In this study, we tested six chimpanzees in eye-tracking tasks to examine whether they specifically attended to strange arms or legs of chimpanzee pictures, compared to normal arms or legs, to determine whether they possessed visual body representation. Compared with looking durations toward the normal body parts, the chimpanzees had significantly longer looking times toward the human arms and legs in place of the original chimpanzee arms and legs. This suggests that the chimpanzees noticed that the human parts were strange. They also showed longer looking times towards the misplaced parts than towards the normal parts, but the difference just failed to meet significance.
The “misplaced” condition and “replaced by a human part” condition showed different contrasts against the control condition: the former did not reach significance, although close, while the latter showed significance. The longest attention to the human parts is probably due to the inconsistency of the shapes, or their interests on human parts. Chimpanzees are able to detect an odd stimulus out of uniform distractors (Tomonaga 1998). Human arms and legs are hairless, and look differently from chimpanzee arms and legs, although the overall shapes are similar. In the other two experimental conditions, the manipulation was done with the body parts of the same chimpanzee body. Therefore, the special look of human body parts may have grabbed more attention. The chimpanzees we tested were very familiar with humans. They see and interact with multiple humans every day, and they could see humans in the institute and on the street. Therefore, they had been exposed to human body parts. That said, it is unlikely that they have seen the whole naked arms and legs of humans as were shown in the task; but the experience of exposure to partial human arms and legs may have triggered them to pay more attention to the human parts appearing on chimpanzee bodies and replacing chimpanzee body parts. Therefore, for the results in the “replaced by a human part” condition alone, the longer looking time could be attributed to their body representation, visual inconsistency, or interests to human bodies. To rule out the possibility that they showed longer looking time in this condition solely because of visual inconsistency or interests to humans, more control conditions could be added, or chimpanzees who have less exposure of partially naked humans could be tested. Nevertheless, when we combine all the results, we still tend to think chimpanzees may be able to detect strangeness in terms of body representation, because of the strong tendency of longer looking times towards the misplaced body parts than the normal parts and the tendency of shorter time to first fixation in the “misplaced” leg condition than the “normal” leg condition.
In the analyses of time-to-first-fixation data, we found that there was a significant interaction between condition and body part. The pairwise comparison showed significant differences in three pairs in leg data: the “misplaced” condition had shorter time to first fixation on AOIs than those in all three other conditions. It is possible that the misplaced legs make the whole body configuration look much stranger than a normal body as well as a body with its leg replaced by another part in the original typical position, leading to a much quicker detection. This was not the case for arm manipulation, and this is where the difference of the effect of condition lies for arm and leg manipulations.
The difference of results between arm and leg manipulations was not found in fixation duration, but time to first fixation, as mentioned above. In this specific case, the quicker detection to misplaced legs than legs in other conditions, but not in arms, may come from the fact that legs do not move in the same amplitude as arms. When chimpanzees move in a quadrupedal posture on the ground or in a bipedal posture when climbing, their arms and legs move in similar ranges. However, when manipulating objects on the ground, they reach for objects in places that are a bit far from them using arms not legs, and it could be seen as if the arms were “misplaced” from a distance (e.g., Hayashi and Matsuzawa 2003; Hayashi et al. 2005); chimpanzees also often raise their arms for social communications (Hobaiter and Byrne 2011), but they seldom “raise” their legs. The different function and use of arms and legs could cause chimpanzees detect misplaced legs more quickly.
In this study, we did not manipulate other body parts, such as head and torso. It will be interesting to further examine how their representation differ across various body parts. In a broader comparison counting all body parts, the difference between arms and legs may not be as large as that between head and limbs, or other contrasts. Studies asking children to recognize, name, and point at body parts do not demonstrate large differences between arms and legs, but the performances for eyes was much earlier in the development (MacWhinney et al. 1987; Waugh and Brownell 2015; Witt et al. 1990). Atypical body parts may suggest injury and care, so it is meaningful to examine whether and how knowledge for body parts differ, and which factors are related to this, such as function of the parts. Also, it will be interesting to examine the body representation of other species, too, e.g., preys. Do chimpanzees (and humans) have certain body representation and anatomy knowledge about their preys’ bodies, and do the knowledge help with efficient foraging and feeding?
All the manipulations in this study created strange images that will not occur in real life, yet the chimpanzees did not show significant differences in all manipulated conditions compared to control. One of the reasons could be due to the limited sample size. The significance of the random effect participant ID in both analyses of time-to-first-fixation data and fixation-duration data also indicates individual difference (Figs. 4, 5, 6, 7). If more individuals were tested, the results might have been more consistent. Because of the limited sample size, the conclusions should be generalized with caution, and data from more chimpanzee individuals or populations will be helpful to understand chimpanzees’ perception for atypical body parts.
The participants in this study were captive chimpanzees with a lot of exposure of humans. As discussed above, these individuals might be more sensitive to human body parts on chimpanzee bodies, compared to captive chimpanzees with limited human exposure or wild chimpanzees. However, the experience with humans may not affect chimpanzees’ body representation too much, according to our previous findings (Gao et al. 2020; Gao and Tomonaga 2020b). We tested the same chimpanzees, who were very familiar with humans, to see if they show the inversion effect for human bodies. We used humans in bipedal postures doing Tai chi, but the chimpanzees did not show any inversion effect. We then used bipedal humans showing daily postures (waving hands, walking, etc.), and the chimpanzees showed the inversion effect to these bodies, suggesting that visual experience is important to them. We also used images of crawling humans and horses in quadrupedal postures, which the chimpanzees had never seen previously, but they showed the inversion effect. Their limited inversion effects to humans, a familiar species, and the inversion effects to the quadrupedal animals that they had no visual experience about, suggest a strong tendency to refer to embodied cues, i.e., cues from their own bodies, in their body perception. Therefore, experience with humans may not affect chimpanzees’ body representation for conspecific bodies too much.
The random effect, picture ID, was significant in fixation-duration analysis (Fig. 8). This suggests that the results vary across the pictures. There are several outlier points, but not many. It is possible that the significance is related to the limited data we have: not every picture in each condition received a lot of fixations. As will be discussed below, it is inevitable to have many trials without any fixations in a chimpanzee experiment, and future studies could use more trials for more useful data points. Nevertheless, because the 20 pictures (with different kinds of manipulations) were used across conditions, this significant effect of picture ID does not interfere with the significance of condition, the main effect in the analysis.
There were several other limitations in this study. The number of trials in which the chimpanzees showed fixations to AOIs was less than half of the total trial numbers. Chimpanzees do not consistently look at the screen during a task. When they do, they typically view face and genital areas of a chimpanzee picture, while they allocate less attention to other body parts (Kano et al. 2015). More importantly, because the AOIs in this study were arms or legs (i.e., a small proportion of the whole picture), it is reasonable to have many trials without fixations on AOIs. Nevertheless, future studies in a similar setting could use a larger stimulus set to ensure more data points.
When we prepared the stimuli, the pictures were chosen randomly, and the body parts for manipulations were chosen based on picture editing convenience; we hoped to minimize editing to avoid any effects of unnatural picture manipulations. Overall, we edited 10 left arms, 10 right arms, 5 left legs, and 15 right legs. This should not fundamentally affect the experiment, because chimpanzees were in various positions (e.g., sitting, walking to the left, walking to the right, and bipedal standing), and left/right discrimination was less prominent than in a situation involving only bipedal animals. Nonetheless, future studies should carefully consider left/right bias to ensure a more balanced experimental design.
In summary, our results showed a significant longer looking time towards human body parts on chimpanzee bodies, and two non-significant tendencies: (1) shorter latencies for fixating misplaced legs, and (2) longer looking times towards misplaced parts, compared to normal body parts. These detections of strange body parts indicate that chimpanzees might have a body representation of the typical chimpanzee body. Conspecific body representation has ecological value. For example, it helps animals discriminate among conspecific individuals and individuals of other species (and then they can decide whether to fight them, socially interact with them, prey and feed on them, or ignore them). Strangeness on body parts of living individuals can indicate injury, and body representation can help trigger emotional and behavioral changes to facilitate care for these individuals (Hirata et al. 2017; Matsumoto et al. 2016; Sato et al. 2019). From an evolutionary perspective, evidence of body representation among chimpanzees indicates that the common ancestor of chimpanzees and humans might also have this type of visual representation. Of course, this conclusion needs to be supported by more data from more participants and from studies with further examinations besides arm and leg manipulations. Nevertheless, if this is true, it will lead to many interesting questions. Because both chimpanzees and humans are highly social species, and both encounter many other individuals, it is important to investigate whether body representation originates from the accumulated visual experience of conspecifics’ bodies. Investigation of this point requires examination of more solitary species, such as orangutans, as well as examination of body representation development. If the representation is present in solitary species or develops before intensive social interactions with other individuals, body representation may have more fundamental functions in animals’ life as follows: apart from aiding interactions with other individuals, body representation may be involved in many self-centered activities (Shapiro 2019). Further investigations of body representation and its interactions with other psychological processes are important for understanding how animals coordinate themselves with the outside world.