercular region of the inferior frontal during the experiment as to the expected effects of the gyrus, just below the level of the inferior frontal TMS on performance and the experimenter avoided sulcus. The pilot study referred to above provided referring to the visual WM task as a ‘‘control’’ task. structural MRI scans from which the motor site could be located and the target site estimated from cortical TMS features (Figure 5). The TMS coil was held by the experimenter firmly over the target site by aligning Apparatus the spot marking the center of the coil with the A MagStim Super-Rapid (MagStim, Whitland, UK ) TMS appropriate spot on the subject’s head. The handle generator connected to a 50-mm figure-of-eight coil of the coil was generally aligned in the rostrocaudal

296 Journal of Cognitive Neuroscience Volume 16, Number 2

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 of a block of trials and the new coil was positioned in the same way as the first. The level of stimulation used at the target site was variable and limited by the tolerance of each subject to the discomfort associated with rTMS pulse trains. How- ever, the parameters never exceeded the safety guidelines outlined by Wassermann (1998). For Experiment 1 the rTMS pulse train was set at 13.3 Hz for 375 msec, and for Experiment 2 it was set at 10 Hz for 500 msec Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 because the higher frequency was found to be particu- larly unpleasant over the more anterior site. Sample trains of rTMS were administered in steps of 5%, starting from 100% of aMT, until the subject found the level uncomfortable enough to be potentially distracting during task performance. The mean level of stimulation for the subjects in Experiment 1 was 120% of aMT (range 100–130%). These values correspond to approximately 55–75% of the generator’s total power output. For Experiment 2, the levels used were generally lower (mean 110%, range 100–120%) but were the same for both sites. This is because TMS at more anterior frontal sites is associated with greater discomfort due to facial muscle activation resulting in jaw movements and forced eye blinks. Two subjects could not tolerate stimulation over this region due to such effects and only received TMS over the posterior target.

‘‘Brainsight’’ For Experiment 2, high-resolution structural MRI scans were obtained for all subjects. A computerized frameless stereotaxy system (‘‘Brainsight,’’ Rogue Research, Mon- treal) was employed to localize the target sites for stimulation in this experiment. This method greatly improves the anatomical localization for this kind of study and can be used to help target functional brain activity recorded before the TMS session. Briefly, reflective markers attached to the subject’s head are moni- Figure 5. Identification of the TMS sites used in Experiment 1. (A) tored by infrared sensitive cameras to ensure that Coronal (left) and sagittal (left hemisphere) views taken from a movements of the subject’s head are correlated in real structural MRI series of a subject in whom we attempted to ascertain time with the MRI data set. The MRI data are coregis- the optimal site for speech disruption. Various cortical features are labeled to indicate the landmarks used to identify Broca’s area in this tered to the head by reference to easily visible cranial study. AR/HR=ascending and horizontal rami of the lateral fissure; landmarks. Reflective markers attached to the TMS coil CS=central sulcus; IFS=inferior frontal sulcus; IPS=inferior branch of allow the system to track the position of the coil relative the precentral sulcus. (B) A surface rendering of the MRI series shown to the head MRI and optimal coil positioning can be in Figure 5A with symbols superimposed to indicate the TMS target determined for each target site. sites. X shows the approximate location of the hand area of the motor cortex. O denotes the posterior (opercular) target for the TMS used in Identification of the appropriate sulci and gyri that Experiment 1. The more anterior (triangular, T) division of Broca’s area assist in defining the classic Broca’s area was made by was deliberately avoided in this experiment. (C) Coronal (left) and reference to the atlas of Duvernoy, Bourgouin, Cabanis, sagittal (left hemisphere) views from a normalized MRI brain showing and Cattin (1999) and an MRI examination technique the location of the activation reported when subjects silently repeat the developed by Bacon-Moore, Crosson, Gokay, Leonard, sound ‘‘lah’’ at 2 Hz (Paulesu and Passingham, unpublished observations). The coordinates are given in Tailarach space. and Foundas (2000). The boundary between the pars opercularis and pars triangularis divisions of Broca’s area was a line drawn along the anterior ascending ramus of plane with some minor (max 108) adjustments to the sylvian fissure (Figure 5A). It should be noted that minimize subject discomfort. If a coil needed to be this method provides a limited anatomical definition of exchanged due to warming, this was done at the end areas that can only be precisely denoted with further

Nixon et al. 297

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 the acoustic stimulation associated with the TMS trains used in the previous experiments, but did not induce the peripheral somatosensory effects associated with TMSoverprefrontalsites.Thesubjectsweremade aware before the experiment began that they would not be receiving any invasive stimulation.

Data Analysis Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 Error rate and RT were chosen as dependent variables for analysis of the effects of TMS on task performance. Repeated measures t tests were performed on the unprocessed data to show any significant differences between blocks of TMS (or sham TMS) and non-TMS trials for each condition. To examine differences between the two tasks, similar tests were performed on normalized data obtained by subtracting the values for non-TMS trials from those of the TMS trials. In addition, one-sample t tests were used to reveal if any of the changes in ER or RT for each task differed significantly from the baseline (i.e., no change). Finally, a series of single-factor ANOVAs, corrected for multiple compari- sons (Tukey test) were performed on the normalized data to examine differences in accuracy and RT between all three experiments.

Acknowledgments This work was supported by a grant from the Wellcome Trust (UK ). Reprint requests should be sent to P. D. Nixon, Department of Experimental Psychology, University of Oxford, South Parks Road, Oxford OX1 3UD, UK, or via e-mail: philip.nixon@psy. ox.ac.uk.

REFERENCES Figure 6. Identification of the TMS sites used in Experiment 2. (A) The locations (white circles) of the stimulation points in all six subjects for Amunts, K., Schleicher, A., Burgel, U., Mohlberg, H., Uylings, H. TMS applied over posterior Broca’s area (pars opercularis). The points B. M., & Zilles, K. (1999). Broca’s region revisited: were determined by reference to MRI scans of each individual subject Cytoarchitecture and intersubject variability. The Journal of but are plotted here onto a left hemisphere sagittal view (x = À48) Comparative Neurology, 412, 319–341 from a standard brain MRI (MNI template). All the sites lay between Ashbridge, E., Walsh, V., & Cowey, A. (1997). Temporal aspects y = +11 to +21 and z = +6 to +20, thus falling within the higher of visual search studied by transcranial magnetic stimulation. probability contour maps for the pars opercularis defined in Tomaiuolo Neuropsychologia, 35, 1121–1131. et al. (1999). The crosshairs show the position of the anterior Bacon Moore, A., Crosson, B., Gokcay, D., Leonard, C. M., & commissure ( y = 0). (B) Location of stimulation points for TMS Foundas, A. L. (2000). Defining pars triangularis. Society For targeted over the anterior part of Broca’s area (pars triangularis/ Neuroscience Abstracts, 26, 464.15. orbitalis). These points lie between y = +24 and +37, z = À7 and +3. Bartres-Faz, D., Pujol, J., Deus, J., Tormos, J. M., Keenan, J. P., & Pascual-Leone, A. (1999). Identification of the brain areas from which TMS induces speech arrest in normal subjects. reference to cortical cytoarchitecture (Amunts et al., NeuroImage, 9, S1051. 1999). The sites chosen for stimulation in the six sub- Brasil-Neto, J. P., Pascual-Leone, A., Valls-Sole, J., Cohen, L. G., jects are shown in Figure 6. & Hallett, M. (1992). Focal transcranial magnetic stimulation and response bias in a forced-choice task. Journal of Neurology, Neurosurgery and Psychiatry, 55, 964–966. Sham TMS Breier, J. I., Simos, P. G., Zouridakis, G., & Papanicolaou, A. C. (1999). Temporal course of regional brain activation Experiment 3 was run in identical fashion to associated with phonological processing. Journal of Clinical Experiment 2 except that sham TMS stimulation was and Experimental Neuropsychology, 21, 465–476. Chein, J. M., & Fiez, J. A. (2001). Dissociation of verbal working given by orienting the center of the coil 908 from its memory system components using a delayed serial recall normal position tangential to the scalp. This reproduced task. Cerebral Cortex, 11, 1003–1014.

298 Journal of Cognitive Neuroscience Volume 16, Number 2

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 Cohen, J. D., Perlstein, W. M., Braver, T. S., Nystrom, L. E., Noll, visuo-motor information in humans. Experimental Brain D. C., Jonides, J., & Smith. E. E. (1997). Temporal dynamics Research, 118, 435–438. of brain activation during a working memory task. Nature, McConnell, K. A., Nahas, Z., Kozel, F. A., Lorberbaum, J. P., 386, 604–607. Shastri, A., Bohning, D. E., & George, M. S. (2000). The TMS Davachi, L., Maril, A., & Wagner, A. D. (2001). When keeping in motor threshold depends on the distance from coil to mind supports later bringing to mind: Neural markers of underlying cortex: A replication in healthy adults. phonological rehearsal predict subsequent remembering. NeuroImage, 11, S535. Journal of Cognitive Neuroscience, 13, 1059–1070. McDermott, K. B., Petersen, S. E., Watson, J. M., & Ojemann, J. Demonet, J. F., Chollet, F., Ramsay, S., Cardebat, D., G. (2003). A procedure for identifying regions preferentially Nespoulous, J. D, & Wise, R. (1992). The anatomy of activated by attention to semantic and phonological

phonological and semantic processing in normal subjects. relations using functional magnetic resonance imaging. Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 Brain, 115, 1753–1768. Neuropsychologia, 41, 293–303. Devlin, J. T., Matthews, P. M., & Rushworth, M. F. S. (2003). Mottaghy, F. M., Doring, T., Muller-Gartner, H. W., Topper, R., Semantic processing in the left inferior prefrontal cortex: A & Krause, B. J. (2002). Bilateral parieto-frontal network for combined fMRI and TMS study. Journal of Cognitive verbal working memory: An interference approach using Neuroscience, 15, 71–84. repetitive transcranial magnetic stimulation (rTMS). Duvernoy, H. M., Bourgouin, P., Cabanis, E. A., & Cattin, F. European Journal of Neuroscience, 16, 1627–1632. (1999). The human brain: Surface, three-dimensional Mottaghy, F. M., Krause, B. J., Kemna, L. J., Topper, R., sectional anatomy with MRI and blood supply. New York: Tellmann, L., Beu, M., Pascual-Leone, A., & Muller-Gartner, Springer–Verlag. H. W. (2000). Modulation of the neuronal circuitry Epstein, C. M., Lah, J. J., Meador, K., Weissman, J. D., Gaitan, L. subserving working memory in healthy human subjects by E., & Dihenia, B. (1996). Optimum stimulus parameters for repetitive transcranial magnetic stimulation. Neuroscience lateralized suppression of speech with magnetic brain Letters, 280, 167–170. stimulation. Neurology, 47, 1590–1593. Pascual-Leone, A., Tarazona, F., Keenan, J., Tormos, J. M., Epstein, C. M., Meador, K., Loring, D. W., Wright, R. J., Hamilton, R., & Catala, M. D. (1999). Transcranial magnetic Weissman, J. D., Sheppard, S., Lah, J. J., Puhalovich, F., stimulation and neuroplasticity. Neuropsychologia, 37, Gaitan, L., & Davey, K. R. (1999). Localization and 207–217. characterization of speech arrest during transcranial Pascual-Leone, A., Tourmos, J. M., Keenan, J., Tarazona, F., magnetic stimulation. Clinical Neurophysiology, 110, Canete, C., & Catala, M. D. (1998). Study and modulation of 1073–1079. cortical excitability with transcranial magnetic stimulation. Fiez, J. A., & Petersen, S. E. (1998). Neuroimaging studies of Journal of Clinical Neurophysiology, 15, 333–343. word reading. Proceedings of the National Academy of Paulesu, E., Frith, C. D., & Frackowiak, R. S. J. (1993). The Sciences, U.S.A., 95, 914–921. neural correlates of the verbal component of working Fiez, J. A., Raife, E. A., Balota, D. A., Scwarz, J. P., Raichle, M. E., memory. Nature, 362, 342–345. & Petersen, S. E. (1996). A positron emission tomography Paulesu, E., Goldacre, B., Scifo, P., Cappa, S. F., Gilardi, M. C., study of the short-term maintenance of verbal information. Castiglioni, I., Perani, D., & Fazio, F. (1997). Functional Journal of Neuroscience, 16, 808–822. heterogeneity of left inferior frontal cortex revealed by fMRI. Flitman, S. S., Grafman, J., Wasserman, M. D., Cooper, B. A., NeuroReport, 8, 2011–2016. O’Grady, J., Pascual-Leone, A., & Hallett, M. (1998). Paus, T., Jech, R., Thompson, C. J., Comeau, R., Peters, T., & Linguistic processing during repetitive transcranial magnetic Evans, A. C. (1997). Transcranial magnetic stimulation during stimulation. Neurology, 50, 175–181. positron emission tomography: A new method for studying Georgiewa, P., Rzanny, R., Gaser, C., Gerhard, U. J., Vieweg, U., connectivity of the human cerebral cortex. The Journal of Mentzel, H. J., Kaiser, W. A., & Blanz, B. (2002). Phonological Neuroscience, 17, 3178–3184. processing in dyslexic children: A study combining Poldrack, R. A., Wagner, A. D., Prull, M. W., Desmond, J. E., functional imaging and event related potentials. Glover, G. H., & Gabrielli, J. D. (1999). Functional Neuroscience Letters, 318, 5–8. specialization for semantic and phonological processing in Grezes, J., & Decety, J. (2001). Functional anatomy of the left inferior prefrontal cortex. NeuroImage, 10, 15–35. execution, mental simulation, observation, and verb Price, C. J., Indefrey, P., & Turennout, M. (1999). The neural generation of actions: A meta-analysis. Human Brain architecture underlying the processing of written and Mapping, 12, 1–19. spoken word forms. In C. M. Brown & P. Hagoort (Eds.), The Hershenson, M. (1963). Reaction time as a measure of neurocognition of language (pp. 211–240). Oxford: OUP. intersensory facilitation. Journal of Experimental Price, C. J., Wise, R. J. S., Warburton, E. A., Moore, C. J., Psychology, 63, 289–293. Howard, D., Patterson, K., Frackowiak, R. S., & Friston, K. J. Jahanshahi, M., & Rothwell, J. (2000). Transcranial magnetic (1996). Hearing and saying: The functional neuro-anatomy stimulation studies of cognition: An emerging field. of auditory word processing. Brain, 119, 919–931. Experimental Brain Research, 13, 1–9. Raichle, M. E., Fiez, J. A., Videen, T. O., MacLeod, A. K., Pardo, J. Jonides, J., Schumacher, E. H., Smith, E. E., Koeppe, R. A., Awh, V., Fox, P. T., & Petersen, S. E. (1994). Practice related E., Reuter-Lorenz, P. A., Marshuetz, C., &Willis, C. R. (1998). changes in human brain functional anatomy during The role of the parietal cortex in verbal working memory. non-motor learning. Cerebral Cortex, 4, 8–26. The Journal of Neuroscience, 18, 5026–5034. Rumsey, J. M., Horwitz, B., Donohue, B. C., Nace, K., Maisog, J. Maeda, F., Keenan, J., Tormos, J. M., Topka, H., & M., & Andreason, P. (1997). Phonological and orthographic Pascual-Leone, A. (2000). Interindividual variability of the components of word recognition. A PET-rCBF study. Brain, modulatory effects of repetitive transcranial magnetic 120, 739–759. stimulation on cortical excitability. Experimental Brain Rushworth, M. F. S., Ellison, A., & Walsh, V. (2001). Research, 133, 425–430. Complementary localization and lateralization of orienting Marzi, C. A., Miniussi, C., Maravita, A., Bertolasi, L., Zanette, G., and motor attention. Nature Neuroscience, 4, 656–661. Rothwell, J. C., & Sanes, J. N. (1998). Transcranial magnetic Shapiro, K. A., Pascual-Leone, A., Mottaghy, F. M., Gangitano, stimulation selectively impairs interhemispheric transfer of M., & Caramazza, A. (2001). Grammatical distinctions in the

Nixon et al. 299

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021 left frontal cortex. Journal of Cognitive Neuroscience, 13, Topper, R., Mottaghy, F. M., Brugmann, M., Noth, J., & Huber, 713–720. W. (1998). Facilitation of picture naming by focal transcranial Signoret, J. L., Castaigne, P., Lhermitte, F., Abelanet, R., & magnetic stimulation of Wernike’s area. Experimental Brain Lavorel, P. (1984). Rediscovery of Leborgne’s brain: Research, 121, 371–378. Anatomical description with CT scan. Brain & Language, Walsh, V., & Cowey, A. (2000). Transcranial magnetic 22, 303–319. stimulation and cognitive neuroscience. Nature Reviews, 1, Stewart, L., Meyer, B. U., Frith, U., & Rothwell, J. (2001). Left 73–79. posterior BA37 is involved in object recognition: A TMS Warburton, E., Wise, R. J., Price, C. J., Weiller, C., Hadar, U., study. Neuropsychologia, 39, 1–6. Ramsay, S., & Frackowiak, R. S. J. (1996). Studies with Stewart, L., Walsh, V., Frith, U., & Rothwell, J. (2001). TMS positron emission tomography of noun and verb retrieval in

produces two dissociable types of speech disruption. normal subjects. Brain, 119, 159–180. Downloaded from http://mitprc.silverchair.com/jocn/article-pdf/16/2/289/1758045/089892904322984571.pdf by guest on 18 May 2021 NeuroImage, 13, 472–478. Wassermann, E. M. (1998). Risk and safety of repetitive Stewart, L., Walsh, V., & Rothwell, J. (2001). Motor and transcranial magnetic stimulation: Report and suggested phosphene thresholds: A transcranial magnetic stimulation guidelines from the International Workshop on the Safety of correlation study. Neuropsychologia, 39, Repetitive Transcranial Magnetic Stimulation, June 5–7, 415–419. 1996. Electroencephalography and Clinical Tomaiuolo, F., MacDonald, J. D., Caramanos, Z., Posner, G., Neurophysiology, 108, 1–16. Chiavaras, M., Evans, A. C., & Petrides, M. (1999). Willmes, K., & Poeck, K. (1993). To what extent can aphasic Morphology, morphometry and probability mapping of the syndromes be localized? Brain, 116, 1527–1540. pars opercularis of the inferior frontal gyrus: An in vivo MRI Zattore, R. J., Evans, A. C., Meyer, E., & Gjedde, A. (1992). analysis. European Journal of Neuroscience, 11, Lateralisation of phonetic and pitch discrimination in speech 3033–3046. processing. Science, 256, 846–849.

300 Journal of Cognitive Neuroscience Volume 16, Number 2

Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/089892904322984571 by guest on 26 September 2021