Modulation of Motor Excitability During Speech Perception: the Role of Broca’S Area

Modulation of Motor Excitability during Speech Perception: The Role of Broca’s Area

Kate Watkins and Toma´ˇs Paus

Abstract & Studies in both human and nonhuman primates indicate regions that modulate the excitability of the motor system that motor and premotor cortical regions participate in during speech perception. Our results show that during auditory and visual perception of actions. Previous studies, auditory speech perception, there is increased excitability of using transcranial magnetic stimulation (TMS), showed that motor system underlying speech production and that this perceiving visual and auditory speech increased the excitability increase is significantly correlated with activity in the posterior of the orofacial motor system during speech perception. Such part of the left inferior frontal gyrus (Broca’s area). We propose studies, however, cannot tell us which brain regions mediate that this area ‘‘primes’’ the motor system in response to heard this effect. In this study, we used the technique of combining speech even when no speech output is required and, as such, positron emission tomography with TMS to identify the brain operates at the interface of perception and action. &

INTRODUCTION perceptions to phonetic gestures and that such a system The division between the brain regions involved in might be at the origin of the evolution of human speech action production and perception is becoming increas- (Rizzolatti & Arbib, 1998; Gallese et al., 1996). ingly blurred as evidence accumulates that in the In humans, transcranial magnetic stimulation (TMS) primate brain, motor and premotor regions also par- applied over the primary motor cortex has been used ticipate in action perception. In the domain of speech to probe its excitability during visual perception of perception this notion is not new. In fact, Liberman actions (Aziz-Zadeh, Maeda, Zaidel, Mazziotta, & Iacobo- and Mattingly (1985), in their motor theory of speech ni, 2002; Gangitano, Mottaghy, & Pascual-Leone, 2001; perception, first raised the possibility that perception Strafella & Paus, 2000; Fadiga, Fogassi, Pavesi, & Rizzo- of speech involves access to the speech production latti, 1995). In such studies, motor-evoked potentials are system. elicited in a target muscle by stimulating the appropriate In the monkey, a number of studies have identified region of the primary motor cortex. During perception neurons in the ventral premotor cortex that are active of actions that involve the target muscle, motor-evoked during production of actions and visual perception of potentials increase in size, indicating that the motor the same actions (Gallese, Fadiga, Fogassi, & Rizzolatti, system underlying production of the perceived action 1996; DiPellegrino, Fadiga, Fogassi, Gallese, & Rizzolatti, is in a state of increased excitability, or lowered thresh- 1992). These ‘‘mirror neurons’’ may play a role, there- old. Electro- and magnetoencephalography during visual fore, in the recognition, understanding, and imitation of perception of hand movements also reveal changes in actions (Umilta et al., 2001; Rizzolatti, Fadiga, Fogassi, & the primary motor cortex, which are similar to those Gallese, 1999). Recently, neurons in the ventral premo- seen during movement execution (Cochin, Barthelemy, tor cortex of the monkey were found to be active not Roux, & Martineau, 1999; Hari et al., 1998). only when the monkey sees another individual execut- In the speech domain, several studies using methods ing an action but also when it hears the sound associ- similar to those described earlier have examined motor ated with the action (e.g., the sound of paper tearing or cortex changes during visual and auditory perception of a stick dropping; Kohler et al., 2002). These results (Watkins, Strafella, & Paus, 2003; Fadiga, Craighero, indicate that the premotor cortex participates not only Buccino, & Rizzolatti, 2002; Nishitani & Hari, 2002; in visual but also in auditory perception of actions. Sundara, Namasivayam, & Chen, 2001). In one such Furthermore, it has been proposed that a similar system study, we applied TMS over the face area of the primary exists in the human allowing the matching of phonetic motor cortex to elicit a motor-evoked potential in the orbicularis oris muscle of the lips (Watkins et al., 2003). The size of the motor-evoked potential increased during Montreal Neurological Institute, McGill University both auditory and visual speech perception compared to

D 2004 Massachusetts Institute of Technology Journal of Cognitive Neuroscience 16:6, pp. 978–987 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 control conditions. This increase in motor excitability tentials recorded from the lip muscles (see Figure 1 and during speech perception was only evident for stimula- Methods for further details). tion over the left hemisphere and not for stimulation over the right hemisphere; this pattern is consistent with the RESULTS known specialization of the left hemisphere for speech. In a control experiment, we demonstrated that the size of Eight subjects were scanned using PET. In each subject, the motor-evoked potentials elicited in a muscle of the three scans were obtained for each of the four condi- right hand did not differ among these conditions, sug- tions, with the exception of one subject in whom one gesting that speech-perception-related changes in excit- Speech scan was rejected because the coil moved. ability are specific to the muscles involved in speech Twenty pulses of TMS were applied during each scan. production. In a similar study, Fadiga et al. (2002) showed Motor-evoked potentials were recorded in response to increased motor-evoked potentials recorded from the the TMS pulses using continuous electromyography. tongue when subjects listened to speech sounds that The average sizes of motor-evoked potentials, elicited require movement of the tongue in their production. by TMS, and recorded from the orbicularis oris muscle in The aforementioned TMS studies demonstrate that the Speech, Lips, and Eyes conditions were expressed as perception influences the state of the motor system percentages of those recorded in the Control condition. involved in production. These studies, however, cannot Within-subjects analysis of variance revealed a significant identify the brain region or regions that mediate such effect of condition, F(2,14) = 9.31, p = .003, which was changes in the motor system. In the present study, we due to significantly greater motor-evoked potentials for combined TMS with positron emission tomography the Speech condition relative to the Eyes, t(7) = 5.02, (PET) to identify the brain regions mediating the p = .005, and to the lips, t(7) = 3.55, p = .028 con- changes in motor excitability during perception of ditions (see Figure 2). The two visual conditions did not speech. On the basis of findings in monkey premotor differ significantly. These results confirm that listening to cortex and the homology between this region and the speech increases the excitability of the motor system posterior part of the human inferior frontal gyrus, underlying speech production. In this study, no signifi- known as Broca’s area, we hypothesized that the latter cant change in motor excitability was seen during visual would be one region involved in the modulation of perception of speech-related lip movements. motor excitability during speech perception. Comparison of the PET scans obtained in the Speech, We applied TMS over the face area of the left primary Lips, and Eyes conditions with those obtained in the motor cortex to measure motor excitability during (1) control condition revealed, respectively, the brain regions listening to speech (Speech condition), (2) viewing of involved in listening to speech, viewing speech-related speech-related lip movements (Lips condition), (3) view- lip movements or viewing eye-and-brow movements. ing of eye and brow movements (Eyes condition), and As expected, listening to speech was associated with (4) listening to and viewing noise (Control condition). At strong activation of the superior temporal gyrus bilater- the same time, we scanned the brain using PET to ally, which extended along its length to the uncus in identify regions in which changes in activity, as indexed the anterior temporal lobe (see Figure 3 and Table 1). by cerebral blood flow, correlated with changes in Viewing movements of the face, either lip or eye-and- excitability, measured by the size of motor-evoked po- brow movements, was associated with activation in the

Figure 1. Experimental design. Schematic shows examples of the visual and auditory stimuli used in each of the four conditions. PET data were acquired over 60 sec during the application of 20 magnetic pulses to the primary motor cortex face area.

Watkins and Paus 979 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 whether the slope of the regression line at each voxel was significantly different from zero. For the Speech conditions, significant positive correlations were seen in the opercular region of the left inferior frontal gyrus (area 44/45), the left putamen, a very medial portion of the left parietal operculum, and the left cerebellum (lobules IV, V, and VI) (Table 2 and Figure 4). Significant negative correlations were seen in the left supramarginal gyrus (in a very lateral portion of the parietal operculum) and the area of the right precentral gyrus that corresponds functionally to the primary motor face area (Fox et al., 2001). The mean voxel values from a 6-mm Figure 2. Motor excitability changes. The graph shows the mean of sphere centered at each peak location were extracted the change in size of the motor-evoked potential for the Speech, Lips, from the CBF images and entered into a stepwise and Eyes conditions relative to the Control condition. Error bars multiple regression analyses with the size of the motor- represent standard errors of the mean. evoked potential as the dependent variable. The only significant predictor of the size of the motor-evoked potential was the voxel value at the peak in the left inferior occipitotemporal regions bilaterally (see Figure 3 inferior frontal gyrus, R2 = .26; F(1,19) = 7.51; p = .012. and Table 1). In the conditions involving either auditory Figure 4 shows the significant linear relationship be- or visual speech perception, namely, the Speech and Lips tween the size of the motor-evoked potential and the conditions, there was additional activation in the left normalized CBF data at that location. inferior frontal gyrus. Even though the motor-evoked potential data showed The regions where changes in blood flow were related no significant increase in excitability of the motor system to changes in motor excitability were revealed by regres- for the other conditions (see above), similar regression sion analyses between the normalized motor-evoked analyses were carried out with the data obtained in each potential sizes and the normalized CBF images obtained condition. The only significant correlation (p <.05, during each scan. The resulting t-statistic map indicated corrected) in these analyses was a negative one between

Figure 3. Results of the PET subtractions. Maps reflecting significant increases in CBF in the Speech (red), Lips (green), and Eyes (blue) conditions relative to the Control condition are superimposed on transverse slices through a template image (left side) and a semiopaque surface- rendered image (right side). The maps were thresholded at t > 3.5. In the upper right corner of each slice is the Z coordinate in Talairach space. L = left hemisphere.

980 Journal of Cognitive Neuroscience Volume 16, Number 6 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 Table 1. Peak Coordinates of CBF Differences in PET Broca’s Area: Structural and Functional Anatomy Subtractions In addition to its well-known participation in speech and Region x y z t language functions (Mohr, 1976; Rasmussen & Milner, Speech minus Control 1975; Broca, 1861), Broca’s area is a good candidate for a region that modulates motor cortex excitability during 1 L inferior frontal gyrus (BA 45) À52 22 20 4.02 action perception; the role of Broca’s area and the L uncus À24 À6 À18 7.84 ventral premotor cortex in perception of action has been demonstrated in both human and nonhuman L superior temporal gyrus À56 À14 6 16.90 primates (Kohler et al., 2002; Buccino et al., 2001; R uncus 20 À7 À15 4.84 Decety et al., 1997; Gallese et al., 1996; Grafton, Arbib, R superior temporal gyrus 58 À13 3 16.67 Fadiga, & Rizzolatti, 1996; Rizzolatti et al., 1996; DiPelle- grino et al., 1992). Similarly, this region is activated during action imitation, particularly by goal-directed Lips minus Control actions (Koski, Iacoboni, Dubeau, Woods, & Mazziotta, L inferior frontal gyrus (BA 45/47) À42 27 À2 3.771 2003; Koski et al., 2002; Iacoboni et al., 1999). Also, the connectivity of this region with sensory cortices in the L inferior occipital gyrus À41 À76 À8 7.60 temporal and parietal lobes is known (see Petrides & L middle occipital gyrus (superior) À29 À90 21 5.52 Pandya, 2002a, 2002b). The exact homology between the cortical areas within R fusiform gyrus 40 À71 À12 8.07 the human inferior frontal gyrus and those in the monkey is unclear, however (Aboitiz & Garcia, 1997; Eyes minus Control Petrides & Pandya, 1994; Barbas & Pandya, 1987). Human area 44, the most caudal part of Broca’s area, L inferior occipital gyrus À40 À80 À5 9.73 is located on the pars opercularis, and seems to be the R fusiform gyrus 38 À44 À15 5.89 most likely homologue of monkey ventral premotor 38 À76 À11 9.99 cortex (area 6) (Rizzolatti & Arbib, 1998; Petrides & Pandya, 1994). The boundary between area 44 and the R intraparietal sulcus 29 À85 21 4.84 more rostrally located area 45 lies within the ascending Results are reported for peaks with t > 4.5 ( p < .05 corrected). ramus of the lateral fissure (Amunts et al., 1999). The morphology of this region is very variable, however 1Peaks with t > 3.0 ( p < .001, uncorrected) in the left inferior frontal region. L = left; R = right. (Tomaiuolo et al., 1999). In our study, the stereotaxic coordinates (À39, 18, 12) of the peak voxel within

the size of the motor-evoked potential and the CBF Table 2. Peak Coordinates for CBF Correlations with Motor values in the right precentral gyrus (Talairach coordi- Excitability during Auditory Speech Perception nates: 28, À21, 60) for the Lips condition. Region x y z t DISCUSSION Positive correlations The results of this study show that during auditory L inferior frontal gyrus (BA 44/45) À39 18 12 3.52 perception of speech the increased size of the motor- L putamen À19 À4 9 3.21 evoked potential obtained by stimulation over the face L parietal operculum (very medial) À39 À31 23 3.16 area of the primary motor cortex correlated with cerebral blood flow in the posterior part of the left inferior L cerebellum frontal gyrus, namely, Broca’s area. We suggest, there- Lobule V/IV À7 À54 À9 3.86 fore, that the excitability of the motor system during Lobule VI À16 À69 À15 3.99 auditory speech perception is related to an increase in activity in Broca’s area and that these data support the idea that Broca’s area plays a role in modulating motor Negative correlations cortex activity during perception of speech. Even though activity in a number of regions correlated with motor L supramarginal gyrus À63 À25 20 À3.13 (lateral parietal operculum) excitability during auditory speech perception, the results of a multiple regression analysis indicated that the R precentral gyrus (face area) 40 À642À3.48 activity in Broca’s area was the best predictor of motor Results are reported for peaks with t > 4.5 ( p < .05 corrected), except excitability. The following discussion, therefore, focuses for peaks with t > 3.0 ( p < .001, uncorrected) in motor- or language- on the Broca’s region and, in particular, its connectivity. related regions. L = left; R = right.

Watkins and Paus 981 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 Figure 4. Results of the regression analysis between rCBF and motor excitability during auditory speech perception. Maps reflecting a significant positive relationship between CBF and motor excitability are superimposed on sagittal and coronal slices through the peak in the left inferior frontal gyrus. The maps were thresholded at t > 2.5. In the upper right corner of each slice is the x (sagittal slice) or y coordinate in Talairach space. The graph shows the significant linear relationship between CBF values extracted from a sphere (6-mm radius) centered at this peak and the average size of the motor-evoked potential during auditory speech conditions in each subject. Note: CBF data for the Speech condition were not available for one of the eight subjects due to coil movement.

Broca’s area that was identified by regression analyses production is modulated by activity of neurons in area between motor excitability during auditory speech per- 44, which receive projections from neurons in the ception and brain activity are located between those of parietal operculum, which in turn receives projections the centers of gravity of the cytoarchitectural maps for from primary, secondary, and association auditory areas areas 44 (À43, 12, 20) and 45 (À43, 26, 17) (Binkofski in the posterior superior temporal cortex. Area 44 is in et al., 2000). It is therefore not possible to determine the a position to modulate the motor system either via relative contribution of neural activity in areas 44 and 45 connections within the ventral premotor cortex or di- to the observed blood-flow changes in this region. rectly. This model is in accord with one derived from Examination of the respective patterns of connectivity magnetoencephalography findings during visual percep- of areas 44 and 45, however, suggests that it is activity tion and execution of verbal and nonverbal lip forms in area 44 that most likely modulates excitability of the (Nishitani & Hari, 2002). From the relative timing of motor system during auditory speech perception. Area areas of activation during visual observation of lip forms, 44 and the ventral premotor cortex (area 6) receive these authors showed that activity progressed from the inputs from the supramarginal gyrus, and adjacent pari- occipital cortex to the superior temporal region, the etal operculum (Petrides & Pandya, 2002a) and the inferior parietal lobule, Broca’s area, and the primary posterior insula (Mesulam & Mufson, 1982), among motor cortex. Our data suggest that a similar network is other regions. Interestingly, in our regression analyses, engaged during auditory speech perception. activity in a region of the parietal operculum, which was located quite medially, showed a significant positive PET Subtraction Analyses relationship with motor excitability. Wise et al. (2001) have reanalyzed a number of imaging studies of speech For the PET data, the subtraction of the control condi- processing and concluded that the medial left tempor- tion from each of the other conditions revealed activity oparietal junction ‘‘acts as an interface between posteri- in areas consistent with perception of the different or temporal cortex and motor cortex for speech.’’ In stimuli. For auditory speech conditions, the superior contrast to area 44, the more rostrally located area 45 in temporal cortex, where primary and secondary auditory the monkey receives strong inputs directly from the areas are located, was active bilaterally and extensively. auditory-related superior temporal gyrus and the multi- Furthermore, there was activity in the left inferior frontal modal areas of the superior temporal sulcus (Petrides & gyrus. This is consistent with the results of previous Pandya, 2002b; Deacon, 1992) and secondary auditory studies reporting activity in this region for stimuli that cortex (Romanski et al., 1999). We cannot rule out the required semantic or thematic processing of receptive possibility, therefore, that these direct projections addi- speech (Newman, Just, Keller, Roth, & Carpenter, 2003; tionally contributed to the activity in Broca’s area, which Bookheimer, 2002; Dapretto & Bookheimer, 1999; correlated with motor excitability in our study. In hu- Stromswold, Caplan, Alpert, & Rauch, 1996); the stimuli mans the projections from areas 44 and 45 to primary used in our study (continuous prose passages or stories) motor cortex are not known. In the monkey, however, would certainly involve semantic and thematic pro- ventral premotor cortex (area 6), the most likely homo- cessing. For the visual speech conditions, a very infe- logue of human area 44, projects to the precentral gyrus rior and anterior portion of the left inferior frontal (area 4) (Barbas & Pandya, 1987). gyrus was active. This is consistent with the results of Based on these considerations, we propose that the several previous studies in hearing subjects (Paulesu excitability of the motor system underlying speech et al., 2003; Calvert & Campbell, 2002; Campbell et al.,

982 Journal of Cognitive Neuroscience Volume 16, Number 6 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 2001) reporting activation in this area during silent are required, however, to determine whether this area speech reading. During the conditions in which subjects is necessary for the modulation of motor excitability, and viewed either speech-related lip movements or move- whether the posterior left inferior frontal gyrus and the ments of the eyes and brow, activity was seen in the motor system underlying speech production are ne- ventral temporal and occipital regions, which in previous cessary for speech perception. Also, questions remain studies have been activated specifically by processing of about the existence and location of different neuronal face stimuli (Puce, Allison, Bentin, Gore, & McCarthy, populations within the inferior frontal gyrus that might 1998; Kanwisher, McDermott, & Chun, 1997). commonly represent both perception and production of actions involving different body parts. For example, does one portion of the inferior frontal gyrus modulate the Changes in Motor Excitability during activity of different motor-output systems or can sub- Speech Perception divisions within this region be made? The increased excitability of the speech production system during auditory speech perception seen in this study replicates the findings from two previous TMS METHODS studies (Watkins et al., 2003; Fadiga et al., 2002). We did not, however, find significant increases in motor excit- Subjects ability during visual perception of speech-related lip Eight right-handed, English-speaking subjects (5 women, movements as previously reported using TMS (Watkins 3 men; age range = 19.75–30.17 years; average age = et al., 2003) and magnetoencephalography (Nishitani, & 24.43 years) participated in this study. They were medi- Hari, 2002). This may have been due to an unforeseen cation free and had no personal or family history of fourfold increase in the size of our visual stimulus in the seizures or other neurological disorder. All subjects gave environment of the PET scanner compared to the TMS informed consent to their participation in this study, laboratory, which was due to positioning the monitor which was approved by the Research Ethics Board of the directly above and in front of the subject so that it could Montreal Neurological Hospital/Institute. be viewed comfortably from the gantry of the scanner. It is also worth noting that in our previous study, the Experimental Setup increase in excitability during visual speech perception was smaller and more variable across subjects than TMS was applied over the face area of the primary during auditory perception. The increased excitability motor cortex to measure the excitability of the motor of the motor system during auditory speech perception system underlying speech production. At the same is consistent with the notion of a shared representation time, regional cerebral blood flow (rCBF) was mea- for perception and production of speech and with the sured with the 15O-labeled water-bolus PET method. motor theory of speech production (Liberman & Mat- Subjects passively listened to or viewed stimuli in the tingly, 1985). The increased motor excitability seen in following four conditions: this study and in our previous one (Watkins et al., 2003) Speech condition: Subjects listened to continuous may reflect an internal repetition of the perceived speech while viewing visual noise. speech by the speech production system that does not Lips condition: Subjects viewed speech-related lip result in overt movement of the articulators and pro- movements while listening to white noise. ceeds, therefore, at a much faster rate than overt speech. Eyes condition: Subjects viewed eye-and-brow move- This internal speech might improve the listener’s ability ments while listening to white noise. to understand and even anticipate the heard speech. Control condition: Subjects viewed visual noise and listened to white noise. Concluding Remarks The continuous speech stimulus was a digital recording The results of our combined PET and TMS study indicate of a female speaker reading continuous prose. The lip that the excitability of the motor system underlying movements performed by this speaker while reading speech production is modulated by activity in the pos- were recorded and comprised the visual stimulus for the terior part of the left inferior frontal gyrus, most likely Lips condition. For all conditions, subjects were told to area 44, during auditory speech perception. This finding give their full attention to the visual and auditory stimuli is also consistent with the idea that an auditory equiv- and that they would be questioned about the stimuli at alent of the ‘‘mirror-neuron’’ system reported in the the end of the experiment. Debriefing following the monkey also exists in humans (Rizzolatti, Fogassi, & experiment did not reveal any systematic attentional Gallese, 2001). This region within the left inferior frontal bias among the different conditions. Auditory stimuli gyrus appears to ‘‘prime’’ the motor system underlying were played through insert earphones at a comfortable speech production in response to speech perception listening level (~80-dB SPL). Visual stimuli were pre- even when no production is required. Further studies sented on a large monitor at a distance of approximately

Watkins and Paus 983 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 1 m from the subject’s face. The order of the four lowest motor threshold was obtained while the muscle conditions (ABCD) within a block was randomized and was contracted. The active motor threshold was deter- three blocks of different random orders were used for mined as the intensity at which TMS elicited at least 5 each subject. The order within the first block was out of 10 motor-evoked potentials with at least 50-AV reversed for the last block and the middle block was amplitudes. The intensity of the stimulator was then set selected so that no two successive conditions were the to 120% of this active motor threshold for the stimula- same. A different random order was used for each tions applied during the PET scan. At the time of each subject. In each subject, three 60-sec PET scans were scan, 20 pulses of TMS were applied at intervals of obtained for each of the conditions above (12 scans per between 5 and 8 sec over a 130-sec period; these subject). At the same time, 20 pulses of magnetic stimulation parameters are extremely unlikely to result stimulation were applied at intervals of 5–8 sec and in any TMS-induced effects on motor cortex excitability. the resulting motor-evoked potential was recorded The average active motor threshold (percentage of using electromyography (see Figure 1). maximum stimulator output) for stimulation over the face area of left primary motor cortex was 47.88% (range 30–58%). Electromyography For each subject, surface electrodes (Ag/AgCl, 10-mm Positron Emission Tomography diameter) were attached to the orbicularis oris muscle to record electromyographic (EMG) activity. This muscle PET scans were obtained with a CTI/Siemens (Erlan- is rarely at rest, so in order to stabilize baseline EMG gen, Germany) HR + 63-slice tomograph scanner op- activity during the study, subjects were trained for erated in a three-dimensional acquisition mode. To approximately 10 min to produce a constant level of protect the photomultipliers in the PET detectors from contraction of the lip muscles by pursing them while the effects of the coil-generated magnetic field, a well- receiving visual feedback indicating the amount of EMG grounded cylindrical insert consisting of four layers of activity. They were trained until a satisfactory constant 0.5-mm-thick mu metal (a nickel–iron alloy that is very level of contraction of between 20% and 30% of their efficient for screening magnetic fields), whose outer maximum voluntary contraction was obtained. During diameter matched the inner diameter of the scanner’s the study, subjects were asked to produce this level of patient port, was used as a shield (Paus et al., 1997). contraction while being stimulated. Contraction of the With the subject in the scanner, a bite-bar attached to lip muscles lowered the motor threshold, thereby allow- the head holder was customized and locked in place. ing lower levels of stimulation to be used compared with This helped to minimize head movement during the those for the resting muscle. Continuous electromyo- scanning session. The TMS coil was positioned over the graphy recordings were acquired during each scan with area where the lowest motor threshold was obtained a 10-channel TMS-compatible system (Virtanen, Ruo- (see above). Once the subject, the bite-bar, and the coil honen, Naatanen, & Ilmoniemi, 1999), which uses a assembly were in place, a 10-min transmission scan was sample-and-hold circuit that pins the amplifier output performed. This scan was used later to correct the to a constant level for 2.5 msec, starting 100 Asec before emission scans for the attenuation of gamma rays due the pulse, with about 3-msec recovery time. The ampli- to all the objects in the scanner. For each emission scan, fier’s bandwidth was 0.1 to 500 Hz and the signal was while the subject listened and viewed the stimuli rele- sampled at 1.45 kHz. vant to the condition, 10 mCi of 15O-labeled water were injected into the left antecubital vein. At the same time, a sequence of 20 TMS pulses were applied and EMG was Transcranial Magnetic Stimulation recorded from the orbicularis oris muscle. During the Magnetic stimuli were generated by a Magstim 200 unit first scan in one subject, it became apparent that the coil and delivered by a figure-eight coil (each wing 3.5-in. had moved relative to the subject’s head. The coil was diameter; Magstim, Dyfed, UK) connected through a repositioned before continuing the experiment and a BiStim module. The coil was placed tangential to the transmission scan was obtained at the end of the skull, such that the induced current flowed from poste- scanning period. The data from the first scan for this rior to anterior under the junction of the two wings of subject were rejected. the figure-eight coil. We located the face area of the primary motor cortex by first locating the hand area Magnetic Resonance Imaging (where application of TMS elicited a muscle twitch in the contralateral hand) and then moving the coil ventrally On a separate day, a magnetic resonance imaging (MRI) and slightly anteriorly until we observed a motor-evoked scan of the subject’s brain was acquired using a Siemens potential in the contralateral orbicularis oris muscle with Vision 1.5-T system. A high-resolution T1-weighted scan a latency of ~10 msec. The coil was held in a fixed was obtained (3-D gradient-echo scan with 160–170 position by a mechanical arm over the area where the sagittal slices, 1-mm isotropic resolution, TR = 22 msec,

984 Journal of Cognitive Neuroscience Volume 16, Number 6 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 TE = 10 msec, flip angle = 308), coregistered with the 1994). To assess the significance of rCBF differences PET scans (Woods, Mazziotta, & Cherry, 1993) and used between conditions, subtraction maps were generated. to provide a transformation of each subject’s brain into A t statistic was calculated at each voxel by dividing the standardized stereotaxic space via an automatic feature- mean CBF difference by its standard deviation pooled detection algorithm (Collins, Neelin, Peters, & Evans, across all brain voxels (Worsley, Evans, Marrett, & Nee- 1994). lin, 1992). In addition, a regional regression map was generated for each condition that assessed the significance of the relationship between the normalized motor- Data Analysis evoked potential data and rCBF. These maps revealed brain regions where changes in blood flow correlated TMS/EMG Data with changes in motor excitability. For the Lips, Eyes, and Rectified EMG sweeps starting 100 msec before the TMS Control conditions, the data consisted of 24 normalized pulse and ending 200 msec after were analyzed individ- CBF maps obtained in eight subjects, scanned three ually for each scan in each subject using Matlab 5.2 (The times each; for the Speech condition, the data consisted Mathworks Inc., Natick, MA). The size (area under the of 21 maps obtained in seven subjects. The relationship curve of the rectified EMG signal, mV Â msec) of single between normalized motor-evoked potential size and motor-evoked potentials and the average height of CBF was assessed by means of an analysis of covariance, the preceding baseline EMG activity (100 msec before with subjects as a main effect and motor-evoked poten- the TMS pulse) were measured off-line for each of the tial size as the covariate. The subject effect was removed 20 pulses per scan, per subject. Because the subjects and the parameter of interest was the slope of the effect were contracting the lip muscles, the amount of base- of motor-evoked potential size on normalized CBF. A line EMG activity was linearly related to motor-evoked t-statistic map was generated that tested whether, at a potential size. Even though our subjects were trained to given voxel, the slope of the regression was significantly maintain a contraction of the muscles during stimula- different from zero. The presence of a significant peak tion, we used analysis of covariance to adjust the mean was evaluated in both the subtraction and regression t motor-evoked potential size for the corresponding base- maps by a method based on three-dimensional Gaussian line EMG activity in each condition and in each subject. random-field theory, which corrects for the multiple Repeated measures analyses of variance indicated no comparisons involved in searching across a volume significant difference in baseline EMG activity among (Worsley et al., 1992). Values equal to, or exceeding, a conditions. To assess the significance of differences in criterion of t = 4.5 were considered as significant (p < motor excitability among conditions, the mean motor- .05, corrected). Values greater than t = 3.0 were also evoked potential size for the Speech, Lips, and Eyes considered as significant (p < .001, uncorrected) but conditions were expressed as percentages of the mean only for regions predicted in advance. In the subtraction for the Control condition. Within-subjects analysis of maps the left inferior frontal region was predicted to variance was used to analyze these data. To assess the show activation in the speech and lips conditions; in the relationship between motor excitability and normalized regression analyses, CBF values in motor- and language- rCBF, the size of each motor-evoked potential was related cortices were predicted to be related to motor normalized using the mean and standard deviation excitability. across all conditions for each subject separately and equated to a distribution with an arbitrary mean of 100 and standard deviation of 15. The mean normalized Acknowledgments motor-evoked potential size was calculated for each scan We thank the staff at the McConnell Brain Imaging Centre for in each subject (three per condition per subject) and their assistance with this study. We also thank Drs. Marie- entered into regression analyses with the PET data (see He´le`ne Grosbras, Mortimer Mishkin, and Michael Petrides for below). In one subject, only two mean MEP sizes were useful discussion concerning this manuscript. The Canadian obtained in the Speech condition because of coil move- Institutes for Health Research and the Canadian Foundation ment; these data were not included in the regression for Innovation supported this research. analyses.

REFERENCES PET/TMS Data Aboitiz, F., & Garcia, R. (1997). The anatomy of language The CBF images were reconstructed with a 14-mm revisited. Biological Research, 30, 171–183. Hanning filter, normalized for differences in global Amunts, K., Schleicher, A., Burgel, U., Mohlberg, H., Uylings, CBF, coregistered with the individual MRIs (Woods H. B., & Zilles, K. (1999). Broca’s region revisited: Cytoarchitecture and intersubject variability. Journal of et al., 1993) and transformed into standardized stereo- Comparative Neurology, 412, 319–341. taxic space (Talairach & Tournoux, 1988) by means of an Aziz-Zadeh, L., Maeda, F., Zaidel, E., Mazziotta, J., & Iacoboni, automated feature-matching algorithm (Collins et al., M. (2002). Lateralization in motor facilitation during action

Watkins and Paus 985 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 observation: A TMS study. Experimental Brain Research, Gallese, V., Fadiga, L., Fogassi, L., & Rizzolatti, G. (1996). 144, 127–131. Action recognition in the premotor cortex. Brain, 119, Barbas, H. & Pandya, D. N. (1987). Architecture and frontal 593–609. cortical connections of the premotor cortex (area 6) in the Gangitano, M., Mottaghy, F. M., & Pasucal-Leone, A. (2001). rhesus monkey. Journal of Comparative Neurology, 256, Phase-specific modulation of cortical motor output during 211–228. movement observation. NeuroReport, 12, 1489–1492. Binkofski, F., Amunts, K., Stephan, K. M., Posse, S., Schormann, Grafton, S. T., Arbib, M. A., Fadiga, L., & Rizzolatti, G. (1996). T., Freund, H. J., Zilles, K., & Seitz, R. J. (2000). Broca’s Localization of grasp representations in humans by PET: 2. region subserves imagery of motion: A combined Observation compared with imagination. Experimental cytoarchitectonic and fMRI study. Human Brain Mapping, Brain Research, 112, 103–111. 11, 273–285. Hari, R., Forss, N., Avikainen, S., Kirveskari, E., Salenius, S., Bookheimer, S. (2002). Functional MRI of language: New & Rizzolatti, G. (1998). Action of human primary motor approaches to understanding the cortical organization of cortex during action observation: A neuromagnetic study. semantic processing. Annual Review of Neuroscience, 25, Proceedings of the National Academy of Sciences, U.S.A., 151–188. 95, 15061–15065. Broca, P. (1861). Remarques sur le sie`ge de la faculte´ Iacoboni, M., Woods, R. P., Brass, M., Bekkering, H., Mazziotta, du langage articule´; suivies d’une observation J. C., & Rizzolatti, G. (1999). Cortical mechanisms of human d’ aphemie. Bulletin de la Socie´te´ d’ Anatomie de Paris, imitation. Science, 286, 2526–2528. 6, 330–357. Kanwisher, N., McDermott, J., & Chun, M. M. (1997). The Buccino, G., Binkofski, F., Fink, G. R., Fadiga, L., Fogassi, L., fusiform face area: A module in human extrastriate cortex Gallese, V., Seitz, R. J., Zilles, K., Rizzolatti, G., & Freund, H. J. specialized for face perception. Journal of Neuroscience, (2001). Action observation activates premotor and 17, 4302–4311. parietal areas in a somatotopic manner: An fMRI study. Kohler, E., Keysers, C., Umilta, M. A., Fogassi, L., Gallese, V., European Journal of Neuroscience, 13, 400–404. & Rizzolatti, G. (2002). Hearing sounds, understanding Calvert, G. A., & Campbell, R. (2002). Reading speech from still actions: Action representation in mirror neurons. Science, and moving faces: The neural substrates of visible speech. 297, 846–848. Journal of Cognitive Neuroscience, 15, 57–70. Koski, L., Iacoboni, M., Dubeau, M. C., Woods, R. P., & Campbell, R., MacSweeney, M., Surguladze, S., Calvert, G., Mazziotta, J. C. (2003). Modulation of cortical activity during McGuire, P., Suckling, J., Brammer, M. J., & David, A. S. different imitative behaviors. Journal of Neurophysiology, (2001). Cortical substrates for the perception of face actions: 89, 460–471. An fMRI study of the specificity of activation for seen speech Koski, L., Wohlschlager, A., Bekkering, H., Woods, R. P., and for meaningless lower-face acts (gurning). Cognitive Dubeau, M. C., Mazziotta, J. C., & Iacoboni, M. (2002). Brain Research, 12, 233–243. Modulation of motor and premotor activity during imitation Cochin, S., Barthelemy, C., Roux, S., & Martineau, J. (1999). of target-directed actions. Cerebral Cortex, 12, 847–855. Observation and execution of movement: Similarities Liberman, A., & Mattingly, I. G. (1985). The motor theory of demonstrated by quantified electroencephalography. speech perception—revised. Cognition, 21, 1–36. European Journal of Neuroscience, 11, 1839–1842. Mesulam, M.-M., & Mufson, E. J. (1982). Insula of the Old Collins, D. L., Neelin, P., Peters, T. M., & Evans, A. C. (1994). World monkey: III. Efferent cortical output and comments Automatic 3D intersubject registration of MR volumetric on function. Journal of Comparative Neurology, 212, data in standardized Talairach space. Journal of 38–52. Computer-Assisted Tomography, 18, 192–205. Mohr, J. P. (1976). Broca’s area and Broca’s aphasia. In M. Dapretto, M., & Bookheimer, S. Y. (1999). Form and content: Whitaker & H. Whitaker (Eds.), Studies in neurolinguistics Dissociating syntax and semantics in sentence (vol. 1, pp. 201–233). New York: Academic Press. comprehension. Neuron, 24, 427–432. Newman, S. D., Just, M. A., Keller, T. A., Roth, J., & Carpenter, Deacon, T. W. (1992). Cortical connections of the inferior P. A. (2003). Differential effects of syntactic and semantic arcuate sulcus cortex in the macaque monkey brain. Brain processing on the subregions of Broca’s area. Cognitive Research, 573, 8–26. Brain Research, 16, 297–307. Decety, J., Grezes, J., Costes, N., Perani, D., Jeannerod, M., Nishitani, N., & Hari, R. (2002). Viewing lip forms: Cortical Procyk, E., Grassi, F., & Fazio, F. (1997). Brain activity during dynamics. Neuron, 36, 1211–1220. observation of actions. Influence of action content and Paulesu, E., Perani, D., Blasi, V., Silani, G., Borghese, A. N., subject’s strategy. Brain, 120, 1763–1777. De Giovanni, U., Sensolo, S., & Fazio, F. (2003). A DiPellegrino, G., Fadiga, L., Fogassi, L., Gallese, V., & functional-anatomical model for lip-reading. Journal of Rizzolatti, G. (1992). Understanding motor events: A Neurophysiology, 30, 2005–2013. neurophysiological study. Experimental Brain Research, Paus, T., Jech, R., Thompson, C. J., Comeau, R., Peters, T., 91, 176–180. & Evans, A. C. (1997). Transcranial magnetic stimulation Fadiga, L., Craighero, L., Buccino, G., & Rizzolatti, G. (2002). during positron emission tomography: A new method for Speech listening specifically modulates the excitability of studying connectivity of the human cerebral cortex. tongue muscles: A TMS study. European Journal of Journal of Neuroscience, 17, 3178–3184. Neuroscience, 15, 399–402. Petrides, M., & Pandya, D. N. (2002a). Association pathways of Fadiga, L., Fogassi, L., Pavesi, G., & Rizzolatti, G. (1995). the prefrontal cortex and functional observations. In D. T. Motor facilitation during action observation: A magnetic Stuss & R. T. Knight (Eds.), Principles of Frontal Lobe stimulation study. Journal of Neurophysiology, 73, Function (pp. 31–50). New York: Oxford University Press. 2608–2611. Petrides, M., & Pandya, D. N. (2002b). Comparative Fox, P. T., Huang, A., Parsons, L. M., Xiong, J. H., Zamarippa, cytoarchitectonic analysis of the human and the macaque F., Rainey, L., & Lancaster, J. L. (2001). Location-probability ventrolateral prefrontal cortex and corticocortical profiles for the mouth region of human primary connection patterns in the monkey. European Journal motor-sensory cortex: Model and validation. Neuroimage, of Neuroscience, 16, 291–310. 13, 196–209. Petrides, M., & Pandya, D. N. (1994). Comparative architectonic

986 Journal of Cognitive Neuroscience Volume 16, Number 6 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021 analysis of the human and the macaque frontal cortex. Sundara, M., Namasivayam, A. K., & Chen, R. (2001). In F. Boller & J. Grafman (Eds.), Handbook of Observation-execution matching system for speech: A Neuropsychology: Vol. 9 (pp. 17–58). Amsterdam: Elsevier. magnetic stimulation study. NeuroReport, 12, 1341–1344. Puce, A., Allison, T., Bentin, S., Gore, J. C., & McCarthy, G. Talairach, J., & Tournoux, P. (1988). Co-planar Stereotaxic (1998). Temporal cortex activation in humans viewing eye Atlas of the Human Brain: 3-Dimensional Proportional and mouth movements. Journal of Neuroscience, 18, System: An Approach to Cerebral Imaging. Stuttgart, 2188–2199. Germany: Thieme. Rasmussen, T., & Milner, B. (1975). Clinical and surgical Tomaiuolo, F., MacDonald J. D., Caramanos, Z., Posner, G., studies of the cerebral speech areas in man. In K. J. Zulch, Chiavaras, M., Evans, A. C., & Petrides, M. (1999). O. Creutzfeldt, & G. C. Galbraith (Eds.), Cerebral Morphology, morphometry and probability mapping of Localization (pp. 238–257). Berlin: Springer-Verlag. the pars opercularis of the inferior frontal gyrus: An Rizzolatti, G., & Arbib, M. A. (1998). Language within our grasp. in vivo MRI analysis. European Journal of Neuroscience, 11, Trends in Neurosciences, 21, 188–194. 3033–3046. Rizzolatti, G., Fadiga, L., Fogassi, L, & Gallese, V. (1999). Umilta, M. A., Kohler, E., Gallese, V., Fogassi, L., Fadiga, L., Resonance behaviors and mirror neurons. Archives of Keysers, C., & Rizzolatti, G. (2001). I know what you Italian Biology, 137, 85–100. are doing. A neurophysiological study. Neuron, 31, 155–165. Rizzolatti, G., Fadiga, L., Matelli, M., Bettinardi, V., Paulesu, Virtanen, J., Ruohonen, J., Naatanen, R., & Ilmoniemi, R. J. E., Perani, D., & Fazio, F. (1996). Localization of grasp (1999). Instrumentation for the measurement of electric representations in humans by PET: 1. Observation versus brain responses to transcranial magnetic stimulation. execution. Experimental Brain Research, 111, 246–252. Medical and Biological Engineering and Computing, 37, Rizzolatti, G., Fogassi, L., & Gallese, V. (2001). 322–326. Neurophysiological mechanisms underlying the Watkins, K. E., Strafella, A. P., & Paus, T. (2003). Seeing and understanding and imitation of action. Nature Reviews hearing speech excites the motor system involved in speech Neuroscience, 2, 661–670. production. Neuropsychologia, 48, 989–994. Romanski, L. M., Tian, B., Fritz, J., Mishkin, M., Golman-Rakic, Wise, R. J. S., Scott, S. K., Blank, S. C., Mummery, C. J., Murphy, P. S., & Rauschecker J. P. (1999). Dual streams of auditory K., & Warburton E. A. (2001). Separate neural subsystems afferents target multiple domains in the primate prefrontal within ‘Wernicke’s area.’ Brain, 124, 83–95. cortex. Nature Neuroscience, 2, 1131–1136. Woods, R. P., Mazziotta, J. C., & Cherry, S. R. (1993). MRI-PET Strafella, A. P., & Paus, T. (2000). Modulation of cortical registration with automated algorithm. Journal of excitability during action observation: A transcranial Computer-Assisted Tomography, 17, 536–546. magnetic stimulation study. NeuroReport, 11, 2289–2292. Worsley, K. J., Evans, A. C., Marrett, S., & Neelin, P. (1992). Stromswold, K., Caplan, D., Alpert, N., & Rauch, S. (1996). A three-dimensional statistical analysis for CBF activation Localization of syntactic comprehension by positron studies in human brain. Journal of Cerebral Blood Flow emission tomography. Brain and Language, 52, 452–473. and Metabolism, 12, 900–918.

Watkins and Paus 987 Downloaded from http://www.mitpressjournals.org/doi/pdf/10.1162/0898929041502616 by guest on 27 September 2021