Abstract
Many nonhuman animals produce facial expressions which sometimes bear clear resemblance to the facial expressions seen in humans. An understanding of this evolutionary continuity between species, and how this relates to social and ecological variables, can help elucidate the meaning, function, and evolution of facial expression. This aim, however, requires researchers to overcome the theoretical and methodological differences in how human and nonhuman facial expressions are approached. Here, we review the literature relating to nonhuman facial expressions and suggest future directions that could facilitate a better understanding of facial expression within an evolutionary context.
Many species of nonhuman animals have relatively static, immovable faces that do not have the capacity to change in relation to behavioural or environmental stimuli (Diogo, Abdala, Lonergan, & Wood, 2008). Most mammalian species, however, can produce facial movements (Diogo, Wood, Aziz, & Burrows, 2009) which form meaningful and adaptive components of the animal’s behavioural repertoire, and are often termed facial expressions. Yet the way in which human and nonhuman facial expressions are studied can differ greatly. In part, this is due to methodological constraints, but the study of human and nonhuman facial expression has also long been rooted in different theoretical disciplines, with scant communication between the two. Nevertheless, continuity between nonhuman and human facial expressions is palpable, and similarity in form and function is often explicit and measurable. A full understanding of facial expression (in terms of both proximate and ultimate causation, sensu Tinbergen, 1963) requires looking beyond humans, and toward closely (and not so closely) related species.
Nonhuman facial expressions are often studied as displays (sensu Huxley, 1914) where their adaptive function in terms of information transfer is of primary concern (Smith, 1977). There is a strong focus on the receiver, as signals are often moulded and constrained by the senses of the receiving individual (receiver psychology: Guilford & Dawkins, 1991). In contrast, human facial expressions are usually studied with the psychology of the sender in mind, stemming from Darwin’s early accounts of facial expression as an inevitable behavioural counterpart to felt emotion (Darwin, 1872). Despite some attempts to emphasise the importance of the communicative value of facial expression (Fridlund, 1994; Schmidt & Cohn, 2001), the study of human facial expression is sometimes almost synonymous with the study of emotion (e.g., Ekman, 1994), thus focussing on the feeling state of the sender. Such an association between felt emotion and expression is rarely assumed in animal research: emotion can be “inferred from behavioural changes and evolved communication signals. There exists no one-on-one relation between an emotion and ensuing behaviour, however” (de Waal, 2011).
Emotion is also often defined very broadly by animal scientists: “emotion can be thought of as a process that facilitates appropriate responses to a wide range of both internal and environmental situations” (Parr & Waller, 2007), or intrinsically tied to functional behavioural systems (Panksepp, 1998). Facial expressions are deemed to be predictive of the general motivation and/or tendency of the organism to engage in a series of actions (van Hooff, 1972), and so there is more interest in their role in social outcomes than in any underlying feeling state per se. This approach to facial expression in animal research has clear parallels to the behavioural ecology approach to human expression (Fridlund, 1994) and is probably also similar to the “action tendency” view of emotion (Frijda, 1986) where emotions are conceptualised as tendencies to engage in behaviour. Interestingly, the dominant evolutionary approach to human facial expression (that of basic, universal emotions: Ekman, 1994) seeks phylogenetic patterns, but does not take a similarly behavioural ecological approach.
Facial Expression in Nonprimates
Nonmammal species are rarely considered in the study of facial expression, but there are a few compelling examples that could be considered to be facial expressions (in a sense). The facial components of birds’ heads are not particularly mobile and they do not have facial muscles (Diogo et al., 2008), yet many species display various expressive signals which involve the face (Morris, 1956). These displays tend to involve movements of the crest and/or opening and orientation of the bill and are frequently used in the context of aggression or reproduction (e.g., Kumar, 2010). Some reptile species, such as rainbow skinks (Carlia jarnoldae), produce head-bobbing as a submissive signal during territorial fighting and courtship. Also, a throat flash is displayed by males by exposing their gular (throat) coloration while making a sweeping movement of the head (Langkilde, Schwarzkopf, & Alford, 2003).
Mammalian facial expressions appear much more varied and complex. For example, canid facial displays involve various combinations of movements of the ears, mouth, lips, and eye gaze shifts, often accompanied by vocalisations (Fox, 1970). Submissive expressions are characterised by the head being lowered, the ears flattened, and the lips horizontally retracted. Expressions of aggression involve the opening of the mouth, direct staring, and forward erection of the ears. These species also exhibit a “playface” which is produced during play, characterised by the lips being drawn back horizontally, the teeth being exposed, the ears more or less erected, and eyes partially closed. Brown bears (Ursus arctos) also produce a “playface,” where the mouth is open and relaxed, in addition to threat displays involving different orientations of the head and ears, variable degrees of mouth opening, and roaring vocalisations (Egbert & Stokes, 1976). Ungulates display a “flehmen” display, in which the head is raised, the upper lip drawn back extensively (sometimes puckered) exposing the teeth and gums, and the nostrils are wrinkled and closed (Moehlman, 1998). This complex facial display is strongly associated with olfactory communication and marking behaviour (Stahlbaum & Houpt, 1989). Miller (1975) described facial expressions in two species of pinnipeds (walruses, seals, etc.), such as submissive displays characterised by extreme erection of the vibrissae (whiskers), wide opened eyes, a hanging lower jaw with the lower lip and corners of the mouth retracted. During aggressive displays the mouth is less widely opened, the lips tense and the corner of the mouth not retracted, or drawn forward to form an “aggressive pucker.”
Facial Expression in Nonhuman Primates
One taxon of animals has received more attention than others in the study of facial expression—primates. There are several comprehensive descriptions of the facial repertoires of nonhuman primates (e.g., Andrew, 1963; Chevalier-Skolnikoff, 1973; van Hooff, 1967). The majority of social species exhibit some form of facial expressions, involving movements of the jaws, lips, ears, eyelids, and other facial landmarks. Some expressions, such as the bared-teeth display, where the lips are retracted and teeth exposed, are ubiquitous in the primate order. The specific meaning and context of the bared-teeth expression can vary between species (in relation to power asymmetries, see what follows), but it has long been argued that these similar expressions are homologous with each other, as well as with the human smile (van Hooff, 1972). Likewise, relaxed open mouth displays are frequently found in the repertoires of different primate species, usually associated with play, and thought to be homologous with human laughter (van Hooff, 1972) and the playful open mouth expressions seen in other mammals (e.g., canids: Bekoff, 1974; bears: Egbert & Stokes, 1976; rats: Panksepp & Burgdorf, 2003). Some facial expressions are unique to certain species, or a group of related species, and do not have obvious counterparts in humans or other animals. The “lipsmack” expression, for example, is a dynamic display found in many old-world monkeys, particularly macaques and baboons, but not apes or new-world monkeys (van Hooff, 1967). “Lipsmacking” is a dynamic display where the lips are protruded and smacked together rapidly (often producing an auditory element), and is associated with affinitive behaviours (Easley & Coelho, 1991).
The overarching aim of many scientists interested in facial expression, particularly the early pioneers (Andrew, 1963; Chevalier-Skolnikoff, 1973; Darwin, 1872; van Hooff, 1967), was to draw explicit comparisons between species, in an effort to trace homology and understand the phylogenetic relationship. Thus, there have been several key studies examining the extent to which similar appearing facial expressions are underpinned by shared anatomical structures (i.e., facial musculature: Bolwig, 1964; Huber, 1931), a factor that may reveal homology (Preuschoft & van Hooff, 1995). Recent dissection studies (e.g., Burrows, Waller, Parr, & Bonar, 2006; Diogo et al., 2009) and intramuscular stimulation studies (e.g., Waller, Parr, Gothard, Burrows, & Fuglevand, 2008) have reported extensive similarity in the form and movement of facial musculature in various primate species, suggesting they have very similar capacity for facial movement.
Interestingly, species of primate that we might not expect to need extensive facial musculature still exhibit great similarity to other primate species. Hylobatid species, for example, show similar facial muscles and facial movement capacity to other primate species (Burrows, Diogo, Waller, Bonar, & Liebal, 2011; Waller, Lembeck, Kuchenbuch, Burrows, & Liebal, 2012) despite living in small family groups; orang-utans (Pongo spp.) show similarity despite having dispersed, often solitary social organisation (Caeiro, Waller, Zimmerman, Burrows, & Davila Ross, 2012); and galagos (Otelemur spp.) show similarity despite being nocturnal (Burrows & Smith, 2003). Thus, these species have minimal opportunity for facial communication. Extensive facial musculature could, therefore, be the ancestral condition, possibly evolved for some other purpose, but has been co-opted for facial expression as an exaptation (a trait selected for one function, but later used for another). Alternatively, these species may be using facial expression in particularly subtle ways that do not necessarily correlate with frequency of social interaction.
Recently, scientists have made anatomically based coding systems for explicit and quantifiable comparisons between species, based on a system used in humans (Facial Action Coding System, FACS: Ekman & Friesen, 1978; Ekman, Friesen, & Hager, 2002). FACS is an observational coding scheme to classify the facial movements associated with individual (or combined) facial muscle contractions. Users learn to recognise the surface movements associated with each muscle contraction, and can thus describe facial expressions in terms of their component movements (as opposed to meaning or interpretation). Systems have now been developed for chimpanzees (Pan troglodytes; ChimpFACS: Vick, Waller, Parr, Smith Pasqualini, & Bard, 2007), rhesus macaques (Macaca mulatta; MaqFACS: Parr, Waller, Burrows, Gothard, & Vick, 2010), hylobatids (GibbonFACS: Waller et al., 2012), and orangutans (Pongo spp.; OrangFACS: Caeiro et al., 2012). Paul Ekman and colleagues created the original FACS in an effort to standardise and objectivise comparisons between people, heeding Darwin’s observation that “the study of expression is difficult, owing to the movements being often extremely slight and of a fleeting nature. A difference may be clearly perceived, and yet it may be impossible, at least I have found it so, to state in what the difference consists” (Darwin, 1872, p. 13).
Using ChimpFACS, morphological comparisons have now been made between human anger and the chimpanzee bulging-lip face, human smiles and the chimpanzee bared-teeth display, human laughter and the chimpanzee relaxed open-mouth display, and screams in both species (Parr & Waller, 2007). The only prototypical chimpanzee expression that does not seem to have a morphological counterpart is the pant-hoot, which is used in contexts of excitement (Parr, Cohen, & de Waal, 2005).
One aspect of nonhuman facial expressions that has, to date, been examined purely in nonhuman primates and not in other animals, is the underpinning cognitive mechanism. For example, using a computerised match-to-sample paradigm, Parr, Hopkins, and de Waal (1998) showed that chimpanzees can discriminate between different conspecific facial expressions. Such studies demonstrate that these facial expressions are perceived as meaningful, discrete stimuli to chimpanzees. Parr, Waller, and Heintz (2008) conducted similar experiments using computerised model chimpanzee faces morphed into different anatomically correct facial expressions, based on ChimpFACS. The ability of the chimpanzees to discriminate between expressions was analysed in terms of the component movements each expression contained, which demonstrated that chimpanzees (similar to humans) were using both configurational and featural cues in facial expression processing. Similarly, in an attempt to understand the “meaning” of facial expressions, Parr (2001) asked chimpanzees to match facial expressions to images of positive and negative items (e.g., food vs. veterinary procedures). To date, this is one of very few studies explicitly addressing how nonhuman primates understand facial expressions in terms of positive and negative associations.
Physical, Ecological, and Social Correlates of Facial Expression
Although much of the work on the evolution of facial expression focuses on homology (i.e., identifying structures/behaviours that have arisen through common descent), there is also value in exploring analogy (i.e., identifying similarities that have arisen in response to similar selection pressures). Given the problem of phylogenetic inertia (that nonadaptive traits can be retained through descent despite being suboptimal), it could be argued that analogy is more enlightening in terms of function than homology. Inevitably, it is sometimes hard to disentangle the two, but evaluating the importance of social and ecological variables in the evolution of facial expression in animals allows us to identify those factors which influence it (Egbert & Stokes, 1976; Fox, 1970; Miller, 1975).
The most comprehensive analyses come from Dobson’s (2009a, 2009b) studies of facial mobility in anthropoid primates, where Dobson uses facial mobility as a proxy for complexity in facial expressions (Changizi, 2003). Using the human FACS (Ekman et al., 2002) applied to the different primate species under study, Dobson (2009a) showed that body size was positively correlated with facial mobility, even when taking phylogeny into account. Because species with larger individuals are better equipped to process subtle facial movements (Kiltie, 2000), this allometric constraint on facial mobility is likely to be exercised on its perception. Thus, the adaptive value of an increased facial mobility, and hence a potentially increased range of facial expression, might be significant for species with large individuals but negligible for species with smaller individuals (Dobson, 2009a).
Some species deviate from this allometric pattern (e.g., gorillas [Gorilla gorilla] are twice the size of chimpanzees but produce the same number of facial movements). Because social and ecological factors are important selective pressures driving the evolution of social cognition and communication (Dunbar, 1993; Maestripieri, 1999), they could explain the observed discrepancies. In a follow-up study, Dobson (2009b) examined this hypothesis by examining the impact of group size and ecology on facial mobility. Phylogenetically informed correlations, controlling for body size, revealed that facial mobility increases in parallel with group size. Facial expressions and communicative signals in general are mainly used to establish and maintain social relationships, which is crucial for group cohesion. As groups become larger, sustaining group cohesion becomes more challenging and consequently, behaviours used to facilitate group cohesion will be selected for (Dunbar, 1993; Maestripieri, 1999; McComb & Semple, 2005).
One would expect that species living in open areas would rely more heavily on visual communication than arboreal species because of visibility constraints. Surprisingly, in Dobson’s study (2009b), species’ ecological milieu did not significantly correlate with facial mobility. One explanation is that arboreality is correlated with body size (the larger species tend to be more terrestrial), or simply, a dichotomous classification (arboreal vs. terrestrial) is not precise enough to reveal a significant result. It is also possible that facial expression is still adaptive in arboreal environments in close-range social interaction.
The nature and quality of social relationships within a species (i.e., the social style of a species) can also impact the morphology, function, and dynamics of facial expressions (power asymmetry hypothesis of motivational emancipation: Preuschoft & van Hooff, 1995, 1997). Tolerant species are characterised by weak influence of dominance and kinship on patterns of aggression and affiliation (Thierry, 2007). In these species, facial displays of submission and affiliation (e.g., the silent bared teeth and the relaxed open mouth) converge functionally and morphologically toward a highly bidirectional affiliative display (e.g., the open mouth and bared teeth in tonkean macaques [M. tonkeana]: Preuschoft, 1995). On the other hand, when power asymmetries are strong (e.g., rhesus macaques), individuals need to signal their subordinate status to higher ranking individuals via a salient and unambiguous signal in order to avoid the cost of an unnecessary conflict. In these species, the signal is strictly unidirectional (directed only from subordinates to dominants).
Increased sociality and group cohesion are also known to lead to increased vocal complexity (rodents: Blumstein & Armitage, 1997; nonhuman primates: McComb & Semple, 2005), and this could well apply to nonvocal signals as well. In fact there is evidence from comparative studies of macaques that more tolerant species possess an increased behavioural repertoire and more sophisticated social interactions compared to more despotic species (Maestripieri, 1999). In despotic societies, hierarchy and kinship constrain social relationships and so a despotic social style gives little room for elaborate affiliative interaction (e.g., by reducing the number of potential partners). Thus, ambiguity needs to be reduced (because of the risks of aggression), which can explain a reduced and/or more rigid communicative repertoire.
Current Limitations
Studies of animal facial expressions are complicated by a number of methodological problems. First, facial expressions are sometimes hard to disentangle from vocalisation and gestures. Without extremely detailed video analysis, it is difficult to differentiate between articulatory gestures necessary to produce the accompanying vocalisation, and the mouth movements that are part of a ritualised display (Partan & Marler, 2005). This issue poses a particular problem when scientists try to classify communicative signals according to a single modality, which is the dominant and classical approach (Slocombe, Waller, & Liebal, 2011).
Second, facial expressions are graded signals, meaning that their appearance varies to reach peak intensity, and can oscillate between different configurations. While some authors associate this gradation with variation of the senders’ motivational state (Preuschoft & van Hooff, 1995), the perception of graded visual signals has received little attention and thus, arbitrariness in the classification of facial display is difficult to avoid (but see Parr et al., 2005; Parr et al., 1998). Graded vocalisations can be perceived as discrete signals in some primate species (e.g., Green, 1975), as can facial expressions in some species (Parr et al., 2005).
Third, facial expressions may also merge into one another to produce unritualised blends. In chimpanzees, the context in which a blended signal occurs corresponds specifically to only one of the parent expressions and suggests that the blending reflects conflicting motivation from the sender (Parr et al., 2005). The extent to which these blended facial expressions demonstrate flexibility in production and perception, however, is relatively unknown.
Most of the detailed studies that have enabled researchers to identify and tackle these issues and thus expand the field of animal communication have focused on primates. However, there is a priori no reason to think that the importance of multimodal communication and the use of graded and blended signals are irrelevant to nonprimate species. Much more work is therefore needed to obtain a better idea of the complexity of animal facial displays.
Conclusion and Future Directions
Nonhuman animals produce facial expressions in social circumstances; few scientists debate this statement. What they communicate, however, is more contentious. There is a division between researchers studying facial expressions as an emotional display (Ekman, 1993) and those considering facial expressions as adaptations designed to allow receivers to predict the behaviour of senders (Smith, 1977) and/or to allow senders to manipulate receivers (Guilford & Dawkins, 1991). These two perspectives are not mutually exclusive, but rather represent two different levels of analysis: causation and function. A focus on both, of course, is necessary to fully understand the phenomenon (Tinbergen, 1963). We suggest three future directions that could facilitate greater communication between human and nonhuman facial expression research. First, both proximate mechanisms and ultimate function could be considered simultaneously. Second, facial expression exists in many species other than nonhuman primates, and so further examination of this could reveal a broader evolutionary context. Third, facial expressions are part of a multimodal communicative and expressive system, and if we studied them as such, we may demonstrate continuities between species that have been previously overlooked.
