Abstract

Some neuroimaging studies have reported activation by imitation in the left Brodmann area 44 (BA 44), a part of Broca's area considered to be a neural substrate for speech production. However, in these previous studies the subjects were required to perform the same movements repeatedly so that the experimental stimuli could be viewed as cues when to do the given movements rather than a specification of what to do. Activation in the left BA 44 has also been observed in delayed motor execution tasks. This may confound the activity in BA 44 for imitation tasks used in the former studies. We tested the involvement of bilateral BA 44 and BA 45 in imitation and delayed execution tasks by functional magnetic resonance imaging with cytoarchitectonically defined BA 44 and BA 45 as volumes of interest. Our tasks required the subjects to perform a transformation from visual information (photographs of hand postures or symbolic specification of the postures) into a new hand movement in each trial. A three-way analysis of variance was performed with factors instruction, execution timing and area. The results revealed significant main effect by execution timing and by area, but not by instruction. We conclude that Broca's area does not play a pivotal role in imitation.

Introduction

Broca's area has been considered as a neural base for speech production (Broca, 1861; Mohr et al., 1978). A rigorous anatomical definition of the area has not been established, but it seems acceptable to suppose that Broca's area is comprised of Brodmann area 44 and 45 (BA 44 and BA 45), or the pars opercularis and trianguralis in the left hemisphere (Aboitiz and García, 1997; Amunts et al., 1999). Beyond language related functions, recent neuroimaging studies have suggested a new role of BA 44, a part of Broca's area, as a neural substrate for imitation (Krams et al., 1998; Iacoboni et al., 1999, 2001; Nishitani and Hari, 2000, 2002; Tanaka et al., 2001; Koski et al., 2002, 2003; Tanaka and Inui, 2002; Carr et al., 2003; Grèzes et al., 2003). However, we propose that the tasks employed in former neuroimaging studies on imitation were inappropriate. We define visuomotor imitation as follows: ‘to copy a visually demonstrated movements or postures of human body parts by one’s own body parts'.

In order to find the neural basis of imitation, experimental tasks must be designed to have a good variety of target movements and require that the subject make a transformation of the visual information into a new movement in each trial. If the subject repeats identical imitative movements, visuomotor transformation is not necessarily required in the later trials. Instead, the subject can perform the movement using an identical motor program. In such trials, the visually presented movements may be regarded as cues that trigger the learned movement. Therefore, the target movements to be imitated should have a good variety so that the subject cannot skip the visuomotor transformation in every trial.

We should exclude execution timing components from imitation tasks. In any motor behavior, ‘when to do’ and ‘what to do’ components exist, but we consider that a ‘what to do’ component is indispensable for imitation. Correct reproduction of postures without the exact control of execution timing may be classified as imitation, but movement without correct reproduction of postures cannot be accepted as successful imitative behavior. Tasks that require the subject to ‘imitate’ the timing of execution of a given movement do not involve visuomotor transformations into a movement of his/her own body part. Rather, the subject extracts the trigger for commencement of the movement from the visual scene, while he/she uses the same motor program repeatedly. Therefore we should exclude the ‘when to do’ component from the imitation task. Another reason to exclude timing properties from an imitation task comes from neuroimaging studies that reported BA 44 activation in tasks that did not share an imitative element at all. One such non-imitative element is the attention to the timing property of a stimulus. Schubotz and colleagues (Schubotz et al., 2000; Schubotz and von Cramon 2001) demonstrated that the perception of auditory or visually presented temporal patterns involves bilateral BA 44 (Schubotz et al., 2000) and that attention to the timing property of three consecutively presented visual objects activates BA 44 bilaterally (Schubotz and von Cramon 2001). Futhermore, we must also exclude timing properties related to delayed execution. A PET study revealed left BA 44 activation in a delayed execution condition compared with a prompt execution condition, even if the subjects used their left hand (Rushworth et al., 2001). Brain activity related to delayed execution was also studied using event-related fMRI by Toni et al. (1999, 2002). In the study of 1999, they showed ‘what to do’ with one of four shapes as an instruction cue. After a variable delay of 1.28 to 12.8 s, an auditory cue triggered the execution of the movement (Toni et al., 1999). The left frontal operculum was one of the activated areas associated with the trigger cue. Their second study used a similar experimental paradigm with a small modification in the variable delay and found blood oxygenation level dependent (BOLD) signal increase associated with the trigger cue in the posterior part of the right inferior frontal gyrus (Toni et al., 2002). Thoenissen and colleagues conducted a similar event-related fMRI experiment with the variable delay of 1.0–21 s (Thoenissen et al., 2002), in which they reported BOLD signal increase associated with a delay period in the left pars opercularis. These experiments indicate that the inferior frontal gyrus, possibly including BA 44, could be involved in delayed execution.

Grasping movements are inappropriate targets for imitation when one tests involvement of BA 44 in imitation since the movements may confound the activation even in self-initiated conditions as demonstrated by two neuroimaging studies of human precision grip (Ehrsson et al., 2000, 2001). A magnetoencephalograpy (MEG) study used pinching hand movements of the index finger and the thumb as an imitation task. Activity was observed in BA 44 for both imitation and non-imitation conditions in which both conditions employed self-paced pinching hand movements (Nishitani and Hari, 2000). An fMRI study by Grèzes et al. (2003) employed a grasping movement for imitation and found BA 44 activity.

Based on our definition and discussion above, we criticize that the tasks employed in most former studies did not satisfy the prerequisites to investigate the human brain mechanism for imitation since these studies lacked novelty for the visuomotor transformation in each trial, possibly contaminated the activation of BA 44 by the execution timing property, or used grasping movements (Krams et al., 1998; Iacoboni et al., 1999, 2001; Nishitani and Hari 2000, 2002; Koski et al., 2002, 2003; Carr et al., 2003; Grèzes et al., 2003). The tasks used in fMRI studies by Tanaka and colleagues satisfy our criteria for imitation. They reported BA 44 activation in finger configuration imitation tasks but not in hand/arm configuration imitation tasks (Tanaka et al., 2001; Tanaka and Inui, 2002).

By synthesizing the results of these earlier neuroimaging studies, we have formulated two hypotheses. The first hypothesis is that Broca's area is not a crucial neural component for imitation. The second hypothesis is that Broca's area is involved in delayed execution. In order to test these two hypotheses, we have performed an fMRI study employing a 2 × 2 factorial design for hand posture making tasks. One factor used an instruction method to create hand postures: imitation and symbolic instruction. For the other factor, motor execution with or without random delay were introduced in order to address the delay effect or ‘when to do’ effect on Broca's area. Digitized photographs of 96 right hand postures were shown to the subject to imitate. Identical postures were instructed by symbols on the fingers of a digitized photograph of an out-stretched hand with the palm up. Red or blue squares on the fingers indicated whether to extend that finger or to touch the thumb with that finger respectively. A symbol that specified one of four wrist configurations appeared at the lower right position of the hand. We performed a volume of interest (VOI) analysis using cytoarchitectonically defined left and right BA 44 and 45 (Amunts et al., 1999). None of the studies listed above used cytoarchitectonically defined BA 44 and 45 for localizing activations. Mean BOLD signals in the four conditions were extracted from the VOIs. These means were analyzed with a three-way analysis of variance (ANOVA) with factors Instruction (imitation/symbolic instruction), Execution timing (prompt execution/delayed execution) and Area (left BA 44, right BA 44, left BA 45 and right BA 45). Whole brain functional data of all subjects were analyzed with a random effect model to find neural networks that subserve hand movements required in the experimental conditions compared with the baseline condition. The main effect of the instruction method and main effect of execution timing were also tested.

Materials and Methods

Subjects

Nine young right-handed healthy subjects were examined (six male and three female). Handedness was assessed with the Edinburgh Inventory (mean = 96.6, range = 80–100; Oldfield, 1971). The mean age was 25.6 years and the range was 22–30 years. All, but one subject had no history of neurological disorders. One female subject had a petit mal before the age of 15 years. The experimental procedures were approved by the Research Ethics Committees of the Karolinska Hospital and written informed consents were given by all subjects.

Stimuli

Ninety-six right hand postures were selected for imitation. The hand postures consisted of combinations of finger and wrist configurations. For finger configurations, either pointing or touching the thumb were used as rules. These rules produced 18 and nine distinct finger configurations respectively. Wrist configurations produced four variations: the palm up or down, and the flexion or extension of the wrist (Fig. 1). A combination of finger and wrist configurations generated 108 distinct hand postures. We eliminated 12 postures that were relatively difficult to make based on an evaluation by four subjects who were not scanned. The remaining 96 hand postures were represented by symbols on a picture of a hand which showed the palm and fingers stretched out (Fig. 1). Finger configurations were indicated by red or blue squares on the tip of the finger to indicate whether to stick out or to touch the thumb respectively. Wrist configurations were indicated by white figures that appeared on the lower right side of the picture. Images of the hand postures were created using digitized photographs of the right hand of a model in natural color with a black background (Fig. 1). The size of the hand images was 5 × 5°. The duration of the picture presentation was 2 s. A 300 Hz beep sound was delivered for 300 ms to both ears through headphones as a cue to make hand postures. The beep sounded at the same time as the photograph presentation, or 3, 3.5, 4 or 4.5 s later. One trial lasted 5 s.

Figure 1.

Examples of hand postures used in the experiment. Each pair indicates an identical posture: the left side picture is for imitation; the right side picture is for symbolic instruction. Red squares mean ‘stick this finger out’. Blue squares mean ‘touch this finger with the thumb’. White symbols at the lower right of the hand specify wrist configurations: (A) (no symbol), palm upside; (B) palm down; (C) flexion of the wrist; (D) extension of the wrist.

Figure 1.

Examples of hand postures used in the experiment. Each pair indicates an identical posture: the left side picture is for imitation; the right side picture is for symbolic instruction. Red squares mean ‘stick this finger out’. Blue squares mean ‘touch this finger with the thumb’. White symbols at the lower right of the hand specify wrist configurations: (A) (no symbol), palm upside; (B) palm down; (C) flexion of the wrist; (D) extension of the wrist.

Experimental Design

The stimulus presentation was programmed with the Presentation 0.53 software (Neurobehavioral Systems, Inc., Albany, CA) on a Windows notebook PC (Toshiba Dynabook G5/X16PME). Stimuli were projected through a liquid crystal display (LCD) projector (Philips Hopper XG20 Impact) onto the back of a screen that stood at the foot of the gantry of the MRI scanner. To prevent head motion, the forehead and chin of each subject was fixed with straps and cushions. The subjects viewed the images on the screen through binoculars attached to the head-coil. The binoculars had a magnification factor of three. Feedback of hand movements was not provided to the subjects.

Experimental Conditions

A 2 × 2 factorial design with factors Instruction (imitation/symbolic instruction) and Execution timing (prompt execution/delayed execution) was employed. Therefore, the experiment was comprised of four conditions each requiring the subject to make hand postures.

Imitation/Prompt Execution Condition (I)

In this condition the subject was required to mimic immediately a hand posture presented to them in a digitized photograph of a hand. An auditory ‘beep’ cue (300 Hz) accompanied the visual presentation.

Imitation/Delayed Execution Condition (ID)

This task was similar to task I, except that the subject was required to delay the motor execution. An auditory ‘beep’ cue, delivered 3 to 4.5 s after the photograph presentation, indicated when the subject should mimic the posture.

Symbolic Instruction/Prompt Execution Condition (S)

In this condition the subject was instructed to make a hand posture specified by symbols on a picture of a hand with the palm and fingers stretched out. Red or blue squares on the fingers indicated that the subject should either extend the finger or touch the thumb respectively. Wrist configurations were specified by symbols that appeared on the lower right side of the hand. The visual symbolic instruction and the beep were presented simultaneously.

Symbolic Instruction/Delayed Execution Condition (SD)

This condition was the same as the S task, except that a delay of 3–4.5 s was introduced between the visual symbolic instruction and the motor execution. An auditory cue was used to indicate when the subjects should perform the mimic posture.

Baseline Condition (B)

A baseline condition was included in the experiment. The subject was instructed to fixate their eyes on a small white fixation point on a black background.

The fMRI scanning was performed as a block design. One trial had a duration of 5 s and one block had eight trials, so one block had a 40 s duration. The four experimental conditions and the baseline condition were repeated three times in a pseudorandom order. One session consisted of fifteen blocks that lasted 10 min. For each scanning session, half of the 96 postures were presented as the I and ID tasks, and the other half were presented as the S and SD tasks. Another session was made employing the same halves, but the condition allocations were swapped. We made four fMRI scanning sessions using this rule. The order of sessions in the fMRI experiment was counter balanced amongst subjects. Hand movements of all subjects were recorded on video tapes during the fMRI scanning. Before the scan, the subjects practiced each of the conditions outside of the scanner. The I, ID, S, and SD conditions were practiced until they were performed with 100% accuracy. The practice lasted 30–40 min.

Image Acquisition

EPI data were acquired with a 1.5 T Signa HORIZON (GE, Milwaukee, WI, USA). The whole brain was covered with 4 mm thick 34–36 axial images without slice gaps (TR = 4.0 s, TE = 50 ms, FOV = 24 × 24 cm2, 64 × 64 matrix). We had four sessions of fMRI scanning with 154 volumes per session. A total of 616 volumes were collected in ∼90 min. Structural high-resolution T1 images of all subjects were collected before the experimental sessions with a three-dimensional SPGR sequence (TR = 13 ms, TE = 4.2 ms, FOV = 24 × 24 cm2, 256 × 256 matrix, coronal 156 slices, 1.4 mm thick).

Analysis

Preprocessing

The first four volumes of each session were discarded to eliminate magnetic saturation effects; a total of 150 volumes per session were used. The data analysis was carried out using SPM99 (Friston et al., 1994, 1995a,b; Worsley and Friston, 1995) implemented in Matlab (Mathworks Inc., USA) on a SUN Ultra platform. As a preprocessing step, the EPI images were realigned to the first image. Mean EPI images were created for each subject to estimate the normalization parameters. The mean EPI images were co-registered to the same subjects' individual T1 image. Normalizing an individual T1 image to the standard brain template (Roland et al., 1994) was processed in two steps: (i) estimation of the normalization parameters; and (ii) writing the normalized images with the parameters. This parameter transformed the T1 and all the EPI volumes into a common stereotaxic space to allow a cytoarchitectonically defined VOI analysis and multi-subject analysis. The images were resampled into 3 × 3 × 3 mm3 voxels with sinc interpolation. For the whole brain analysis, all volumes were smoothed with an 8 mm full width at half maximum isotropic Gaussian kernel prior to the statistical calculation in order to improve the signal to noise ratio and to compensate for the anatomical variability among individual brains.

VOI Analysis

The cytoarchitectonic population map is a digitized three-dimensional population map of cytoarchitectonic areas from 10 post-mortem human brains in which borders of areas were determined by statistically significant changes in laminar density patterns of neuronal cell bodies (Roland and Zilles, 1998; Schleicher et al., 1999). A probability map of the cytoarchitectonic areas was created from the population map by choosing voxels that had higher probability of belonging to one area than to any other area (Bodegard et al., 2001). We defined volumes of interest for BA 44 and 45 in both hemispheres according to a cytoarchitectonic probability map (Amunts et al., 1999; Fig. 2). BA 44 was defined as a set of voxels that were assigned to BA 44 more often than any other areas among 10 brains. Because the labeling of cytoarchitectonic areas anterior and superior to BA 45 was not yet completed, we used a probability map truncated at 30% to define BA 45. This gave almost the same border between BA 44 and BA 45, but the extension of BA 45 into anterior and superior space was more conservative than using a non-truncated (>0% exclusive) probability map.

Figure 2.

VOIs for ANOVA BA 44 (yellow) and 45 (red) defined by a probability map of cytoarchitectonics delineations of 10 post-mortem brains (Amunts et al., 1999).

Figure 2.

VOIs for ANOVA BA 44 (yellow) and 45 (red) defined by a probability map of cytoarchitectonics delineations of 10 post-mortem brains (Amunts et al., 1999).

The Marsbar-0.21toolbox (Brett et al., 2002), implemented within SPM99, was used for statistical calculation for the VOI analysis. Non-spatially smoothed fMRI data were used in this VOI analysis to avoid contamination of the signal of BA 44 and BA 45 from surrounding voxels. The global mean intensity of each session was normalized to 100. Each subjects' haemodynamic response, induced by the experimental condition was modeled with a box car function delayed by 6 s convolved with a haemodynamic function. The mean signal intensity of each condition was estimated according to the general linear model. Confounds by global signal changes were removed by applying a high pass filter with a cut-off cycle of 160 s. Estimated mean signal intensities of the VOIs in each condition of each subject were used as variables in the three-wayANOVA with factors Instruction (imitation/symbolic instruction), Execution timing (prompt execution/delayed execution) and Area (left BA 44, right BA 44, left BA 45 and right BA 45).

Whole Brain Analysis

The fMRI data were smoothed with an 8mm isotropic Gaussian filter and was used for individual analysis with SPM99 using the same model as the VOI analysis. A group analysis was implemented according to the random effect model (Holmes and Friston, 1998) in which each subject's response was considered as a variable drawn from the population. Each experimental condition was compared against the baseline condition (I > B, ID > B, S > B and SD > B). The main effect of the Instruction method (imitation > symbolic instruction, I + ID > S + SD; symbolic instruction > imitation, S + SD > I + ID) and the main effect of Execution timing (delayed execution > prompt execution, ID + SD > I + S; prompt execution > delayed execution, I + S > ID + SD) were also tested. Statistical inferences were made at the cluster-level by thresholding at t = 5 in each voxel and a P < 0.05 for spatial extent corrected for multiple non-independent comparisons in accordance with the Gaussian random field theory (Worsley and Friston, 1995). A cytoarchitectonic probability map (Roland and Zilles, 1998) was used to label the activated clusters for available areas: somatosensory areas 3a, 3b, 1 (Geyer et al., 1999); area 2 (Grefkes et al., 2001); motor areas 4a and 4p (Geyer et al., 1996); visual areas 17 and 18 (Amunts et al., 2000); Broca's areas 44 and 45 (Amunts et al., 1999); V5: visual area 5 (not yet published); and BA 6 (not yet published).

Results

Behavioral Performance

The mean ± SD percentage of correct responses was 97.4 ± 1.08%. Errors were typically observed in the first trial of delayed conditions (ID and SD) as responding before the beep. Carrying over of this error to the next trial was rare. One posture in the symbolic instruction condition caused error responses because the posture was used in both imitation and symbolic instruction condition with two different meanings. We cannot analyze the performances of one subject and a part of the performance (24 trials of 384 trials, 6.25%) of another subject due to problems in the video recording, but we observed good performances of these two subjects during the practice sessions and the scanning sessions.

VOI Analysis

Signal increase in each condition relative to the baseline condition was calculated for each VOI (Fig. 3). A three-way analysis of variance (ANOVA) with factors Instruction (imitation/symbolic instruction), Execution timing (prompt execution/delayed execution), and Area (left BA 44, right BA 44, left BA 45 and right BA 45) was conducted on the data. There were significant main effects of Execution timing [F(1,8) = 29.55, P < 0.01, delayed execution > prompt execution] and Area [F(3,24) = 10.90, P < 0.01], but not in instruction. A significant interaction between Instruction and Area was found [F(3,24) = 6.94, P < 0.01]. The effect of Area was significant for both imitation [F(3,24) = 14.16, P < 0.01] and symbolic instruction [F(3,24) = 7.87, P < 0.01]. A significant difference among areas was found for the imitation condition as R44 > L44 ≈ R45 > L45 and for the symbolic instruction as R44 > L44> R45 > L45.

Figure 3.

Mean VOI signal intensities for each condition. Mean signal intensities of four VOIs (left BA 45, left BA 44, right BA 45 and right BA 44) in four conditions (I, ID, S and SD) are shown. Error bars indicate standard error.

Figure 3.

Mean VOI signal intensities for each condition. Mean signal intensities of four VOIs (left BA 45, left BA 44, right BA 45 and right BA 44) in four conditions (I, ID, S and SD) are shown. Error bars indicate standard error.

Whole Brain Analysis

Each condition was contrasted with the baseline condition (I > B, ID > B, S > B, and SD >B). Figure 4 shows activated clusters projected on the brain surface revealed by contrasts of each condition against the baseline condition. Activation in left BA 44 and BA 45 appeared in the delayed execution condition (ID >B, SD >B) but not in the prompt execution condition (I > B, S > B). On the other hand, right BA 44 activity was found in all of the contrasts. Right BA 45 activation was observed in the contrasts I > B, ID > B and SD > B, but not in S > B. This analysis revealed a whole brain network to perform these tasks and consists of the prefrontal area, premotor area, sensorimotor area, parietal area, temporo-occipital area, thalamus, and the cerebellum. We observed a trend for the delayed execution conditions in which there were larger clusters when compared with the corresponding prompt execution conditions.

Figure 4.

Surface rendered brains for activations in each condition (I, ID, S, and SD) compared with the baseline condition (B). Top left: I > B. Bottom left: ID > B. Top right: S > B. Bottom right: SD > B. Statistical inference was made at the cluster-level by thresholding at t = 5 in each voxel, then P < 0.05 for the spatial extent corrected for multiple non-independent comparisons in accordance with the Gaussian random field theory.

Figure 4.

Surface rendered brains for activations in each condition (I, ID, S, and SD) compared with the baseline condition (B). Top left: I > B. Bottom left: ID > B. Top right: S > B. Bottom right: SD > B. Statistical inference was made at the cluster-level by thresholding at t = 5 in each voxel, then P < 0.05 for the spatial extent corrected for multiple non-independent comparisons in accordance with the Gaussian random field theory.

Main Effect of Instruction

An activated cluster was found in the left visual area 5 (V5) for the contrast of imitation versus symbolic instruction (I + ID > S + SD; Table 1). Other brain regions, such as white matter, survived the threshold, but we neglected these clusters because they were ‘false’ activations. Those clusters were not found in either contrasts of I > B or ID > B that showed a neural network to perform imitation. Therefore, these clusters were not activated by imitation. This activation is explained by a signal decrease in symbolic instruction conditions (S and SD) relative to the baseline condition (B). We do not know why some brain regions have signal decreases in some tasks, but we have often observed this phenomenon. In the contrast of I + ID > S + SD, we compared signal intensities in imitation conditions and in symbolic instruction conditions. If a region has a lower signal in symbolic instruction conditions relative to the baseline condition, the difference of signals between imitation conditions and symbolic instruction conditions can be significant even if the region is not activated by imitation. Considering the aim of the contrast, we treat such a case as a ‘false’ activation. In the statistical comparison of symbolic instruction versus imitation (S + SD > I + ID, Table 2), the lingual gyrus was activated bilaterally.

Table 1

Imitation versus symbolic instruction (I + ID > S + SD)

Cluster p (corrected)
 
Number of voxels in the cluster
 
Voxel level Z
 
Coordinates of peaks x, y, z (mm)
 
  anatomical location
 
0.006 25 4.39 −48 −69 left V5 

 

 
4.19
 
−42
 
−81
 
−3
 
left V5
 
Cluster p (corrected)
 
Number of voxels in the cluster
 
Voxel level Z
 
Coordinates of peaks x, y, z (mm)
 
  anatomical location
 
0.006 25 4.39 −48 −69 left V5 

 

 
4.19
 
−42
 
−81
 
−3
 
left V5
 

V5: visual area 5. Statistical threshold was set to P < 0.05 corrected for multiple comparisons with cluster level inference. Anatomical location was determined by the cytoarchitectonic probability map (Roland and Zilles, 1998).

Table 2

Symbolic instruction versus imitation (S + SD > I + ID)

Cluster p (corrected)
 
Number of voxels in the cluster
 
Voxel level Z
 
Coordinates of peaks x, y, z (mm)
 
  Anatomical location
 
Cytoarchitectonic area
 
0.000 63 4.05 21 −72 −9 right lingual gyrus 18 
  3.8 27 −69 −21 right lingual gyrus  
0.000 53 3.91 −84 −6 right lingual gyrus 17 
  3.7 −90 −9 left lingual gyrus 17 

 

 
3.56
 
6
 
−75
 
−9
 
right lingual gyrus
 
17
 
Cluster p (corrected)
 
Number of voxels in the cluster
 
Voxel level Z
 
Coordinates of peaks x, y, z (mm)
 
  Anatomical location
 
Cytoarchitectonic area
 
0.000 63 4.05 21 −72 −9 right lingual gyrus 18 
  3.8 27 −69 −21 right lingual gyrus  
0.000 53 3.91 −84 −6 right lingual gyrus 17 
  3.7 −90 −9 left lingual gyrus 17 

 

 
3.56
 
6
 
−75
 
−9
 
right lingual gyrus
 
17
 

Statistical threshold was set to P < 0.05 corrected for multiple comparisons with cluster level inference. Cytoarchitectonic area was determined by the cytoarchitectonic probability map (Roland and Zilles, 1998).

Main Effect of Execution Timing

The contrast of delayed execution versus prompt execution (ID + SD > I + S; Table 3) revealed activated clusters in the bilateral dorsal premotor areas, left sensorimotor area, bilateral parietal areas, left insula, bilateral superior temporal sulci, bilateral pars opercularis (BA 44; Fig. 5), bilateral supplementary motor areas, and the right cerebellum (Table 3). The reverse contrast (I + S > ID + SD) did not give any suprathreshold clusters.

Figure 5.

Activation in BA 44 revealed by the contrast of delayed execution condition against prompt execution condition (ID + SD > I + S). Note that activated clusters appear in BA 44 bilaterally. An activated cluster was also found in the posterior part of the right superior temporal sulcus. Statistical inferences were made at the cluster-level by thresholding at t = 5 in each voxel, then P < 0.05 for the spatial extent corrected for multiple non-independent comparisons in accordance with the Gaussian random field theory. The x Talairach coordinate of the slice is given in the top left corner of each slice.

Figure 5.

Activation in BA 44 revealed by the contrast of delayed execution condition against prompt execution condition (ID + SD > I + S). Note that activated clusters appear in BA 44 bilaterally. An activated cluster was also found in the posterior part of the right superior temporal sulcus. Statistical inferences were made at the cluster-level by thresholding at t = 5 in each voxel, then P < 0.05 for the spatial extent corrected for multiple non-independent comparisons in accordance with the Gaussian random field theory. The x Talairach coordinate of the slice is given in the top left corner of each slice.

Table 3

Delayed execution versus prompt execution (ID + SD > I + S)

Cluster p (corrected)
 
Number of voxels in the cluster
 
Voxel level Z
 
Coordinates of peaks x, y, z (mm)
 
  Anatomical location
 
Cytoarchitectonic area
 
0.000 576 4.52 −39 −54 36 left intraparietal sulcus  
  4.16 −27 −15 60 left dorsal premotor area 
  4.13 −39 −27 45 left primary motor area 4a 
0.000 384 4.56 −15 54 right supplementary motor area 
  4.45 54 15 right inferior frontal gyrus 44 
  4.41 42 −12 33 right primary motor area 4p 
0.000 275 4.77 36 −45 36 right intraparietal sulcus  
  4.33 45 −45 33 right intraparietal sulcus  
  4.17 48 −39 36 right intraparietal sulcus  
0.000 88 4.33 −42 24 left inferior frontal gyrus 44 
  3.92 −48 left inferior frontal gyrus 44 
0.000 65 4.34 −9 −60 51 left superior parietal lobule  
  4.23 −6 −54 54 left postcentral gyrus 3b 
  3.8 −15 −57 45 left superior parietal lobule  
0.000 48 4.1 −36 21 27 left inferior frontal gyrus 45 
  4.03 −42 21 33 left inferior frontal gyrus 45 
  3.92 −39 24 30 left middle frontal gyrus  
0.000 40 4.51 63 −39 right posterior superior temporal sulcus  
  4.1 57 −45 12 right posterior superior temporal sulcus  
0.002 30 3.98 36 −51 −36 right Cerebellum  
  3.65 42 −57 −30 right Cerebellum  
0.01 22 3.91 −36 27 left insula  
  3.68 −39 33 left insula  
  3.44 −30 21 left insula  
0.047
 
15
 
4.33
 
−51
 
−24
 
3
 
left anterior superior temporal sulcus
 

 
Cluster p (corrected)
 
Number of voxels in the cluster
 
Voxel level Z
 
Coordinates of peaks x, y, z (mm)
 
  Anatomical location
 
Cytoarchitectonic area
 
0.000 576 4.52 −39 −54 36 left intraparietal sulcus  
  4.16 −27 −15 60 left dorsal premotor area 
  4.13 −39 −27 45 left primary motor area 4a 
0.000 384 4.56 −15 54 right supplementary motor area 
  4.45 54 15 right inferior frontal gyrus 44 
  4.41 42 −12 33 right primary motor area 4p 
0.000 275 4.77 36 −45 36 right intraparietal sulcus  
  4.33 45 −45 33 right intraparietal sulcus  
  4.17 48 −39 36 right intraparietal sulcus  
0.000 88 4.33 −42 24 left inferior frontal gyrus 44 
  3.92 −48 left inferior frontal gyrus 44 
0.000 65 4.34 −9 −60 51 left superior parietal lobule  
  4.23 −6 −54 54 left postcentral gyrus 3b 
  3.8 −15 −57 45 left superior parietal lobule  
0.000 48 4.1 −36 21 27 left inferior frontal gyrus 45 
  4.03 −42 21 33 left inferior frontal gyrus 45 
  3.92 −39 24 30 left middle frontal gyrus  
0.000 40 4.51 63 −39 right posterior superior temporal sulcus  
  4.1 57 −45 12 right posterior superior temporal sulcus  
0.002 30 3.98 36 −51 −36 right Cerebellum  
  3.65 42 −57 −30 right Cerebellum  
0.01 22 3.91 −36 27 left insula  
  3.68 −39 33 left insula  
  3.44 −30 21 left insula  
0.047
 
15
 
4.33
 
−51
 
−24
 
3
 
left anterior superior temporal sulcus
 

 

Statistical threshold was set to P < 0.05 corrected for multiple comparisons with cluster level inference. Cytoarchitectonic area was determined by the cytoarchitectonic probability map (Roland and Zilles, 1998).

Discussion

Activation of Broca's Area

The results of the VOI analysis indicate that activities in left BA 44 and BA 45, or in Broca's area, are present in all conditions. However, an ANOVA shows that the activity is not influenced by the instruction method for making the hand postures, but is affected by the execution timing of the movement. Therefore, we can say that BA 44, or Broca's area, is not specifically activated by imitation. The discrepancy between our results and that of former studies should be sought in the differences in the experimental tasks (Krams et al., 1998; Iacoboni et al., 1999, 2001; Koski et al., 2002, 2003; Carr et al., 2003; Grèzes et al., 2003). We have pointed out in the introduction that the tasks for imitation must have a good variety of movements to copy; otherwise the stimuli can be viewed as triggers for execution timing of a given movement, which was not fulfilled in the preceding studies. In contrast, we employed an imitation task with 96 distinct hand postures as targets to imitate. This enabled us to observe brain activity associated with visuomotor transformation from a visually presented target posture into a hand movement to imitate the posture. From another point of view, the difference between the tasks is in the manner of presentation of targets: animated hand (Iacoboni et al., 1999, 2001; Koski et al., 2002, 2003; Grèzes et al., 2003) or still hand postures. Visual analysis of a moving hand is more complex than that of a still hand posture, since the spatial relations among body parts keep changing and the observer has to update the state of the hand posture. Therefore, a visuomotor transformation of animated hand movements is more difficult than that of still hand postures. Imitation with real time tracking of target movements will require the correct reproduction of the timing aspects of the movements. These timing aspects are not required in our imitation tasks using still hand postures.

Imitation

Neuropsychological studies of imitation deficit after brain damage, namely ideomotor apraxia, have attributed the deficit to a left parietal lesion, not to Broca's area (De Ajuriaguerra et al., 1960; De Renzi, 1989; Haaland et al., 2000). We found activation in the left parietal area in contrasts of each condition against the baseline condition (Fig. 4), but the effect of instruction was insignificant in this area (Tables 1 and 2). These results indicate that both imitation and movements for symbolic imitation need the left parietal area. The left parietal area seems to have an important role in movements specified by external stimuli, such as imitation and movement on verbal command (ideomotor apraxia), or movements due to symbolic instructions. The contrast of imitation versus symbolic instruction (I + ID > S + SD) revealed the left visual area 5 (V5). This is in agreement with the result of the PET study by Hermsdörfer et al. (2001) that reported activation in the temporo-occipital junction area, presumably V5, during the discrimination of hand and finger gestures. It is not clear why static posture stimuli can activate V5, which seems to be a neural correlate of visual motion processing, but our results and those of Hermsdörfer et al. could imply that the area is involved in the visual analysis of hand or finger configurations. Alternatively, this area might correspond to the ‘extrastriate body area (EBA)’ that is selective for visual processing of the human body (Downing et al., 2001). It is obvious that V5 or EBA are not dedicated selectively to imitation, but this area may have an essential role in imitation and a lesion to this area may cause a disturbance in imitation.

Some researchers consider that the goal directed action or target directed action is relevant in human imitative behavior. In goal directed actions, imitation is achieved by understanding the goal of the visually presented actions and then performing the movements by selecting the appropriate motor programs from action repertoires (Bekkering et al., 2000; Koski et al., 2002; Wohlschläger et al., 2003). When one uses this strategy for imitation, the most important goal of the action is copied exactly, but less dominant goals can be omitted (Bekkering et al., 2000; Wohlschläger et al., 2003). They observed that imitation in pre-school children was more goal-directed than an exact copy of the movements. This is a good example to understand the nature of imitation as being goal or target directed (Bekkering et al., 2000). According to this theory, we can view that the hand posture itself is the goal in our study (Wohlschläger et al., 2003). However, in the current study, it is not clear if the subjects performed imitation by understanding the goal and selected the appropriate motor program from pre-existing action repertoires. These imitative behaviors may recruit different cortical networks and may include BA 44. Since our experimental design does not allow the discussion of this matter, further studies are needed to elucidate possible differences in brain networks between imitation of body postures per se and imitation of the goal.

Neural Substrates for Delayed Execution

Our results indicate that a cortical network that includes BA 44 and the superior temporal sulcus subserve delayed execution. Activation in the superior temporal sulcus has been found during the delay period in delayed execution tasks using event-related MRI techniques (Toni et al., 1999, 2002; Thoenissen et al., 2002). The cortex lining the superior temporal sulcus was also activated in an imitation study with fMRI (Iacoboni et al., 2001). Since the experimental task employed by Iacoboni and colleagues' was similar to their former study (Iacoboni et al., 1999), our criticism noted in the introduction section applies again. Observed activation might be attributed to attention or perception of the timing property of the moving stimuli. We consider that neither Broca's area nor the superior temporal sulcus has a pivotal role in imitation. These two regions are parts of the neural network subserving delayed execution.

Function of BA 44

Why does BA 44 participate in delayed execution? We can take the view that the contrast of delayed execution versus prompt execution shows preparatory activity for movement. It is then possible to reconcile the traditional view of BA 44 as a speech production area, but as a neural base of movement preparation. Speech consists of a chain of syllables or words in order. The brain should prepare each movement for the syllables with a proper sequence. Each movement for a syllable has to wait its turn, that is, delayed movement from the beginning of the utterance. Speech is such a complex sequential movement that must be performed within a short time that preparatory activity is crucial. Challenges to the classical view of Broca's area are one of the most interesting attempts of using neuroimaging methods.

I thank Professor Per Roland for useful discussion and helpful comments on the manuscript. Special thanks to Jeremy Young for useful comments and kind suggestions to improve the manuscript. Many thanks to Sara Bengtsson and Martin Sundström for kind help in the MRI scanning. This work was supported by the EU-project NEUROGENERATOR, for the advancement of Neuroinformatics (QLG 1999 00677).

References

Aboitiz F, García RV (
1997
) The evolutionary origin of the language areas in the human brain. A neuroanatomical perspective.
Brain Res Brain Res Rev
 
25
:
381
–396.
Amunts K, Schleicher A, Burgel U, Mohlberg H, Uylings HB, Zilles K (
1999
) Broca's region revisited: cytoarchitecture and intersubject variability.
J Comp Neurol
 
412
:
319
–341.
Amunts K, Malikovic A, Mohlberg H, Schormann T, Zilles K (
2000
) Brodmann's areas 17 and 18 brought into stereotaxic space where and how variable?
Neuroimage
 
11
:
66
–84.
Bekkering H, Wohlschlager A, Gattis M (
2000
) Imitation of gestures in children is goal-directed.
Q J Exp Psychol A
 
53
:
153
–164.
Bodegard A, Geyer S, Grefkes C, Zilles K, Roland PE (
2001
) Hierarchical processing of tactile shape in the human brain.
Neuron
 
31
:
317
–328.
Brett M, Anton JL, Valabregue R, Poline JB (
2002
) Region of interest analysis using an SPM toolbox.
Neuroimage
 
16
:abstr 497.
Broca MP (
1861
) Remarques sur le siege de la faculté du langage articulé, suivies d'une observation d'aphemie (Perte de la Parole).
Bull Mem Soc Anat Paris
 
36
:
330
–357.
Carr L, Iacoboni M, Dubeau MC, Mazziotta JC, Lenzi GL (
2003
) Neural mechanisms of empathy in humans: a relay from neural systems for imitation to limbic areas.
Proc Natl Acad Sci USA
 
100
:
5497
–5502.
De Ajuriaguerra J, Hecaen H, Angelergues R (
1960
) Les apraxies Variétés cliniques et latéralisation lésionelle.
Rev Neurol
 
102
:
566
–594.
De Renzi E (
1989
) Apraxia. In: Handbook of neuropsychology (Boller F, Grafman J, eds), Vol. 2, pp. 245–263. Amsterdam: Elsevier.
Downing PE, Jiang Y, Shuman M, Kanwisher N. (
2001
) A cortical area selective for visual processing of the human body.
Science
 
293
:
2470
–2473.
Ehrsson HH, Naito E, Geyer S, Amunts K, Zilles K, Forssberg H, Roland PE (
2000
) Simultaneous movements of upper and lower limbs are coordinated by motor representations that are shared by both limbs: a PET study.
Eur J Neurosci
 
12
:
3385
–3398.
Ehrsson HH, Fagergren E, Forssberg H (
2001
) Differential fronto-parietal activation depending on force used in a precision grip task: an fMRI study.
J Neurophysiol
 
85
:
2613
–2623.
Friston KJ, Jezzard P, Turner R (
1994
) Analysis of fMRI time-series.
Hum Brain Mapp
 
1
:
153
–171.
Friston KJ, Holmes AP, Poline JB, Grasby PJ, Williams SC, Frackowiak RS, Turner R (
1995
a) Analysis of fMRI time-series revisited.
Neuroimage
 
2
:
45
–53.
Friston KJ, Holmes AP, Worsley KJL, Poline JB, Frith CD, Frackowiak RSJ (
1995
b) Statistical parametric maps in functional imaging: a general linear approach.
Hum Brain Mapp
 
2
:
189
–210.
Geyer S, Ledberg A, Schleicher A, Kinomura S, Schormann T, Burgel U, Klingberg T, Larsson J, Zilles K, Roland PE (
1996
) Two different areas within the primary motor cortex of man.
Nature
 
382
:
805
–807.
Geyer S, Schleicher A, Zilles K (
1999
) Areas 3a, 3b, and 1 of human primary somatosensory cortex.
Neuroimage
 
10
:
63
–83.
Grefkes C, Geyer S, Schormann T, Roland P, Zilles K (
2001
) Human somatosensory area 2: observer-independent cytoarchitectonic mapping, interindividual variability, and population map.
Neuroimage
 
14
:
617
–631.
Grèzes J, Armony JL, Rowe J, Passingham RE (
2003
) Activations related to ‘mirror’ and ‘canonical’ neurones in the human brain: an fMRI study.
Neuroimage
 
18
:
928
–937.
Haaland KY, Deborah L, Harrington DL, Knight RT (
2000
) Neural representations of skilled movement.
Brain
 
123
:
2306
–2313.
Hermsdörfer J, Goldenberg G, Wachsmuth C, Conrad B, Ceballos-Baumann AO, Bartenstein P, Schwaiger M, Boecker H (
2001
) Cortical correlates of gesture processing: clues to the cerebral mechanisms underlying apraxia during the imitation of meaningless gestures.
Neuroimage
 
14
:
149
–161.
Holmes AP, Friston KJ (
1998
) Generalisability, random effects & population inference.
Neuroimage
 
7
:
S754
.
Iacoboni M, Woods RP, Brass M, Bekkering H, Mazziotta JC, Rizzolatti G (
1999
) Cortical mechanisms of human imitation.
Science
 
286
:
2526
–2528.
Iacoboni M, Koski LM, Brass M, Bekkering H, Woods RP, Dubeau MC, Mazziotta JC, Rizzolatti G (
2001
) Reafferent copies of imitated actions in the right superior temporal cortex.
Proc Natl Acad Sci USA
 
98
:
13995
–13999.
Koski L, Wohlschlager A, Bekkering H, Woods RP, Dubeau MC, Mazziotta JC, Iacoboni M (
2002
) Modulation of motor and premotor activity during imitation of target-directed actions.
Cereb Cortex
 
12
:
847
–855.
Koski L, Iacoboni M, Dubeau MC, Woods RP, Mazziotta JC (
2003
) Modulation of cortical activity during different imitative behaviors.
J Neurophysiol
 
89
:
460
–471.
Krams M, Rushworth MF, Deiber MP, Frackowiak RSJ, Passingham RE (
1998
) The preparation, execution and suppression of copied movements in the human brain.
Exp Brain Res
 
120
:
386
–398.
Mohr JP, Pessin MS, Finkelstein S, Funkenstein HH, Duncan GW, Davis KR (
1978
) Broca aphasia: pathologic and clinical.
Neurology
 
28
:
311
–324.
Nishitani N, Hari R (
2000
) Temporal dynamics of cortical representation for action.
Proc Natl Acad Sci USA
 
97
:
913
–918.
Nishitani N, Hari R (
2002
) Viewing lip forms: cortical dynamics.
Neuron
 
36
:
1211
–1220.
Oldfield RC (
1971
) The assessment and analysis of handedness: the Edinburgh inventory.
Neuropsychologia
 
9
:
97
–113.
Roland PE, Zilles K (
1998
) Structural divisions and functional fields in the human cerebral cortex.
Brain Res Brain Res Rev
 
26
:
87
–105.
Roland PE, Raufelds CJ, Wahlin L, Ingelman L, Andersson M, Ledberg A, Pedersen J, Akerman S, Dabringhaus A, Zilles K (
1994
) Human brain atlas: for high resolution functional and anatomical mapping.
Hum Brain Mapp
 
1
:
173
–184.
Rushworth MF, Krams M, Passingham RE (
2001
) The attentional role of the left parietal cortex: the distinct lateralization and localization of motor attention in the human brain.
J Cogn Neurosci
 
13
:
698
–710.
Schleicher A, Amunts K, Geyer S, Morosan P, Zilles K (
1999
) Observer-independent method for microstructural parcellation of cerebral cortex: a quantitative approach to cytoarchitectonics.
Neuroimage
 
9
:
165
–177.
Schubotz RI, von Cramon DY (
2001
) Functional organization of the lateral premotor cortex: fMRI reveals different regions activated by anticipation of object properties, location and speed.
Brain Res Cogn Brain Res
 
11
:
97
–112.
Schubotz RI, Friederici AD, von Cramon DY (
2000
) Time perception and motor timing: a common cortical and subcortical basis revealed by fMRI.
Neuroimage
 
11
:
1
–12.
Tanaka S, Inui T (
2002
) Cortical involvement for action imitation of hand/arm postures versus finger configurations: an fMRI study.
Neuroreport
 
13
:
1599
–1602.
Tanaka S, Inui T, Iwaki S, Konishi J, Nakai T (
2001
) Neural substrates involved in imitating finger configurations: an fMRI study.
Neuroreport
 
12
:
1171
–1174.
Thoenissen D, Zilles K, Toni I (
2002
) Differential involvement of parietal and precentral regions in movement preparation and motor intention.
J Neurosci
 
22
:
9024
–9034.
Toni I, Schluter ND, Josephs O, Friston K, Passingham RE (
1999
) Signal-, set- and movement-related activity in the human brain: an event-related fMRI study.
Cereb Cortex
 
9
:
35
–49.
Toni I, Shah NJ, Fink GR, Thoenissen D, Passingham RE, Zilles K (
2002
) Multiple movement representations in the human brain: an event-related fMRI study.
J Cogn Neurosci
 
14
:
769
–784.
Wohlschläger A, Gattis M, Bekkering H. (
2003
) Action generation and action perception in imitation: an instance of the ideomotor principle.
Philos Trans R Soc Lond B Biol Sci
 
358
:
501
–515.
Worsley KJ, Friston KJ (
1995
) Analysis of fMRI time-series revisited — again.
Neuroimage
 
2
:
173
–181.