Parietal cortex contributes to body representations by integrating visual and somatosensory inputs. Because mirror neurons in ventral premotor and parietal cortices represent visual images of others' actions on the intrinsic motor representation of the self, this matching system may play important roles in recognizing actions performed by others. However, where and how the brain represents others' bodies and correlates self and other body representations remain unclear. We expected that a population of visuotactile neurons in simian parietal cortex would represent not only own but others' body parts. We first searched for parietal visuotactile bimodal neurons in the ventral intraparietal area and area 7b of monkeys, and then examined the activity of these neurons while monkeys were observing visual or tactile stimuli placed on the experimenter's body parts. Some bimodal neurons with receptive fields (RFs) anchored on the monkey's body exhibited visual responses matched to corresponding body parts of the experimenter, and visual RFs near that body part existed in the peripersonal space within approximately 30 cm from the body surface. These findings suggest that the brain could use self representation as a reference for perception of others' body parts in parietal cortex. These neurons may contribute to spatial matching between the bodies of the self and others in both action recognition and imitation.
The parietal lobe is believed to play a pivotal role in body representation (Jeannerod, 2003). Neuropsychological studies have revealed that damage to right parietal cortex induces anosognosia or asomatognosia. Anosognosic patients exhibit the inability to recognize the presence of motor and sensory defects, frequently concurrent with asomatognosia, which involves impairment of bodily consciousness (Orfei et al., 2007). Multimodal sensory integration, particularly visuotactile integration, is of great importance to this consciousness of one's own body (Thomas, Press, & Haggard, 2006; Beschin & Robertson, 1997).
In monkey experiments, neurons in some areas of parietal cortex have been found to exhibit visuotactile bimodal properties (Murata & Ishida, 2007; Colby, 1998). Bimodal neurons in the fundus of the intraparietal sulcus (IPS) in the ventral intraparietal area (VIP) exhibit tactile receptive fields (RFs) located on the face or head and visual RFs in a location congruent with the tactile RF (Avillac, Deneve, Olivier, Pouget, & Duhamel, 2005; Duhamel, Colby, & Goldberg, 1998; Colby, Duhamel, & Goldberg, 1993). These neurons are thus closely linked to body parts involving peripersonal space. Similar neuronal activity demonstrative of face and/or limb representation was found in the rostral inferior parietal lobe convexity (area 7b), which has anatomical connections to the VIP (Rozzi et al., 2006; Hyvarinen, 1981; Leinonen, Hyvarinen, Nyman, & Linnankoski, 1979; Leinonen & Nyman, 1979).
The discovery of mirror neurons in ventral premotor cortex (F5) and inferior parietal cortex (7b) in the monkey brain provided the first convincing physiological evidence for direct matching between visual images of others' actions and one's own motor representations in a single neuron (Rizzolatti & Craighero, 2004). The functional properties of mirror neurons also strongly support the coexistence of self and others in the same neuronal correlate, contributing to social cognitive function, such as recognition of others' actions, imitation, and communication (Jackson & Decety, 2004; Rizzolatti, Fogassi, & Gallese, 2001). This implies that the bodies of others are represented in the brain along with one's own body. Recently, human imaging studies have revealed that the somatosensory area in one's own brain is activated when observing another person being touched (Blakemore, Bristow, Bird, Frith, & Ward, 2005; Keysers et al., 2004). This suggests that one's own body representation is available for perception of others' body parts via integration of vision and somatosensory maps.
Although mirror neurons are assumed to form the basis of social cognitive functions including action recognition and imitation, how the brain can appropriately map body parts of other individuals during observation of their behavior remains unclear. We expected that others' bodies would be mapped to one's own body representation in the bimodal area of parietal cortex. The present study tested this hypothesis by focusing on visuotactile bimodal areas in the VIP and area 7b in monkeys. The original aim of this study was to determine whether these neurons represent not only one's own body parts but also others' body parts on the map of one's own body.
Three male Japanese monkeys (Macaca fuscata) weighing 5.0 to 8.0 kg were used in this study. We recorded from both hemispheres in one monkey (M18) but only from the left (M17) or right (M22) hemisphere in the other two monkeys. All surgical and experimental protocols were approved by the Animal Care and Use Committee of Kinki University and were in accordance with the Guidelines for Proper Conduct of Animal Experiments of the Science Council of Japan (2006).
After each monkey was familiarized with the laboratory environment and experimenters, the first surgery was performed to enable head fixation under general anesthesia initially induced using ketamine hydrochloride (Ketalar; 5 mg/kg, im) and xylazine hydrochloride (2 mg/kg, im) with atropine sulfate (0.01 mg/kg, im) and maintained using sodium pentobarbital (mean, 15 mg/kg, iv, once every hour). Each monkey was positioned in a stereotaxic apparatus. Small stainless steel bolts were implanted in the skull to allow attachment of a halo ring for head fixation. Postoperatively, animals were administered antibiotics intramuscularly for a period of one week to minimize the possibility of infection. After surgery, each monkey was trained to sit quietly in a primate chair with the head fixed while observing different types of visual stimuli and receiving application of somatic stimuli on various body parts by experimenters. Under the above-described anesthesia, M18 and M22 each underwent a second surgery to implant a magnetic search coil allowing eye position to be monitored (Judge, Richmond, & Chu, 1980).
Following training, an opening of about 2 cm in diameter was made in the skull under the above-described anesthesia using a trephine over the anterior part of inferior parietal cortex. A cylindrical stainless-steel recording chamber (diameter = 20 mm) was implanted under stereotaxic guidance. The center of the chamber and angle in stereotaxic coordinates were as follows: in M17 over the left hemisphere (anterior [A] = 3.0 mm; lateral [L] = 21.0 mm; angle = 50°); in M18 over both the right hemisphere (A = 4.0 mm; L = 19.0 mm; angle = 50°) and left hemisphere (A = 6.0 mm; L = 15.0 mm; angle = 40°); and in M22 over the right hemisphere (A = 5.0 mm; L = 19.5 mm; angle = 50°). Each chamber was fixed to the bone using dental acrylic, filled with saline, and then closed with a plastic cap.
Recording Procedure and Recording Sites
Single-unit recording was performed extracellularly using varnish-insulated tungsten microelectrodes (impedance 1.5–5.0 MΩ at 1 kHz; FHC, USA) advanced into the cortex obliquely (45° or 50° from vertical in the frontal plane) through the dura mater using a hydraulic microdrive manipulator (MO-95, Narishige, Japan).
We searched for visuotactile bimodal neurons in the anterior part of the fundus of the IPS and in the inferior parietal lobule. Before recording from these areas, we mapped the hand and face area in primary somatosensory cortex (SI) to estimate the position of the IPS for penetrations. Neuronal response properties were assessed from the cortical surface when each electrode passed into the lateral or medial bank of the IPS and reached the fundus.
Physiological boundaries of each subregion in the IPS were identified on the basis of previously described response properties and depth within the sulcus (Murata, Gallese, Luppino, Kaseda, & Sakata, 2000; Colby et al., 1993; Andersen, Bracewell, Barash, Gnadt, & Fogassi, 1990; Sakata, Takaoka, Kawarasaki, & Shibutani, 1973). For example, neurons in the anterior part of 7b exhibiting visuotactile bimodal responses in their RFs are principally located on the trunk or bilateral forearms and around the mouth (Hyvarinen, 1981; Leinonen et al., 1979; Leinonen & Nyman, 1979). These neurons were recorded from the lateral convexity within 2 to 3 mm distance from the IPS. Neurons of the VIP in the fundus of the IPS also exhibit bimodal visuotactile responses, in which visual RFs are generally restricted (10–30°), central or peripheral and often bilateral, whereas tactile RFs are mostly located in the head and face regions. These neurons were recorded within the anterior part of the IPS at a depth of more than 4 mm from the cortical surface. Overall, bimodal neurons in both areas often exhibit correspondence in localization of RFs and directional selectivity for both modalities. These characteristics are distinguishable from the functional properties of adjacent areas such as the anterior intraparietal area (AIP), lateral intraparietal area (LIP), medial intraparietal area (MIP), and area 5. For example, neurons of the AIP and LIP do not respond to tactile stimuli. Neurons of the MIP and the convexity of area 5 mainly prefer joint rotation of the hand or forearm, although some neurons respond to visual stimuli and exhibit bimodal visuotactile properties.
Behavioral Testing of Visuotactile RFs in Monkeys
We tested whether each neuron exhibited visuotactile bimodal responses. Single neurons underwent both visual and somatosensory stimulation in an order that was randomized across cells. Types of testing are described below.
Neural activity was recorded while passive somatic stimuli were applied to the body with the monkey's eyes closed by the experimenter's hand. The first somatosensory stimuli consisted of hair deflections by touch, light pressure to stimulate subcutaneous tissues, and slow and fast passive rotations of limb joints. In addition, we attempted to determine whether neurons displayed directional selectivity. Somatosensory RFs were first plotted by repeated presentation of the most effective stimuli on the face and body, and we then carefully defined the borders of RFs on the body parts. To determine the laterality of RFs, we applied preferred stimuli on both contralateral and ipsilateral sides of the same body parts. In cases in which a tactile RF was located bilaterally or at the center of the body part, we quantitatively compared responses to those of different body parts.
First, to exclude neuronal activity related to visual fixation or eye movement, an experimenter presented a piece of food held in his hand or moved the food radially from a different position and at different speeds in front of the monkey. Eye movements were monitored on an oscilloscope with a magnetic search coil in M18 and M22. We discarded neurons responding to the eye movement control or visual fixation.
For visual stimulation, a 20-cm steel rod with a 3-cm diameter white sphere on the end or the experimenter's waving hand was used. To determine the borders and directional selectivity of visual RFs related to the monkey, stimuli were presented close to different body parts with different directions and at different speeds by the experimenter, and the borders of visual RFs near various body parts were then carefully defined.
To determine the extension of RFs in depth, all visual responses were tested for all bimodal neurons when stimuli were presented at various distances from immediately around the monkey's body part to 120 cm distant from it. If neurons exhibited a stable response at each distance from the monkey up to 120 cm, they were discarded because their RFs might involve the extrapersonal space of the monkey. Finally, to determine the laterality of visual RFs, we applied preferred stimuli close to both sides of the same body parts. If a visual RF was located bilaterally or in the center of a body part, we compared responses to those of a different body part, and then identified responses both within and outside the RF.
Mapping of Visual RFs Near the Experimenter's Body Parts
After mapping bimodal RFs on the monkey, the experimenter faced the monkey and sat in a chair at a constant distance of 120 cm to avoid entry of the experimenter's body into the visual RFs in the peripersonal space of the monkey.
First, we examined neuronal responses of bimodal neurons while monkeys were observing visual stimuli applied to the experimenter's body parts. The experimenter who faced the monkey presented a moving stick or waved his hand near his own body parts, and another experimenter presented a moving stick or his own hand in the peripersonal space of the original experimenter. In the same fashion as for mapping of RFs on monkeys, we carefully defined the borders of visual RFs by repeated presentation of the visual stimulus close to various body parts. The stimulus was then applied to each side of the experimenter's body to determine the laterality of visual RFs. We also examined whether neurons could respond during observation of touching of the experimenter's body parts. In some cases, we also tested visual responses to an object mimicking body parts such as a rubber glove for the hand. We presented this object in front of the monkey, and then applied visual stimuli close to this object.
Furthermore, to ensure coding of peripersonal space near both the monkey and the experimenter, another experimenter presented visual stimuli at various distances between the monkey and the original experimenter. If neurons exhibited stable responses to visual stimuli at different positions between the monkey and the original experimenter (from over 30 cm up to 90 cm from the monkey), they were excluded from further analysis, as we could not ensure that their RFs involved only the peripersonal space of the experimenter. In some neurons, to test position invariance of visual RFs near the experimenter's body, visual responses at various distances were recorded when the experimenter shifted position to the left by 35° or to the right by 35°, maintaining a 120-cm distance from the monkey.
After identifying both bimodal RFs on the monkeys' body parts and visual RFs close to the experimenter's body parts in the behavioral tests, we recorded the neuronal responses to stimuli inside RFs and animal behaviors using four small CCD cameras (29 frames/sec from different directions: top, front, both sides) on the same videotape with a four-part screen. Neuronal spikes were isolated and converted to electrical pulses, and then fed into the audio channel of the videotape for off-line analysis. To identify behavioral events, videotapes were edited using a personal computer with DV Storm software (Canopus, Japan). All spike data and behavioral events were analyzed with CED1401plus and Spike2 software (Cambridge Electronic Design, UK) in frame-by-frame fashion to examine correlations between them. The response to each stimulus was defined as the mean firing rate (spikes/sec) in the period from the beginning of stimulation to the end of stimulation. The rest condition comprised background activity before or after stimulation.
For each neuron, mean responses to each stimulus (spikes/s) were calculated for statistical analysis using ANOVA followed by Bonferroni post hoc analyses. In all analyses, values of p < .05 were considered significant.
Bimodal responses were determined not only by observation during behavioral tests but also based on the results of the following statistical comparisons. Neurons were statistically defined as exhibiting both visual and tactile responses when both types of responses were significantly higher than the activity recorded during rest condition. Locations of bimodal RFs were defined by behavioral tests. For neurons exhibiting visual RFs near the experimenter's body in these tests, we also attempted to statistically determine whether neurons were significantly activated by stimuli in those RFs compared with the rest condition. Lateralities of RFs anchored near the experimenter's body parts were usually determined by behavioral tests, and were then tested statistically in many cases.
To assess the visuospatial properties of recorded neurons, stimuli were presented at four spatial locations: (1) within 30 cm of the monkey's body surface (peripersonal space of monkey); (2) 30–60 cm; (3) 60–90 cm; and (4) 90–120 cm from the monkey's body surface. A visual response in the peripersonal space of the monkey was considered present when the response in the given location (1) was significantly higher than in other locations with the experimenter not facing the monkey. Furthermore, a neuron was considered to exhibit visual responses in the experimenter's peripersonal space when the response in the given location (Location 4) was significantly higher than in Locations 2 and 3, with the experimenter facing the monkey. When it was possible to determine whether the neuron exhibited positional invariance of visual RFs near the experimenter's body, visual responses at each distance were statistically compared when the experimenter shifted position 35° to the left or right, while maintaining a distance of 120 cm from the monkey.
Mapping of Visuotactile RFs on Monkeys
We recorded 541 parietal visuotactile bimodal neurons from the three macaque monkeys (133 in M17, 301 in M18, 109 in M22). In agreement with previous reports (Duhamel et al., 1998; Leinonen et al., 1979), we found that bimodal neurons had visual RFs located near tactile ones. Tactile RFs were mostly located on a single body part (n = 451, 83%), such as the face (n = 294), forearm, hand, digit (n = 131; visual–tactile, n = 68; visual–joint, n = 63), or trunk or leg (n = 26). Another 13% (n = 69) were located on multiple body parts including the hemibody, with the majority on the face, upper limb, or trunk (n = 50). In 4% of neurons (21 neurons), the visual RFs were located on a body part different from that of tactile RFs. Most RFs were located on the contralateral side of the body (66%), although some were also found bilaterally or centrally (25%) or ipsilaterally (4%). The remaining RFs could not be placed in any of these categories.
Responses of Body-matching Bimodal Neurons
Forty-four (8%) tested neurons were excluded from analysis because they also responded to visual stimuli presented far (200 cm) from the monkey. On off-line video analysis, we identified 57 neurons that responded to stimuli presented on an experimenter's body part. However, nine of these neurons were excluded because responses were modulated by eye movement or fixation on food presentation. Finally, we identified 48 bimodal neurons that, as qualitatively tested, exhibited both bimodal RFs on the monkey's body and visual RFs close to the experimenter's body. We refer to these neurons as “body-matching bimodal neurons.” When visual stimuli were presented at the same distance from the monkey (120 cm) but in the absence of the experimenter, the responses were weak or disappeared altogether. Body-matching neurons usually responded to any visual stimuli in the RFs close to the experimenter's body. Among them, 23 neurons (48%) could be statistically analyzed and were found to exhibit significant activity during visual stimulation close to the experimenter's body parts, and the locations of RFs near the monkey and the experimenter were examined in detail. Based on the type of spatial congruency of RFs between the monkey and experimenter, we classified these neurons into three types: “mirror-image matching,” “bilateral or central matching,” or “anatomical image matching.”
Figure 1 shows an example of a mirror-matching neuron. This neuron exhibited a tactile RF on the left side of the nose and cheek and a visual RF located close to the same parts of the face (Figure 1A and B) of the monkey. Responses to the left cheek were significantly higher than those to the right cheek in both modalities (one-way ANOVA followed by Bonferroni post hoc test, p < .0001). On testing of extension of the visual RF in depth, visual stimuli within 30 cm from the monkey's left cheek were found to evoke the strongest discharge (p < .0001; Figure 1C). With the stimulus 30 cm away from the monkey, this neuron exhibited less reaction to the stimuli, showing that the visual RF did not extend more than 60 cm from the monkey. In this neuron, the preferred direction of both visual and tactile stimuli was the same, from right to left. With regard to preference in depth movement of visual stimuli, no differences were observed in activity between approaching and receding.
Furthermore, when the experimenter faced the monkey at a distance of 120 cm, visual responses could be evoked by moving stimuli near the experimenter's cheek. The visual RF around the experimenter's face was in a mirror-image position with respect to the monkey's visual RF. The neuron responded to visual stimuli close to the right face of the experimenter, but was less active for the left face (p < .0001; Figure 1D). In addition, the neuron exhibited strong directional selectivity. The neuron preferred rightward direction from the monkey's perspective. Visual responses close to the right cheek of the experimenter were also significantly higher than those to stimuli presented in the same position when the experimenter did not face the monkey (p < .0001; Figure 1C in the “90–120 cm” condition).
Neurons of this class had RFs located bilaterally on body parts or on the center of the body. The neuron in Figure 2 exhibited bimodal RFs on the nose in the center of the face of the monkey (Figure 2A and B). Responses to the center of the face were also significantly stronger than those to the chest of the monkey in both modalities (p < .0001). Visual responses increased as the stimulus neared the face (Figure 2C). Visual stimuli presented within 60 cm of the monkey's nose evoked significantly stronger discharges than those in other conditions (p < .0001). In this neuron, the preferred direction of stimulation was rightward. With regard to preference in depth movement of visual stimuli, no differences were noted in activity between approaching and receding.
The location of the visual RF near the experimenter's face corresponded to the RFs on the monkey, with responses to visual stimuli near the center of the experimenter's face (Figure 2D). Visual responses near the experimenter's face were significantly stronger than those near the chest (p < .0001). The visual RF near the monkey's face was mainly in the peripersonal space, with no response when visual stimuli were at 120 cm distant from the monkey when the experimenter was absent (p < .0001; Figure 2C in the “90–120 cm” condition). This neuron exhibited strong directional selectivity in the same fashion as the monkey (i.e., rightward direction from the perspective of the monkey).
Anatomical Image Matching
Figure 3 shows an example of an anatomical image matching neuron. This neuron exhibited anatomical correspondence of RFs on the forearm of the monkey and the experimenter. The neuron represented bimodal RFs on the left forearm of the monkey (Figure 3A and B). The tactile response on the left forearm was significantly stronger than that on the right arm of the monkey and that at rest condition (p < .0001; Figure 3A). This neuron exhibited directional selectivity, preferring motion of stimuli from proximal to distal in both modalities. When visual stimuli were presented from proximal to distal by the experimenter near the surface of each monkey's arm, visual responses were found in the space only within 30 cm from the left forearm, and not near the right forearm (p < .0001; Figure 3B).
When the experimenter faced the monkey, this neuron exhibited anatomical correspondence to the RF near the left forearm (Figure 3C). Visual responses were found only in the space very close to the experimenter's left arm (p < .0001). Most body-matching neurons exhibited equivalent visual responses when the stimulation was applied by the original experimenter or by another individual. However, this special neuron exhibited its best response when the monkey was observing the experimenter stroking his own left forearm from proximal to distal himself, but when the experimenter presented a moving stick near his own forearm with the same direction, it did not respond. In this neuron, we also tested response to a left-hand rubber glove. We presented this object in front of the monkey, and the experimenter then applied a stroking hand movement on this object. Responses to the actual left arm of the experimenter were significantly stronger than those to the rubber glove (p < .0001).
Locations of RFs on the Body of the Monkey and the Experimenter
Figure 4 shows locations of tactile and visual RFs for 23 body-matching neurons. In total, 10 neurons exhibited mirror-image matching of RFs between the monkey and experimenter. Ten neurons exhibited bilateral or central matching of RFs. Three neurons exhibited anatomical image matching of RFs. The RFs of the 17 neurons in Figure 4A– C were located on the face (Figure 4A: n = 9; Figure 4B: n = 7; Figure 4C: n = 1). Neurons in Figure 4D– F represented RFs on the limbs and digits. The RFs of the neuron in Figure 4D were lateralized in the mirror image, but were incongruent with body parts, appearing for the monkey's left forearm and little finger and the experimenter's finger. The neurons in Figure 4E represent RFs on bilateral body parts of the both the monkey and the experimenter. Two neurons displayed RFs with bilateral matching on the digits and underarm of both the monkey and the experimenter. The other neuron did not exhibit complete matching in body parts, exhibiting responses to the monkey's jaw, forearm, and hand, and both of the experimenter's hands. The RFs of both neurons in Figure 4F were mainly located on the forearm of both the monkey and the experimenter, exhibiting anatomical image matching. The RFs of one neuron extended to the little finger of both the monkey and the experimenter (Figure 4F, bottom).
We estimated recording site from the depth of penetration of the recorded neurons (Figure 5). Most penetrations began from the lateral convexity or vicinity of the crown to the anterior part of the IPS (for the center of the chamber and angle in stereotaxic coordinates, see Methods). More than 75% of all body-matching neurons were recorded at a depth of 4000–11,000 μm (Figure 5, left). The majority of these neurons had RFs located on the monkey's face (Figure 5, right). Because of this location and the recording depth, it is very likely that these neurons were recorded from the VIP area. Conversely, neurons that were recorded more superficially (at a depth of less than 2000 μm) responded, with one exception, to stimulation of digits, hands, and forearms. It is likely that they were in area 7b.
For 23 body-matching neurons, we also examined neural correlates of activity during stimulation of the monkey and the experimenter (Figure 6). Visual responses near the experimenter's body parts were significantly correlated with visuotactile responses by the monkeys (tactile stimulation, r = .55; visual stimulation, r = .64; p = .006, p = .001, respectively), although activity tended to be weaker for the experimenter's body than for the monkey's (monkey tactile, mean = 37.37 spike/sec; monkey visual, mean = 37.29 spikes/sec; experimenter visual, mean = 27.05 spikes/sec, paired t test: p = .001, p = .005, respectively).
Position-invariance of Peripersonal Space Coding
On behavioral tests, the visual RFs of all 48 tested neurons were located in the peripersonal space within 30 cm of the monkey's body, and visual responses could be evoked by moving stimuli in the peripersonal space of the experimenter when he faced the monkey. To ensure peripersonal space coding of body-matching neurons in both the monkey and the experimenter, seven neurons were examined for extension of visual RFs near the monkey and the experimenter when another experimenter presented visual stimuli in the space between the monkey and the original experimenter. In all cases, visual responses were found when stimuli were within 30 cm of the monkey's body part or the experimenter's body part, but was less strong when stimuli were between these areas (p < .01 for both “30–60 cm” and “60–90 cm” from the monkey).
In two of these seven neurons, we also examined whether activity was modified depending on the position of the experimenter. For example, the mirror-matching neuron in Figure 7 was tested for visual RF when the experimenter shifted position from center to 35° to the left or right while maintaining a distance of 120 cm from the monkey. A two-way ANOVA with two factors of position (three levels: left 35°, center, right 35°) and distance (four levels) revealed a significant interaction (p < .05). In cases in which the experimenter was in the center position, visual response within 30 cm of the monkey's body part and of the experimenter's body part were significantly stronger than in intermediate regions. Even if the experimenter shifted positions from “center” to “left” or “right,” this response profile remained unchanged at each position (for all three positions of experimenter: 30 cm and 120 cm > 60 cm and 90 cm, p < .0001), although some differences were noted at 60 or 90 cm depending on the position of the experimenter. In the center position, visual response <30 cm from the monkey was significantly stronger than at 90–120 cm (p < .02). In addition, at 60–90 cm, the response at “left 35°” position was significantly stronger than that in “center” position (p < .01).
The neuron in Figure 8 (same neuron as in Figure 2) was also tested for stability of location of the visual RF near body parts when the experimenter shifted position. This neuron exhibited visual RFs near the center of the face of the experimenter and the monkey, with less response to visual stimuli near the chest (Figure 2A, B, D). Two-way ANOVA with two factors of position (three levels: left 35°, center, right 35°) and body part (three levels: face, chest, and rest) revealed a significant interaction (p < .01). The response profile related to each body part was unchanged in each position. Responses near the “face” were always significantly stronger than those on the “chest” and under the “rest” condition (all positions, p < .0001). In the “face” condition, responses were stronger for the left 35° and center positions than those for the right 35° position (p < .0001 and p < .01, respectively).
Converging findings of monkey and human brain studies suggest that a region in parietal association cortex is involved in visuotactile integration important for bodily consciousness (Murata & Ishida, 2007). Physiological studies of visuotactile body representation have demonstrated that bimodal neurons in area 5 and PEa are related to coding of the egocentric reference frame and may also contribute to monitoring of one's own body (Graziano, Cooke, & Taylor, 2000; Obayashi, Tanaka, & Iriki, 2000; Iriki, Tanaka, & Iwamura, 1996; Sakata et al., 1973). Recent human brain mapping studies have suggested that the inferior parietal lobule and the temporo-parietal junction or extrastriate body area may be involved in processing of the embodied self by multisensory integration (Culham & Valyear, 2006; Blanke & Arzy, 2005; Fogassi & Luppino, 2005). Although it is clear that parietal cortex is involved in representations of self and others, how the brain can appropriately map others' bodies remains unclear.
We examined whether visuotactile bimodal neurons in monkey parietal cortex map others' body parts onto the body representation of the self. Some neurons with RFs anchored on the monkey's body exhibited visual responses near corresponding body parts of an experimenter facing the monkey. Visual RFs were within approximately 30 cm from the body surface. Furthermore, some bimodal neurons in the convexity of the inferior parietal lobe displaying visual RFs near the experimenter's forearm specifically responded to action directed toward the experimenter's body rather than that of a rubber hand glove, suggesting the functional relationship of body-matching neurons with mirror neurons recorded in this area.
Although this study was a preliminary investigation, some alternative explanations of its findings can be excluded. First, the visual responses recorded in this study did not reflect simple object fixation or oculomotor behavior of monkeys. In the first step of behavioral testing, we consistently checked responses related to eye movements or position by presenting food or other visual stimuli. Second, the findings we obtained were not due to visual stimuli within a visual RF extending from the monkey. Because all neurons responded well to a moving stimulus within 30 cm of the monkey, activity decreased gradually to more distant stimuli as far as 120 cm from the monkey. In contrast, visual responses could be evoked at 120 cm when the experimenter faced the monkey. Moreover, some neurons we checked exhibited significantly strong responses only within a space about 30 cm from each body, but not in between these regions (60–90 cm). This strongly suggests that neurons coded only the peripersonal spaces of the monkey and the experimenter.
Third, visual responses near the experimenter were unrelated to visual attention. If this was not the case, we should have observed equal responses during stimulation of any other body parts of the experimenter or presentation of objects resembling body parts (e.g., a glove). In contrast to this, we observed selective responses near individual body parts that appeared to exhibit particular types of correspondence between the monkey and the experimenter.
Finally, the possibility exists that neurons could have been responding to absolute position with respect to the monkey (Galletti & Battaglini, 1989). This seems unlikely, however, because visual RFs maintained selectivity for the same body part even when the position of the experimenter changed. Conversely, some neurons in monkey parietal cortex have been found to be selective for target position in the object-centered frame of reference rather than for retinotopic coordinates (Chafee, Averbeck, & Crowe, 2007). However, the bimodal neurons identified in the present study did not exhibit simple object-centered visual properties. Visual RFs were anchored to tactile RFs, suggesting that these neurons represent self body parts. Bimodal properties indicate a difference between representations of objects and those of the body. To exclude the possibility that some apparent body-matching effects reflect the activation of eye-centered/object-centered visual RFs or eye position/eye movement influences, further controlled experiments are required.
The existence of body-matching neurons supports our hypothesis that self body representations might also be used as references for the bodies of other individuals. A bimodal neuron that matches body parts of the self to those of others may be a source of body-part information for mirror neurons involved in the recognition of others' actions. The VIP, in fact, exhibits numerous anatomical connections with PFG (7b), where parietal mirror neurons were recorded. This suggests that body-matching neurons could be a neural basis for shared body representation between self and others in human parietal cortex. In humans, some studies have demonstrated that secondary somatosensory cortex (SII) is activated during observation of somatosensory events on another's body (Ebisch et al., 2008; Keysers et al., 2004). Notably, the VIP exhibits connections with SII (Lewis & Van Essen, 2000). Furthermore, visuotactile synesthesia was reported in one subject who experienced observation of another person being touched as tactile stimulation on the specular (mirror) corresponding part of her own face (Banissy & Ward, 2007; Blakemore et al., 2005). In addition, the case of an anosognosic patient who denied another patient's paralysis has suggested the importance of own body representations for perception of others' bodies (Ramachandran & Rogers-Ramachandran, 1996). These findings are also consistent with our hypothesis.
Our findings for body-matching neurons demonstrate the neuronal basis of the direct-matching system of mirror neurons, with events on others' bodies automatically simulated using a representation of one's own body parts. For example, recent findings have suggested a role for bimodal neurons in the VIP in predicting touch sensation for objects close to the body by visual information or visual imagery of touch by somatosensory input, to guide or protect the face and brain (Graziano & Cooke, 2006; Sereno & Huang, 2006; Avillac et al., 2005; Pouget, Deneve, & Duhamel, 2002). Furthermore, it appears that body-matching neurons could be used to predict others' perceptual status based on observation of their bodies (Gallese, 2005). For example, when subjects observe an object moving toward/on another's body, they can predict the projected point of impact on the body of the approaching object or determine the visual location of a tactile stimulus. These capacities could contribute to transformation of visual perspective between self and others.
In addition, we speculate that recognition of relationships between an object and space may involve social space coding between self and others. For example, subjects may approach each other to achieve intimacy but move away from one another to maintain a margin of safety. Similarly, Fujii, Hihara, and Iriki (2007) showed that visual and joint bimodal neurons in the anterior wall of the IPS in monkeys exhibited alteration of responses related to spatial relationships among the self, others, and a food item when a dominant monkey and a submissive monkey were placed in socially conflicting circumstances. Body-matching neurons may contribute effectively to the above-noted communicative capacity to map the peripersonal space of self and others.
We conclude that coding of others' body parts occurs in the same area that encodes self body parts, and that the map of self body parts is referred to in perception of others' bodies. Interestingly, the VIP displays numerous connections with the PF/PFG, in which mirror neurons were recorded, and this network could comprise a source of body part information for the mirror neuron system. In imitation learning and shared interpersonal representations, the brain could superimpose others' bodies onto the self representation. Our findings provide important clues to understanding the matching system between self and others' bodies.
This work was supported in part by Grants-in-Aid for Scientific Research on Priority Areas (13210133, 14017090, 15016102, 16015307) and for Scientific Research on Priority Areas “Emergence of Adaptive Motor Function through Interaction between Body, Brain, and Environment” (18047026, 20033022) from the Japanese Ministry of Education, Culture, Sports, Science, and Technology, and by the Narishige Neuroscience Research Foundation and Nakayama Foundation for Human Science. We thank all the staff of the Life Science Laboratory for caring for the animals.
Reprint requests should be sent to Akira Murata, Department of Physiology, School of Medicine, Kinki University, Ohno-Higashi, Osaka-sayama, Japan 589-8511, or via e-mail:email@example.com.