Visual detection of body motion is of immense importance for daily-life activities and social nonverbal interaction. Although neurobiological mechanisms underlying visual processing of human locomotion are being explored extensively by brain imaging, the role of structural brain connectivity is not well understood. Here we investigate cortical evoked neuromagnetic response to point-light body motion in healthy adolescents and in patients with early periventricular lesions, periventricular leukomalacia (PVL), that disrupt brain connectivity. In a simultaneous masking paradigm, participants detected the presence of a point-light walker embedded in a few sets of spatially scrambled dots on the joints of a walker. The visual sensitivity to camouflaged human locomotion was lower in PVL patients. In accord with behavioral data, root-mean-square (RMS) amplitude of neuromagnetic trace in response to human locomotion was lower in PVL patients at latencies of 180–244 msec over the right temporal cortex. In this time window, the visual sensitivity to body motion in controls, but not in PVL patients, was inversely linked to the right temporal activation. At later latencies of 276–340 msec, we found reduction in RMS amplitude in PVL patients for body motion stimuli over the right frontal cortex. The findings indicate that disturbances in brain connectivity with the right temporal cortex, a key node of the social brain, and with the right frontal cortex lead to disintegration of the neural network engaged in visual processing of body motion. We suspect that reduced cortical response to body motion over the right temporal and frontal cortices might underlie deficits in visual social cognition.
Human body motion provides a rich source of visual information about others, which is of tremendous significance not only for a variety of daily-life activities but also for social interaction and nonverbal communication. The temporal cortex and, in particular, the right STS, a keystone of the social brain, is often reported to substantially contribute to body motion processing (Zacks, Swallow, Vettel, & McAvoy, 2006; Morris, Pelphrey, & McCarthy, 2005; Pelphrey, Morris, Michelich, Allison, & McCarthy, 2005; Peuskens, Vanrie, Verfaillie, & Orban, 2005; Vaina & Gross, 2004; Beauchamp, Lee, Haxby, & Martin, 2003; Allison, Puce, & McCarthy, 2000). Magnetoencephalography (MEG) reveals a specific pattern of evoked oscillatory gamma response to a point-light walker that rapidly unfolds in time over the left occipital (100 msec), bilateral parietal (130 msec), and right temporal (170 msec) cortices (Pavlova, Birbaumer, & Sokolov, 2006; Pavlova, Lutzenberger, Sokolov, & Birbaumer, 2004). In healthy adults, electroencephalography (EEG) uncovers negative peaks at latencies of 180–200 msec and 230–360 msec in response to point-light body motion. The later component is associated with the right superior temporal region (Jokisch, Daum, Suchan, & Troje, 2005; Hirai, Fukushima, & Hiraki, 2003). In infants aged 8 months, the averaged negative EEG amplitude in the right hemisphere is greater in response to canonical than to scrambled human motion (Hirai & Hiraki, 2005). While viewing upright as compared with inverted point-light movements of the whole body, infants of this age exhibit larger positive amplitude of the ERPs over the right parietal cortex at a latency of 200–300 msec (Reid, Hoehl, & Striano, 2006). Other brain regions such as the amygdala and the cerebellum also exhibit elevated fMRI response to body motion (Calvo-Merino, Grèzes, Glaser, Passingham, & Haggard, 2006; Ptito, Faubert, Gjedde, & Kupers, 2003; Vaina, Solomon, Chowdhury, Sinha, & Belliveau, 2001; Grossman et al., 2000; Bonda, Petrides, Ostry, & Evans, 1996). Single-cell recording in the macaque monkey, brain imaging in humans, and neuropsychological lesion studies all suggest that visual processing of body motion engages a specialized neural network that differs from processing of other moving configurations (Blake & Shiffrar, 2007; Morris et al., 2005; Jellema, Maassen, & Perrett, 2004; Pelphrey et al., 2003; Puce & Perrett, 2003; Grossman & Blake, 2002). Some components of this network, for example, the right STS, are also activated by Heider-and-Simmel-like animations eliciting visual impression of social interaction (Pavlova, Guerreschi, Lutzenberger, Sokolov, & Krägeloh-Mann, 2008; Schultz, Friston, O'Doherty, Wolpert, & Frith, 2005; Schultz, Imamizu, Kawato, & Frith, 2004; Schultz et al., 2003; Castelli, Happé, Frith, & Frith, 2000).
In the present work, we assessed temporal and topographic changes in neuromagnetic cortical response in adolescents with periventricular leukomalacia (PVL) during detection of camouflaged body motion (Figure 1). PVL is the dominant form of brain injury in individuals who were born premature, and is characterized by gliosis in the white matter and tissue loss with secondary ventricular dilatation (Krägeloh-Mann et al., 1999), thereby affecting the pathways connecting subcortical structures with cortical regions and cortico-cortical connectivity (Skranes et al., 2007). Diffusion tensor imaging (DTI) suggests that PVL might affect the posterior thalamic radiation (Thomas et al., 2005; Hoon et al., 2002) connecting the pulvinar and the lateral geniculate nucleus with the parietal cortex (Behrens et al., 2003).
Our previous work on body motion processing in this patient population shows that: (i) Visual processing of body motion is compromised in adolescents with early periventricular brain damage (Pavlova, Sokolov, Birbaumer, & Krägeloh-Mann, 2006; Pavlova et al., 2005); (ii) These deficiencies are specifically related to the volumetric extent and topography of periventricular lesions (Pavlova, Sokolov, et al., 2006); (iii) Motor experience in human locomotion does not appear to be a necessary prerequisite for the visual analysis of body motion. The visual sensitivity to body motion is not substantially affected by observers' early locomotion restrictions (Pavlova, Staudt, Sokolov, Birbaumer, & Krägeloh-Mann, 2003). By using a one-back repetition paradigm with canonical and spatially scrambled point-light walkers, we demonstrated that, in PVL patients, early (140–170 msec) evoked root-mean-square (RMS) activation over the right parietal lobule in response to biological motion is weaker than in controls (Pavlova, Marconato, et al., 2006). Here, by using a more demanding task to detect biological motion embedded in a simultaneous noise, we address the issue of how periventricular lesions modulate the proper functioning of the large-scale neural network engaged in visual processing of body motion. Adding complex noise to point-light body motion makes the task deliberately more demanding by reducing the sensory evidence available to observers. This helps to uncover fine alterations in time course and topography of cortical MEG activity in response to body motion elicited by disturbances in brain connectivity.
Patients were 11 adolescents (6 girls and 5 boys, age range = 13–16 years) with MRI evidence for PVL. They were born premature between 27 and 33 weeks of gestation. One female patient was excluded from data processing because of additional cortical lesions detectable on her MRI scan. Two patients (1 girl and 1 boy, both with severe PVL) had great difficulties performing the task that resulted in a sensitivity index (d′) equal to or even below zero. The datasets of these patients were excluded from further analysis because the number of correct trials (both hits and correct rejections) submitted for MEG data processing was much lower than in other participants. This left the datasets with eight patients (4 girls and 4 boys, mean ± SD age = 14.75 ± 1.28 years) for the subsequent data analysis. The average volume of PVL in the parieto-occipital region of the left hemisphere was 15.12 ± 10.48 ml, and the volume of the right hemispheric parieto-occipital PVL was 16.54 ± 8.68 ml. The average volume of PVL in the left temporal region was 1.52 ± 1.01 ml, and in the right temporal region it was 1.98 ± 1.41. The average volume of PVL in the left frontal region was 7.59 ± 3.76 ml, and in the right frontal region it was 7.75 ± 3.06. The pairwise interhemispheric differences in PVL volumetric extent in the frontal, parieto-occipital, and temporal regions were not significant. Ten term-born adolescents (4 girls and 6 boys, mean ± SD age = 15 ± 0.94 years) were recruited from the local community. They had MRI scans without any identifiable signs of brain abnormalities, and served as controls. All participants had normal or corrected-to-normal vision and were right-handed. We also controlled for oculomotor dysfunctions that can be observed in PVL patients (Cioni et al., 1997) and may affect performance on visual tasks. Verbal IQ (intelligence quotient) greater than 85 (HAWIK-III, Hamburg-Wechsler-Intelligence-Test-für-Kinder, based on the WISC III, adapted to the German population; Tewes, Rossmann, & Schallberger, 2001) was an inclusion criterion. For patients, verbal IQ scores were in the range of 101–124 (average, 108 ± 7.66), and for controls in the range of 97–127 (average, 117.88 ± 14.83). The groups did not differ in verbal IQ scores (t test, ns). For PVL patients, performance IQ scores were in the range of 73–106 (average, 90.57 ± 11.84), and for controls in the range of 90–106 (average, 100.25 ± 6.25). Although performance IQ scores of PVL patients were lower, there was no significant difference between the groups. The impairments of PVL patients on performance IQ based on visual–perceptual and attentional tasks, and the relationship of these difficulties to visual processing of body motion are reported earlier (Pavlova, Sokolov, Birbaumer, & Krägeloh-Mann, 2008; Pavlova, Lutzenberger, Sokolov, Birbaumer, & Krägeloh-Mann, 2007; Pavlova, Sokolov, & Krägeloh-Mann, 2007; Pavlova, Sokolov, et al., 2006; Pavlova et al., 2005). The outcome of these studies indicates that deficiencies in the visual processing of body motion and in performance on the visual–perceptual tasks constituting performance IQ measured by HAWIK-III are associated with the severity and distinct topography of PVL. All patients attended a mainstream school. The inclusion criteria for all groups of children were also limited by the nature of MEG recording. Children wearing teeth braces could not participate. Neither PVL patients nor controls had a history of psychiatric disorders including autistic spectrum disorders or attention-deficit hyperactivity disorder. Informed written consent was obtained from the participants and their care-providers in accordance with the requirements of the local ethics committee.
Detection Task with Camouflaged Body Motion
Participants were presented with computer-generated point-light configurations created by a modified Cutting's (1978) algorithm. One type of stimulus represented a canonical point-light walker embedded in an array of 33 distracters competing with motions of the walker's dots (Figure 1A). As compared with our earlier psychophysical work (Pavlova, Sokolov, et al., 2006; Pavlova et al., 2003, 2005), the mask level was reduced to 33 (instead of 44) distracting dots to make the task easier and more suitable for MEG recording. The other type of stimulus was a 44-dot mask: 11 dots were added to the walker-absent displays so that their density matched that of the walker-present displays. A canonical point-light walker was composed of 11 dots placed on the joints (ankles, shoulder, etc.) of an invisible human body (Figure 1A). It was seen moving and facing right (sagittal view), with no net translation. A gait cycle was accomplished in 40 frames with a frame duration of 36 msec. The target subtended a visual angle of 9° in height and 6° in width at the most extended point of a gait cycle. A 33-dot distracter consisted of three sets of spatially scrambled dots on the joints of a canonical walker. Within a set, the motion of each dot mimicked the motion of one of the dots defining the point-light target figure. The size, luminance, and phase relations of the dots also remained unchanged. Moving dots were distributed within a region of about 12° in height by 18° in width.
MEG Recording and Data Processing
Each run contained 200 trials with an equal number of walker-present and walker-absent displays. The order of display presentations was randomized. In a yes–no paradigm, by pressing the appropriate key, participants had to detect the presence of a point-light walker. No immediate feedback was given regarding performance. The stimulus appeared for 1 sec on a blank screen with an interstimulus interval that varied randomly between 3.5 and 4.0 sec. To eliminate the influence of motor activity on recorded MEG traces, participants were asked to respond upon the stimulus offset and to avoid responding during the stimulus presentation. A trial was discarded if a participant responded to the stimulus within the stimulus presentation.
A participant was seated in an electromagnetically shielded chamber (Vakuum-Schmelze GmbH, Hanau, Germany). The cortical responses were recorded with the whole-head MEG system (CTF Systems, Vancouver, Canada) composed of 151 hardware first-order magnetic gradiometers distributed with an average distance of 3 cm between sensors. The signals were recorded in DC mode with an antialiasing filter set at 104 Hz and a sampling rate of 312.5 Hz. Recorded epochs lasted from 300 msec before to 1000 msec after the stimulus onset. Both at the beginning and at the end of each recording session, the participant's head position was determined with three localization coils fixed at the nasion and the preauricular sites. Sessions with head movements exceeding 0.5 cm were discarded. All epochs of MEG activity were first automatically and then manually inspected for artifacts. Epochs containing blinks or eye movements (>;±100 μV) were rejected. Detection task performed by participants obligates attention to all types of stimuli and reduces possible attention effects on recorded MEG traces. Each MEG recording session (during presentation of a set of 200 stimuli) lasted 10 to 12 min. The entire experimental session defined by the preparatory period, instruction, familiarization, and MEG recording took about 50 min. For each participant, the artifact-free MEG datasets were averaged separately for correct responses (hits and correct rejections for walker-present and walker-absent displays, respectively). Because we were interested in relatively early brain activation, RMS analysis was performed in equivalent temporal windows of 180–244 and 276–340 msec separately for occipital, parietal, temporal, and frontal regions, and for each hemisphere. These time windows were chosen after visual inspection of RMS curves by three independent MEG experts who were blind in respect to the purposes of the study. Within these time intervals, pronounced alterations of early RMS amplitude were observed. For analysis of visual sensitivity to body motion, the sensitivity index (d′), a standard measure of sensitivity in signal detection theory (Macmillan & Creelman, 2005), was computed for each participant.
In PVL patients, the sensitivity index (d′) to camouflaged body motion was in the range of 0.8–2.84 (mean = 2.29 ± 0.67), and in controls in the range of 1.24–5.05 (mean = 3.13 ± 1.22). As expected, pairwise comparison reveals a significant difference in visual sensitivity between the PVL patients and controls [t(16) = 2.218, p < .041, one-tailed; Figure 2]. No differences were found between PVL patients and controls in cognitive decision criteria (ln β). This indicates that it is a reduction in sensitivity that is responsible for the poorer performance of PVL patients. Participants made very few false alarms, and no difference was found between PVL patients and controls in the false alarm rate (computed as a ratio of the number of false alarms to the total number of stimuli for which this type of error might occur). This indicates that the poorer performance of PVL patients cannot be explained by differences in personality characteristics (e.g., by anxiety). Difficulties in body motion detection by patients were also reflected in response time which was longer for walker-absent than for walker-present displays [mean 566 ± 289 msec, for walker, and 645 ± 261 msec, for mask; t(7) = 3.108, p < .017, one-tailed]. This difference was absent in controls (mean 563 ± 270 msec, for walker, and 591 ± 208, for mask; ns).
The sensitivity to body motion was inversely linked to the volumetric extent of PVL in the parieto-occipital region (Pearson product–moment correlation, r = −.818, p < .02; for both hemispheres together). There was no relationship between the sensitivity to motion and the extent of lesions in the temporal or frontal regions. This result agrees well with our earlier findings obtained with a more complex mask (44 instead of 33 masking dots) and with a different (confidence rating instead of yes/no) psychophysical paradigm (Pavlova, Sokolov, et al., 2006).
Neuromagnetic RMS Response
In the first time window of 180–244 msec, a three-way ANOVA with factors stimulus (walker-present/mask only), group (PVL patients/controls), and cortical region (occipital right, parietal right, temporal right, frontal right, occipital left, parietal left, temporal left, frontal left) was performed on individual RMS values. This analysis revealed a highly significant effect of Region [F(7, 112) = 8.028, p < .0001]. Neither of the main effects of stimulus and group, nor the interactions between factors, were significant. Pairwise comparisons showed that in response to walker-present displays, the RMS amplitude over the right temporal cortex was lower in PVL patients as compared with healthy controls [mean 89.2 ± 8.85 fT (femtotesla) in PVL patients; and 123 ± 18.56 fT in controls, t(16) = 2.13, p < .049; Figure 3A, C]. By contrast, the RMS amplitude did not differ between PVL patients and controls in response to the control mask-only display (mean 85.1 ± 10.52 fT in PVL patients, and 98.1 ± 10.72 fT in controls, t test, ns; Figure 3B and C). This alteration in amplitude, therefore, was specific to body motion processing. In controls, the RMS amplitude over the right temporal cortex was greater to walker-present than to mask-only displays [t(9) = 2.626, p < .027], but did not differ in PVL patients.
In this first time window, the RMS amplitude in controls was also greater in response to walker-present as compared with mask-only displays over the right occipital [t(9) = 2.262, p < .05] and over the left frontal region [t(9) = 3.489, p < .006]. However, it was not greater than in PVL patients. In patients, there were no differences in MEG response between walker-present and mask displays over these regions. The lack of differences in RMS amplitude between walker-present and mask-only displays in patients over the left frontal cortex may be explained by an increased cortical activation to mask-only displays. The left frontal cortex is known to be involved in decision making (e.g., Heekeren, Marrett, Ruff, Bandettini, & Ungerleider, 2006), and PVL patients exhibit difficulties in making correct rejections. These difficulties are reflected at the behavioral level: In patients, but not in controls, response time for mask-only displays was longer than for walker-present displays (see Behavioral findings above). In this time window, there were no other significant differences in RMS amplitude.
At latencies of 276–340 msec, a three-way ANOVA with factors stimulus (walker-present/mask only), group (PVL patients/controls), and cortical region (occipital right, occipital left, parietal right, parietal left, temporal right, temporal left, frontal right, frontal left) was performed on individual RMS values. This analysis revealed a highly significant effect of region [F(7, 112) = 8.482, p < .0001]. The main effect of stimulus also was significant [F(1, 16) = 5.194, p < .036]. Neither the main effect of group, nor interactions between factors were significant. In PVL patients, RMS amplitude over the right temporal region to walker-present displays was greater than to mask-only displays [t(7) = 3.707, p < .007]. However, this amplitude in PVL patients did not differ from the amplitude in healthy controls. In controls, we did not find any differences in RMS amplitude between walker-present and walker-absent displays at these latencies. In controls, but not in PVL patients, RMS amplitude over the left frontal lobe in response to walker-present displays was greater than to mask-only displays [t(9) = 2.433, p < .037]. However, it was not greater than in PVL patients in response to both walker-present and walker-absent displays.
In the time window of 276–340 msec, RMS amplitude over the right frontal cortex in response to point-light walker was lower in PVL patients than in healthy controls (mean 49.1 ± 9.17 fT; and mean 79.6 ± 13.3 fT; p < .031, respectively; Figure 4A and C). By contrast, the amplitude did not differ between PVL patients and controls in response to the control walker-absent displays. This alteration in RMS amplitude was specific for human walking (Figure 4B and C). In controls, the difference in RMS amplitude between walker-present and mask-only displays approached significance [t(9) = 2.063, p < .069], but did not differ in PVL patients.
Relation of Visual Sensitivity to RMS Amplitude
Most striking, in the time window of 180–244 msec, in healthy adolescents, the visual sensitivity to body motion (d′) was negatively linked to the right temporal RMS amplitude (Pearson product–moment correlation, r = −.778, p < .01; Figure 5A). In healthy adolescents, the sensitivity to body motion dropped with increasing RMS amplitude over the right temporal cortex. In PVL patients, there was no link between the sensitivity to body motion and RMS amplitude over this region (r = .305, ns; Figure 5B).
This work is an investigation of cortical neuromagnetic response to body motion in patients with congenital periventricular brain lesions that interrupt structural brain connectivity. From the very beginning in life, such interruption might prevent the proper functioning of the brain networks involved in action observation. By combining visual psychophysics with whole-head MEG recording, we identify two main effects of PVL on cortical activity. Specific for body motion processing, alterations of neuromagnetic response in PVL patients occur over the right temporal cortex at a latency of 180–244 msec, and over the right frontal cortex at a latency of 276–344 msec (Figures 3 and 4).
The early (180–244 msec) neuromagnetic RMS response to camouflaged point-light body motion over the right temporal region is weaker in PVL patients as compared to healthy controls, whereas right temporal response to the control walker-absent displays does not differ between PVL patients and controls. This indicates that cortical MEG response to body motion over the right temporal region is stimulus-specifically modulated by early periventricular lesions. This alteration of the MEG response is difficult to explain by possible changes in gray matter distribution reported in adolescents who were born premature (e.g., Nosarti et al., 2008), because reduced neuromagnetic activity in PVL patients occurs in response solely to body motion. Otherwise, the RMS amplitude in patients would be reduced not only in response to walker-present but also to control mask-only displays. This was not the case in our study.
The stimulus-specific alteration of the neuromagnetic response over the right temporal region in PVL patients is of importance because this area is not only involved in the visual analysis of body motion, but it is also known as the keystone of the social brain (e.g., Tankersley, Stowe, & Huettel, 2007; Allison et al., 2000). This region is reciprocally connected to the orbito-frontal cortex and the amygdala (Adolphs, 2003). Most recent work indicates that healthy perceivers easily reveal information about socially relevant attributes (intentions, emotions, and dispositions) through body motion (Atkinson, Tunstall, & Dittrich, 2007; de Gelder, 2006; Clarke, Bradshaw, Field, Hampson, & Rose, 2005; Atkinson, Dittrich, Gemmel, & Young, 2004; Grèzes, Frith, & Passingham, 2004; Heberlein, Adolphs, Tranel, & Damasio, 2004). It has been shown, for example, that detection of camouflaged human locomotion is modulated by the emotional content of the gait, with the highest visual sensitivity to angry point-light walking (Chouchourelou, Toshihiko, Harber, & Shiffrar, 2006). The right STS plays a crucial role in this ability (e.g., Blake & Shiffrar, 2007; Pelphrey, Morris, & McCarthy, 2004; Pelphrey et al., 2003). For example, an increased fMRI response is observed to fearful body motions as compared to emotionally neutral motions in the right temporal pole and STS (Grèzes, Pichon, & de Gelder, 2007). These data suggest that PVL patients might be impaired not only on action observation but also on some aspects of social cognition revealed through body motion. In accord with this assumption, we have recently found that PVL patients exhibit some difficulties in perception and understanding of others' actions, and the extent of PVL over the right temporal region serves as the best predictor of this impairment (Pavlova, Sokolov, et al., 2008).
Most intriguing, in healthy adolescents, the RMS response over the right temporal cortex is negatively linked to the visual sensitivity to body motion. At first glance, this inverse linkage appears paradoxical because, in healthy adolescents, not only the sensitivity to biological motion is higher but also the RMS amplitude over the right temporal cortex is greater than in PVL patients. An earlier fMRI study reported that in healthy adults, after training to detect camouflaged human locomotion, not only the visual sensitivity to biological motion but also the BOLD signal is elevated (Grossman, Blake, & Kim, 2004). The magnitude of enhancement in the fMRI response is positively correlated with an improvement in the visual sensitivity. Improvement with practice, however, can be accompanied by decreased signal strength in the responsive brain regions (Raichle et al., 1994). The negative link between the RMS amplitude over the right temporal region and the sensitivity might be explained as follows. In both healthy adolescents and PVL patients, for reaching high performance on the detection task, some increase in cortical activity over the right temporal cortex is required. When, however, with perceptual learning the detection task becomes easier, or if from the very beginning it is not so demanding to an individual observer, high performance level does not require a strong cortical response. In PVL patients, however, presumably because of reduced brain connectivity, cortical activity over the right temporal region does not reach the level that is required for optimal performance on the human locomotion detection task. The strong link between the behavioral measures of performance and the cortical response over the right temporal cortex provides direct support for the particular significance of this cortical area for visual processing of body motion.
In PVL patients, at later latencies of 276–340 msec, the neuromagnetic response to human walking over the right temporal cortex was greater than to mask-only displays. However, this response did not differ from that in healthy adolescents. In controls, there was no difference in RMS amplitude between walker-present and mask-only displays. In earlier work, we did show that oscillatory MEG response to camouflaged human locomotion (26.5 Hz) peaked at a latency of 170 msec over the right parieto-temporal cortex, whereas enhancement in PVL patients occurred later, at a latency of 290 msec over the left temporal cortex (Pavlova, Lutzenberger, et al., 2007). We explained this delayed latency of the evoked oscillatory MEG response by alterations in association fibers, which are longer, thinner, and less numerous in PVL patients (Skranes et al., 2007; Thomas et al., 2005; Hoon et al., 2002). However, the difference in the neuromagnetic response to walker-present and mask-only displays, which was found in PVL patients in the second time window, cannot be considered as a shift in cortical activation that is specific for human locomotion because the RMS amplitude in response to body motion in PVL patients did not differ from controls.
The other essential finding is that specific alterations in the RMS amplitude in response to human locomotion in PVL patients occur over the right frontal cortex. Several possibilities might account for this alteration. One of them is that the demanding task to detect a point-light walker embedded in an array of moving dots challenges attentional resources. It has been suggested that PVL patients exhibit behavioral visual attentional deficits (Skranes et al., 2007; Krägeloh-Mann et al., 1999), and these deficits might be reflected in reduced cortical activity over the frontal regions. These regions are implicated in the functioning of attentional networks (e.g., Pessoa, Kastner, & Ungerleider, 2003). The other possibility is that changes in the cortical response are connected to deficient visual working memory in PVL patients because the right frontal cortex is involved in working memory networks (e.g., Crone, Wendelken, Donohue, van Leijenhorst, & Bunge, 2006). It appears difficult to dissociate these two possibilities because visual working memory and attention appear to share a common neural substrate (Mayer et al., 2007). In both cases, however, frontal activation would be reduced in response not only to walker-present but also to control mask-only displays. This was not observed in our study. Instead, decrease in the RMS amplitude in PVL patients was specific to the body motion displays because cortical response to control mask-only displays did not differ between PVL patients and controls. The most plausible explanation, therefore, is that periventricular lesions affect the functioning of the distributed network specialized for body motion processing, and the right frontal cortex is a part of this network. This finding dovetails with previous fMRI data (Saygin, Wilson, Hagler, Bates, & Sereno, 2004) showing increases in the BOLD signal, specific to body movement, over the frontal regions. In accord with this, voxel-based lesion analysis in unilateral stroke patients reveals that not only superior temporal but also frontal areas are of importance for body motion processing (Saygin, 2007). Frontal regions are a part of the mirror neuron system, which has been considered as a core for understanding the actions and intentions of others (Iacoboni & Dapretto, 2006). This system is thought to bridge the gap between the physical self and social perception through motor-simulation mechanisms (Uddin, Iacoboni, Lange, & Keenan, 2007). Specific engagement of the frontal regions in visual processing of body motion found in our study might indicate that these cortical regions are also of importance for social perception through body motion and understanding of others' actions. In accord with this notion, right frontal fMRI activation, along with activation of the right temporal pole and the STS, is greater in response to fearful as compared with neutral whole-body motions (Grèzes et al., 2007). The reduced cortical MEG response to body motion over the right frontal cortex in PVL patients, therefore, might reflect possible deficits in social cognition. High-functioning autistic patients, for example, show no specific fMRI activity in the frontal regions, and brain activity over these regions is inversely related to the severity of impairments in the social domain (Dapretto et al., 2006). A dysfunctional mirror neuron system may underlie deficits in social perception (Williams et al., 2006).
In summary, the present data shed light on the role of brain connectivity for visual processing of body motion. We show that disturbances in structural brain connectivity caused by periventricular brain damage lead not only to reduced visual sensitivity to camouflaged body motion but also to alterations in neuromagnetic RMS response over the temporal and frontal cortices of the right hemisphere. These changes in cortical activity are absent for control displays that do not contain human walking. The visual sensitivity to body motion in controls, but not in PVL patients, is inversely related to the increase in the right temporal RMS amplitude. The strong link between behavioral measures of performance and the cortical response over the right temporal cortex highlights the significance of this area for visual processing of human motion. Overall, the findings indicate that disturbances in brain connectivity, in particular, with the right temporal and frontal cortices cause disintegration of neural networks engaged in the visual processing of body motion.
We thank the participants, PVL patients and controls, their family members, and care providers for kind cooperation. We also thank Jürgen Dax for technical assistance, John C. Baird for valuable advice on the manuscript, and three anonymous reviewers for their helpful comments. This work was supported by the University of Tübingen Medical School (fortüne-Program 1576-0-0 and 1757-0-0 to MP) and, in part, by the Deutsche Forschungsgemeinschaft (DFG, KR 1316/5-2). Christel Bidet-Ildei was a visiting PhD student in M. P.'s laboratory. Her stay was supported by the “Aires Culturelles” Program of the French Research Ministry.
Reprint requests should be sent to Marina Pavlova, Developmental Cognitive and Social Neuroscience Unit, Department of Paediatric Neurology and Child Development, Children's Hospital, University of Tübingen Medical School, Hoppe-Seyler-Str. 1, D-72076, Tübingen, Germany, or via e-mail: firstname.lastname@example.org.