Ability to Recognize Individuals
KeywordsFace Recognition Inversion Effect Vervet Monkey Individual Recognition Holistic Processing
The ability to recognize individuals is a highly advantageous skill for social living species. Although not always clear, an important distinction in the literature is that between individual discrimination and individual recognition. Individual discrimination is the ability to differentiate one individual from all the rest, whereas individual recognition is the ability to differentiate each individual from every other individual and is the most precise form of recognition (Beecher 1989). This entry discusses the evidence for the recognition of conspecific individuals in nonhuman animals, specifically through the auditory and visual systems.
Studying nonhumans provides important insight into the evolution of human sociocognitive skills, such as individual recognition. As in human societies, most social species live in groups that are structured by kinship, dominance, and reproductive status (Smuts et al. 1987), making the ability to recognize others critical to navigating their social world. Animals may use olfactory, auditory, behavioral, or visual cues to recognize others. As much of the research has been conducted in the auditory domain, research on vocal recognition is first discussed briefly. Then, because most social living animals, particularly mammals (including humans), are highly reliant on vision, the majority of the text will focus on recognition within the visual domain.
Of course, there are many different types of recognition, including the recognition of one’s own species, kin recognition (see entries on “Kin Recognition” and “Kin Classification Systems Support Kin Recognition” for more information), the recognition of familiar versus unfamiliar individuals, individual recognition, the recognition of sex, age, and so on. Individual recognition is the most precise form of recognition. Therefore, the discussion will center on individual recognition within one’s own species or the ability to discriminate one particular individual from every other conspecific individual. Individual recognition underlies many sociocognitive processes, facilitates group cohesion, and is prerequisite for more complex behaviors such as reconciliation (see “Reconciliation”). This implies that there was strong evolutionary pressure in group-living species to individually recognize others and remember those with whom they have interacted.
This entry will focus on the nonhuman primate (NHP) literature because we ourselves are primates and, thus, the majority of research on individual recognition in animals has focused on NHPs (although see section “Face Recognition in Non-primates” for evidence in other species). Moreover, some researchers have posited that social problems may have been the most cognitively complex problems that humans may have faced in our recent evolutionary history, and this may help explain our large brain to body ratios compared to the rest of the animal kingdom (e.g., Byrne and Whiten 1988). Thus, NHPs provide us with the opportunity to study the evolutionary function of human sociocognitive skills and visual perception within a comparative framework. Studying other NHPs allows us to determine whether a particular trait is shared with humans due to common evolutionary history (i.e., homology) or due to shared evolutionary pressures (i.e., convergence).
A large portion of the research examining the ability to recognize individuals has been conducted in the auditory domain, primarily through playback experiments in which researchers record naturally occurring vocalizations and then play the prerecorded stimuli back to subjects in order to gauge the subjects’ responses to novel situations or in order to reproduce events that may occur naturally. Playback experiments designed to help elucidate the ways in which NHPs classify each other have found evidence that NHPs recognize their infants, more distant kin, and their neighbors (e.g., Cheney and Seyfarth 1990; Rendall et al. 1996).
For instance, a study conducted on wild vervet monkeys (Cercopithecus aeithiops) played the scream of a 2-year old juvenile in the presence of its mother and two other adult females (who had offspring of their own) while they were separated from their offspring. The mother of the juvenile responded more quickly than control mothers, looking and approaching the speaker from which the call originated, indicating that vervet monkeys recognize their infant as distinct from other infants. Furthermore, it was noted that the control females responded by looking at the infant’s mother, often before the mother made any movement. The anticipatory behavior of the control females suggests that they recognized the relationship between the screaming juvenile and its mother (i.e., third party relationship; Cheney and Seyfarth 1990).
Other evidence in the auditory domain indicates that the ability to recognize individuals may go beyond immediate kin. Rendall and colleagues (1996) first found that adult female rhesus macaques (Macaca mulatta) responded faster and looked longer towards the origin of matrilineal kin calls compared to non-kin, suggesting that they were able to discriminate the calls of matrilineal kin from non-kin. Following this, they implemented a habituation-dishabituation paradigm to test for individual recognition. In this paradigm, subjects typically experience a series of habituation trials during which an exemplar of the stimulus class is repeatedly presented, in this case, various calls of one matrilineal relative. Following habituation, the test stimulus of either the same or different class is presented, such as the call of a second matrilineal relative. A significant difference in response (dishabituation), such as a longer looking time, is taken as evidence that the subject judged the two types of stimuli as different. Following habituation in this study, females responded significantly more to the call from the second matrilineal relative, suggesting vocal recognition at the individual level.
There is also evidence that vocal recognition may extend beyond the boundaries of the group. For example, vervet monkeys reacted more strongly to the calls of individuals from a neighboring group when the calls originated from an inappropriate territory compared to when the calls originated from within the groups’ typical territory. This implies that the monkeys had expectations about where those individuals should be (Cheney and Seyfarth 1990).
Many of the studies described above have been taken as evidence for individual vocal recognition. However, for individual recognition to take place, subjects must not only recognize a call as familiar but also perceive that it belongs to a specific individual (Beer 1970). Although it is certainly possible that individual recognition occurred in the aforementioned studies, subjects may have categorized the vocalizations at a more general level rather than identifying each of these individuals specifically. For instance, mother-offspring recognition may simply involve discrimination between one’s own offspring from all others. This should be considered discrimination, rather than individual recognition. Moreover, the discrimination of one’s own offspring or kin from all others may occur through signature cues such as family specific acoustic cues. Likewise, the recognition of group mates may simply involve the discrimination of familiar versus unfamiliar vocalizations. In the case of the vervet monkeys, vocal recognition of neighbors may reflect an association between a familiar neighbor’s sound and its familiar location.
Thus, while these studies present behaviors that seem to indicate individual recognition, controlled laboratory studies that manipulate the exposure to stimuli allows researchers to better evaluate the importance of particular features and how such information is organized in the brain. To date, however, the majorities of laboratory studies have focused on how animals perceive, process, and organize nonsocial information. The few experimental studies that have employed social stimuli to examine how animals acquire complex social knowledge have primarily been conducted in the visual domain (although intermodal studies are notable exceptions, e.g., Adachi and Hampton 2011).
Visual Face Recognition in Nonhuman Primates
Much of the research on visual recognition has employed two-dimensional photographs as experimental stimuli in place of real-life objects to assess human and nonhuman sociocognitive processes. The use of photographic stimuli is more reliable than presenting real objects or individuals because it allows researchers to manipulate subjects’ exposure to social information, helping rule out alternative hypotheses. Moreover, the use of photographs provides controlled investigation of image qualities such a brightness, contrast, viewpoint, and so forth. A number of studies have used indirect measures, such as looking time duration or habituaton-dishabituation procedures, to examine individual recognition in the visual domain; however, some have used more objective measures, requiring subjects to make a direct response in the task. Although studies that require an objective response from subjects tend to require more extensive training, these studies provide conclusive results for individual recognition, particularly the perception and recognition of faces. Faces are a highly salient class of visual stimuli as they provide primates, both human and nonhuman, and perhaps other species as well (discussed below) with valuable social information such as the age, sex, individual identity, and the emotional state of others (for reviews see Bruce and Young 1998; Leopold and Rhodes 2010; Parr 2011). Thus, the ability to individually discriminate faces and use the information present in faces likely played an important role in primate evolution.
Within the primate order, the majority of neurological, developmental, and behavioral research suggests a common evolutionary route for the visual recognition of faces. First, growing evidence indicates that at least some species of nonhuman primates possess a face processing system that shares similar neural underpinnings with humans. From a large body of behavioral and neurological data, we know that humans possess a specialized mechanism for face processing (see Tsao and Livingstone 2008, for a review). Functional magnetic resonance imaging (fMRI) studies in humans have revealed a system of face-selective areas in the inferotemporal (IT) cortex that are involved in face recognition, including (but not necessarily limited to) the fusiform face area (FFA), the occipital face area (OFA), and an area of the superior temporal sulcus (STS-FA). These areas may be specialized for different functions. For example, the FFA is thought to be involved in processing identity, whereas the OFA is involved in processing face parts and the STS-FA appears to respond selectively to emotional expression and eye gaze and therefore is thought to be involved in the processing of changeable aspects of the face (Tsao and Livingstone 2008).
Electrophysiological studies in rhesus macaques have found regions in the rhesus macaque brain similar in number and relative size to those in humans. Neurons in the STS of the temporal cortex respond to a variety of human and monkey faces, changes in facial expressions, eye gaze, and facial orientation (Tsao and Livingstone 2008). These findings suggest certain homologies between cortical areas in the human and monkey brain suggesting a common neural mechanism for face recognition in primates. Although this may be true for at least some primate species, it is unclear whether a common face-processing system exists for all primates. That is, a basic structure from which species specializations may have evolved.
The behavioral evidence for a common face recognition system among primates has been mixed. For instance, it is unclear to what degree NHPs rely on second-order configuration, which refers to the relative spatial arrangement of facial features unique to each individual that are thought to provide the information necessary for individual face recognition in humans. Humans incorporate both the basic configuration of the components of a face (i.e., the eyes, which are above the nose, which are above the mouth), or first-order configuration cues, and second-order configuration cues (e.g., the relative spacing and positioning of facial features) into a single perceptual whole through a fast-acting process referred to holistic processing. This is exemplified by the inversion effect, in which humans are slower and less accurate in recognizing faces (but not objects) when they are presented in an upside-down orientation compared to an upright orientation, due to the disruption of holistic processing (Valentine 1988). Holistic processing (and the inversion effect) is often taken as evidence that face processing is different from nonface object processing in that faces are represented as a whole rather than as a combination of component parts (eyes, nose, mouth) and the relations between them.
Yet, behavioral evidence of the inversion effect in NHPs is mixed. In chimpanzees (Pan troglodytes), the inversion effect seems to be dependent upon expertise, such that chimpanzees demonstrate the inversion effect for human and chimpanzee faces but not capuchin faces or cars (Parr 2011; see also Tomonaga et al. 1993). However, this does not seem to be the case for monkeys (Parr 2011). Thus, it is possible that the inversion effect may reflect one facet of face processing (i.e., holistic processing) that is an adaptive specialization of the face recognition system of apes and humans, yet further work is needed to rule out the possibility that differences in methodology contributed to inconsistent results. Comparative research on the inversion effect outside the primate order would also be helpful in elucidating the role of holistic processing in face recognition abilities.
On a more general level, behavioral research on NHPs’ ability to individuate conspecific faces have provided relatively consistent results. One of the most direct ways to evaluate NHPs’ ability to individuate faces is to present them with a task in which they must match the same individual across different viewpoints. This task helps rule out the possibility that subjects are relying on irrelevant perceptual features specific to each photograph, such as symmetry or lighting, to discriminate the stimuli and thus provides additional evidence that face recognition is distinct from basic visual processing. Accordingly, positive results obtained from studies employing paradigms that require direct responses from subjects are generally accepted as evidence for individual recognition (see Parr 2011, for a review). Using this type of methodology, all of the species tested thus far, including chimpanzees, orangutans (Pongo spp.), rhesus macaques, crested macaques (Macaca nigra), and capuchin monkeys (Cebus apella), have demonstrated the ability to discriminate conspecific faces (Micheletta et al. 2015; Parr 2011; Talbot et al. 2015, in press).
Previous studies typically examined this ability using unfamiliar faces; however, many of the more recent studies have included familiar facial stimuli as well. The majority of these studies have found differences in performance based on familiarity, suggesting that experience with, exposure to, and the familiarity of faces may play a critical role in influencing face recognition. In humans, changes in lighting, facial expression, or viewpoint of the facial stimuli impair the ability to recognize unfamiliar, but not familiar, faces (Bruce and Young 1998). Likewise, chimpanzees, orangutans, and capuchin monkeys all performed better when individuating familiar conspecific faces across viewpoints compared to unfamiliar conspecific faces (Parr 2011; Talbot et al. 2015, in press). However, one Old World monkey, the crested macaque, discriminated familiar and unfamiliar faces equally well (Micheletta et al. 2015).
The fact that this effect has been observed in New World monkeys, but not Old World monkeys, which are more closely related to Hominoids (apes), suggests one of three possibilities. First, it may have been present in the common ancestor of Hominoids and New World primates but was subsequently selected against in Old World primates. However, this seems improbable given the presumed benefits of recognizing familiar individuals and because traits tend to be preserved in all of the descendants of a common ancestor unless there were strong selective forces working against the trait. To date, we have been unable to identify such a pressure in Old World monkeys.
A second possibility is that the familiarity effect is a convergent trait of the face-processing system, affected by selective social pressures, with species that live in larger, more complex social groups exhibiting greater nuances in face perception. However, this seems unlikely for several reasons. First, crested macaques live in large multi-male, multi-female groups of up to 100 individuals, whereas tufted capuchins groups are significantly smaller, ranging in the teens to low twenties in size. Second, orangutans, a species that only recently became a primarily solitary species, also exhibit the familiarity effect (Talbot et al. 2015), suggesting that once the effect is established, there does not seem to be sufficient selective pressure for losing it. It may be that the face recognition system evolved in the primates and is maintained despite the diversity in social group size and organization, indicating that current group size or organization is not the key feature.
A third possibility is that differences in methodology may have impacted results. The classic methodology for experiments requiring training calls for the use of novel test stimuli to be presented a limited number of times. In tests of individual recognition, therefore, each individual should only be presented once or twice in order to evaluate any potential differences in spontaneous discriminations as a function of familiarity with those individuals, not the test stimuli themselves (Talbot et al. in press; Thompson 1995). However, in the NHP face recognition literature, there is a trend to adapt a loose definition of “novel,” in part due to the scarcity of high-quality photos that met criterion for testing. For example, new photos of the same individuals observed in training are often used during testing, and/or subjects are repeatedly presented with the same stimuli. Although this is a difficult standard to achieve, the novelty of both the photos and the novelty of the individuals represented in those photos is imperative for a true transfer test. Otherwise, it is impossible to rule out nonconceptual factors, such as memorization of the training stimuli, to explain subjects’ performance. Overall, current evidence indicates that despite apparent differences among species, the familiarity effect is nonetheless a widespread phenomenon observed in many species that exhibit individual recognition.
Much evidence shows that faces are highly salient social stimuli for humans and NHPs from a very early age. Infants orient more towards face-like patterns compared to nonface-like patterns (Pascalis and Kelly 2009). These “face-like patterns” can be as simple as three dots arranged in a triangular fashion, reflecting the basic arrangement of the eyes above the nose, which is above the mouth, which is referred to as first-order configuration. First-order configural cues are important for identifying faces at the categorical level, that is, discriminating faces from nonfaces. Both human and NHP infants demonstrate a preference for faces even when they have never seen a face. In a particularly telling study, infant Japanese macaques (Macaca fuscata) raised in an enriched, but face-deprived, environment for 6–24 months demonstrated a preference for both human and monkey faces over other complex visual stimuli, indicating an innate preference for first-order configural cues (Sugita 2008).
Still, other developmental evidence corroborates the importance of familiarity and the role of exposure on individual face recognition. In particular, evidence suggests that early exposure to faces during a critical developmental period may fine-tune cortical networks to become specialized for the prototypical face to which an individual is exposed. Following a face-deprivation period, infant Japanese macaques preferred to look at and selectively discriminated the species that it was first exposed to (either human or conspecific faces). Likewise, 6-month old human babies discriminated both human and monkey faces, but at 9 months of age, they only discriminated human faces. These results elucidate the role of experience in the development of the “species-specific effect,” which has been likened to the “other race effect” in which it is easier to recognize members of one’s own ethic group or species, or, perhaps more accurately, to the prototypical face to which one is frequently exposed (see Pascalis and Kelly 2009, for a review).
There is evidence in humans and NHPs that the species-specific effect and the other-race effect can be reversed with appropriate experience. For example, a classic study found that rhesus macaques exhibited a species-specific effect in which they discriminated conspecifics, but not domestic animals, yet after several months of exposure to the domestic animals, the macaques could discriminate them as well (Humphrey 1974). Korean children reared exclusively with Koreans and later adopted by Caucasian families between the ages of three and nine demonstrated the same own-race effect as Caucasians exhibit as adults, suggesting that this effect may be reversible with experience (Sangrigoli et al. 2005). Moreover, children as young as 3 months old demonstrated the other-race effect, yet, short-term exposure to other-race stimuli was sufficient to cancel this effect (Sangrigoli and Schonen 2004). Taken together, these studies suggest that experience with and exposure to faces greatly influences face processing both within and across species. In addition, there may be a critical period during early developmental during which the face processing system undergoes perceptual narrowing, but with appropriate exposure this is reversible to a certain extent.
Face Recognition in Non-Primates
Face recognition is by no means limited to the Primate order, but compared to human and NHPs, less is known about whether other animals discriminate and process faces and the underlying cognitive and neural processes by which they may do so. Although the literature on individual recognition in non-primate species is still relatively nascent, growing evidence across the animal kingdom suggests that individual recognition and face recognition, in particular, may not be as rare as once thought.
Similar to humans, sheep (Ovis aries) prefer conspecific faces to heterospecific faces, and familiar faces to unfamiliar faces. Using facial cues alone, sheep discriminated between different species, and, within their own species, the breed and sex of conspecifics. Sheep can learn to discriminate individual faces of adult sheep and can remember up to 50 familiar faces over a period of 2 years. However, individual discrimination is poorer on young lambs and unfamiliar individuals. Moreover, sheep rely on configural cues for the recognition of familiar faces, like humans and most NHPs. Neurological evidence suggests that primates and sheep may share similar neural pathways. The temporal cortex of the sheep brain contains cells that respond selectively to faces compared to other visual stimuli. The responsiveness of particular categories of these cells varies based on dominance (the size of horns), familiarity, breed, and potential level of threat (see Tate et al. 2006, for a review). Collectively, these studies provide strong evidence for a similar face processing system in sheep and primates, with the notable exception that particular groups of face-selective cells have specialized functions that are of social and ecological relevance to each species.
In birds, much of the evidence for visual individual recognition is tangential. For instance, rooks (Corvus frugilegus) are social corvids that live in large colonies and form long-term social bonds, suggesting that they have a strong possibility of recognizing group members individually (Emery et al. 2007). Yet, evidence of individual recognition in this species is lacking. Bird and Emery (2008) presented rooks with a strongly affiliated conspecific and an unfamiliar conspecific through multiple modalities (including live, video, and static images) and found that rooks preferred to look at their affiliated conspecific over an unfamiliar conspecific when videos and live stimuli were used. From this evidence, the authors concluded that rooks recognize the individuals in the video. However, this should not be taken as evidence of individual recognition as the rooks may simply be discriminating familiar from unfamiliar individuals, a considerably less precise form of recognition. Perhaps the most convincing evidence of individual recognition in birds comes from budgerigars (Melopsittacus undulates). Budgerigars discriminated conspecific faces from heterospecific faces and discriminated between conspecific faces, suggesting individual discrimination (Brown and Dooling 1992, 1993). While evidence demonstrating that multiple views of the same individual were perceived by budgerigars as the same individual would provide more conclusive for individual recognition, like the other species discussed, conspecific face recognition in budgerigars appears to be reliant on configural processing, as subjects processed faces in their normal configuration more efficiently than inverted faces or faces with scrambled features.
Most fish live in stable social groups or shoals throughout their lives and demonstrate behaviors that are indicative of social intelligence (see Bshary 2011 for a review). For instance, stickleback fish will not cooperate with individuals who have previously cheated in dangerous predator inspections, suggesting individual recognition and memory of conspecifics. Yet the means by which recognition occurs is poorly understood. Olfactory and chemical cues are likely modes of communication in an aquatic environment as water typically aids the dispersal of such cues; however, recent work suggests the importance of visual cues, particularly facial cues, in more precise forms of recognition in fishes. A recent study found that African cichlid fish (Neolamprologus pulcher) discriminated familiar individuals from unfamiliar individuals using facial features (i.e., color patterns) alone (Kohda et al. 2015). Moreover, two species of damselfish (Pomacentrus amboinensis and P. moluccensis) use ultraviolet facial patterns for species discrimination and potentially for individual recognition as well (Siebeck et al. 2010). However, more research needs to be done in order to confirm that fishes individually discriminate conspecifics using these distinctive facial cues.
Some species of paper wasps, like the Northern paper wasp (Polistes fuscatus), exhibit highly variable facial and abdominal color patterns, which are used to individually recognize their nest-mates. Sheehan and Tibbetts (2011) compared the face learning abilities of P. fuscatus and P. metricus, a closely related species that does not possess elevated phenotypic variation like P. fuscatus and has not been shown to individually recognize conspecifics. P. fuscatus more accurately and quickly discriminated pairs of normal conspecific and heterospecific faces compared to manipulated face images (composed of the same colors and patterns of normal faces) and nonface images, whereas P. metricus did not. This suggests that the specialization for face learning in paper wasps is not based on general pattern or shape discrimination and it is not due to species-specific variability in faces. Furthermore, morphological measurements demonstrate that P. metricus has more acute vision than P. fuscatus ruling out the possibility that differences in visual acuity may explain the results. Thus, specialized face learning in paper wasps appears to be a cognitive adaptation that likely evolved in response to social pressures as P. fuscatus colonies are founded by multiple queens and exhibit a linear dominance hierarchy, whereas P. metricus typically nest alone (Sheehan and Tibbetts 2011).
Many animals possess the ability to visually discriminate conspecifics. Collectively, these studies strongly suggest that a face-processing system is not human, or even primate, specific. Current evidence suggests that individual recognition may be a cognitive adaptation that evolved convergently in response to ecological and social pressures. Some species (e.g., sheep and NHPs), currently appear to have more complex forms of recognition, which indicates that recognition may have evolved in a stepwise fashion with more general forms of recognition (e.g., species discrimination) emerging before more precise forms (e.g., individual recognition). It seems likely that certain aspects of face perception, such as the preference towards orienting towards simple face-like patterns or even particular features of the face such as the eyes as this was imperative for predator detection, may share a deeper evolutionary history, whereas, other, more specialized facets of face perception, such as the use of second-order configural cues, may have evolved more recently during primate evolution, explaining its presence in humans and apes but not consistently observed in monkeys. Further work in a variety of species that vary in terms of sociality, ecology, and evolutionary history will clarify which facets of face perception result from homologous or convergent processes.
I thank Sarah F. Brosnan and Darby Proctor for helpful feedback. CFT was funded by NSF GRFP (DGE-1051030).
- Beer, C. G. (1970). Individual recognition of voice and its development in birds. Proceedings of 15th International Orinthonological Congress, 339–359.Google Scholar
- Bruce, V., & Young, A. (1998). In the eye of the beholder: The science of face perception. Oxford: Oxford University Press.Google Scholar
- Byrne, R., & Whiten, A. (1988). Machiavellian intelligence: Social expertise and the evolution of intellect in monkeys, apes and humans. Oxford: Clarendon.Google Scholar
- Cheney, D. L., & Seyfarth, R. M. (1990). How monkeys see the world. Chicago: University of Chicago Press.Google Scholar
- Emery, N. J., Seed, A. M., von Bayern, A. M., & Clayton, N. S. (2007). Cognitive adaptations of social bonding in birds. Philosophical Transactions of the Royal Society of London, Series B, 362, 489e505.Google Scholar
- Kohda, M., Jordan, L. A., Hotta, T., Kosaka, N., Karino, K., Tanaka, H., … & Takeyama, T. (2015). Facial recognition in a group-living Cichlid Fish. PloS one, 10(11), e0142552.Google Scholar
- Smuts, B. B., Cheney, D. L., Seyfarth, R. M., Wrangham, R. W., & Struhsaker, T. T. (1987). Primate societies. Chicago: University of Chicago Press.Google Scholar
- Talbot, C. F., Leverett, K. L., & Brosnan, S. F. (2016). Capuchins recognize familiar faces. Animal Behaviour, 122, 37–45.Google Scholar
- Tate, A. J., Fischer, H., Leigh, A. E., & Kendrick, K. M. (2006). Behavioural and neurophysiological evidence for face identity and face emotion processing in animals. Philosophical Transactions of the Royal Society of London B: Biological Sciences, 361(1476), 2155–2172.CrossRefPubMedPubMedCentralGoogle Scholar
- Thompson, R. K. R. (1995). Natural and relational concepts in animals. In H. Roitblat & J. A. Meyer (Eds.), Comparative approaches to cognitive science (pp. 175–224). Cambridge, MA: Bradford/MIT Press.Google Scholar
- Tomonaga, M., Itakura, S., & Matsuzawa, T. (1993). Superiority of conspecific faces and reduced inversion effect in face perception by a chimpanzee. Folia Primatologica, 61(2), 110–114.Google Scholar