On Visual Kin Recognition and Family Resemblance in Chimpanzees

The results of Parr and de Waal (1999) on male-offspring biased visual kin recognition in chimpanzees was replicated using both human judgements, and a principal components analysis (PCA) applied to pixel-maps. With the same matching to sample paradigm and stimulus sets as used by Parr and de Waal (1999), both humans and the PCA produced the same sex-biased kin recognition as that found with the chimpanzees. The PCA suggested that the basis of the sex-based bias in kin recognition could be a function of the similarity in the “framing” of the photographs of mothers and their sons, but not daughters. Eliminating the potential framing bias, either by cropping the photos (Experiment 2) or rebalancing the recognition foils (Experiment 3), also eliminated the sex bias, but not the recognition of chimpanzee kin with human participants. For kin selection (Hamilton, 1964a, 1964b) to play the rôle posited by evolutionary This work was supported by operating grants from the Natural Sciences and Engineering Research Council of Canada to each of the first two authors. Portions of this work were presented at the joint meeting of the Canadian Society for Brain, Behaviour and Cognitive Science (BBCS) and the Experimental Psychology Society of Britain (EPS) in Cambridge, UK, July 21, 2000, and at the Banff Annual Seminar in Cognitive Science (BASICS) in Banff, Alberta, Canada, May 10, 2002. We thank Leah Christensen for her assistance in collecting the data for Experiments 2 and 3. We also thank Hervé Abdi, Alice O’Toole, Dominique Valentin, and Betty Edelman whose earlier work with autoassociative neural network models of face perception served as a model for the current research. Finally, the first author would like to thank Martin Daly, who first introduced him to the psychological issues underlying the problem of kin recognition. Reprint requests should be sent to Dr. John R. Vokey, Department of Psychology and Neuroscience, University of Lethbridge, Lethbridge, Alberta, Canada, T1K 3M4. Electronic mail should be sent to: [email protected] VISUAL KIN RECOGNITION 2 biologists, there must be some mechanism for the discrimination of kin from non-kin. Suggested mechanisms have ranged from experiential familiarity with nest, troop, or clutch mates of non-dispersing species to explicit genetic markers and shared tropisms in other species (e.g., Greenberg, 1979; Waldman & Adler, 1979; Wu, Holmes, Medina, & Sackett, 1980). Recently, Parr and de Waal (1999) reported an unusual example of visual kin recognition. In their study, five chimpanzee subjects matched facial photographs of mothers and their male offspring at levels above chance when both were unrelated and unfamiliar to the subjects; perhaps surprisingly, no such effect was found for facial photographs of female offspring. The asymmetry in the chimpanzees’ ability to match photographs of mothers to their sons but not daughters was given a functional interpretation by Parr and de Waal (1999) as an adaptive response to the patrilineal structure of chimpanzee communities. In such groups, it is the males who form the stable core of related individuals, whereas adult females are unrelated having immigrated in from other communities at sexual maturity. It is also the males who show high levels of social affiliation and cooperation thereby potentially reaping the kin-selected fitness benefits associated with kin-biased social behavior (cf. Hamilton, 1964a, 1964b). The fact, then, that these chimpanzees were better able to match the faces1 of unfamiliar males as opposed to females to those of their mothers suggests that the patrilineally stratified nature of chimpanzee communities has favoured a specialised mechanism related to faces and face processing that facilitates efficient detection of male relatedness in this species, with no selective pressure for and, hence, relatively inefficient or no recognition of female relatedness (Parr & de Waal, 1999). Two very general paths are available for such a specialised face recognition mechanism to come about. The first is the development of face recognition routines in the perceiver that are specialised for the detection of pre-existing differences in how the faces of sons and daughters resemble their mothers’. For example, developmental differences between sons and daughters may well result in differences in the extent to which certain facial characteristics resemble those of their mothers’, some favouring sons and others favouring daughters (e.g., male development may exaggerate a trait of the mother’s appearance—large ears, say, or flatness of face—that is not similarly exaggerated in female development), and those differences favouring similarity with sons are then capitalised upon by face recognition routines evolutionarily tuned to detect just those differences. The second general path is that characteristics of the faces themselves, either structurally or, more plausibly in our view, through characteristic, identifying behaviours (e.g., pose, expression) have been or through development are modified to bias pre-existing recognition routines toward the detection of the facial similarity of mothers and their male offspring. Thus, for example, sons may copy the poses and facial expressions of their mothers precisely to encourage the detection of relatedness to her and, thereby, other male offspring. As adult females do not live in the To be accurate, the chimpanzees matched photographs of chimpanzee faces; we are sensitive to the issue of whether this result should be read further as one of face perception and individual recognition on the part of the chimpanzees. For the purposes of their functional interpretation, Parr and de Waal (1999) assumed that the results reflected face perception and recognition. For the sake of that argument, we make the same assumption here. However, what evidence there is suggests that the chimpanzee behaviour is indeed one of recognition of the individuals depicted rather than merely the matching photographs, and a skill that may also be found with rhesus monkeys (Parr, Winslow, Hopkins, & de Waal, 2000) and long-tailed macaques (Dasser, 1987, 1988). VISUAL KIN RECOGNITION 3 same social group as their mothers or brothers, there would be no selective pressure for them to adopt the poses and expressions of their mothers or siblings. Of course the two paths could develop concurrently, but in any case the end result is that the faces of male offspring are perceived as more like those of their mothers than are the faces of female offspring. However, if the path toward specialisation were principally the former, then other recognition systems, not specifically tuned to the different ways male and female offspring resemble their mothers, should not discriminate between sons and daughters in kin recognition. That is, they might well be able to detect kin similarity, but not differentially for sons over daughters. Conversely, if the path were principally the latter, then other recognition systems might well be able to respond to it, seeing sons as more facially similar to their mothers than are daughters. To investigate these possibilities, we undertook a series of experiments and a simulation designed to test both the species-specificity of these potential biases in face-processing or face production, and to explore the potential structure of the underlying kin recognition mechanisms. In the first experiment, the same set of photographs and the same matchto-sample procedures used in Parr and de Waal (1999) were employed to test a group of human participants on their ability to match the mother-offspring pairs. Because the human participants had no special previous experience with chimpanzees or photographs of them, nor does it appear likely that their perceptual systems would have been specially tuned to visual cues for male relatedness in photographs of chimpanzees, this approach allowed us to control for these potential influences on such judgements.2