acta ethologica

, Volume 9, Issue 1, pp 1–14

Universals and individuality in facial behavior—past and future of an evolutionary perspective


    • Department of Engineering & Natural Sciences, Biology SectionUniversidade Lusófona de Humanidades e Tecnologias

DOI: 10.1007/s10211-006-0010-x

Cite this article as:
Gaspar, A. acta ethol (2006) 9: 1. doi:10.1007/s10211-006-0010-x


Multiple dimensions of facial expression research are reviewed with emphasis in the study of species’ homology in both facial action single components and patterns or gestalten. Research over the last 30 years has revealed human universal patterns and some cross-species gestalten that appear as homologies. Recently, researchers are looking into the importance of individuality markers in facial behavior, which can play an important role in the fitness of individuals living in complex social systems and thus might be traceable in their evolutionary path. Chimpanzees, bonobos, and humans were in the center of most of this research with research results pointing toward homologies within the HomoPan clade.


Facial expressionBehavior evolutionHuman ethologyChimpanzeeBonobo


Facial behavior is a crucial source of information on what and how individuals are feeling (Ekman 1984) and on what they are likely to do next (Hinde 1985a,b). Recent works even suggest that it is a good predictor of individual differences and of future interactive events (Cohn et al. 2002; Gottman et al. 2004; Grammer 2004).

If we speak of facial behavior at all, it is because many organisms, especially primates, have extraordinarily mobile faces that shift from a pattern of combined signals to another in a nonstop flow that displays complex information at an unprecedented rate (if we disregard human language). This continuity makes facial behavior a hard topic for study, posing many methodological difficulties. This is probably the main reason why it was almost dormant for some 60 years after Darwin’s The Expression of the Emotions in Man and Animals and has advanced at a slow pace since the 1960s.

In 1962, van Hoof (1972) reawakened the subject and elaborated a comprehensive model for the evolution of facial expressions deeply rooted in mechanisms explained by classical Ethology such as ritualization. His observations and those of other primatologists and human ethologists since the sixties have shown how this mobility varies from simple repertoires of basic mouth movements (as in reptiles) to the complex multidimensional mimics of humans or chimpanzees. These findings will be reviewed in Section 2.

There is currently a debate on whether facial behavior conveys information about the internal state of the individual (emotions, moods, or motivational states) or rather indicates only his/her interactive intentions regardless of inner states (see Zajonc 1985; Fridlund 1991, 1994). This, perhaps, will be discussed more apparently than really conflicting views, which will be reviewed in Section  3.

It was repeatedly reported in humans and more recently found in chimpanzees, bonobos, and other Anthropoid primates that facial behavior varies interindividually within the same context, be it an experimental condition or a natural situation in a consistent (intraindividual) manner. These still incipient findings will be reviewed in Section 5. Certain universals in facial behavior seem to overcome these idiosyncrasies and models based on the crucial adaptive value of universals are not hard to sustain and will be discussed in Section 4.

Although the relation between universality and evolution of facial behavior is tight, hardly ever becoming a source of doubt or debate, the relationship between individuality and evolution in facial behavior is not so immediately obvious. Only part of the causal nature of this interindividual variation can be explained by culture, thus, other levels of explanation are required. Personality and social attraction are properties of the individual that were identified as temporarily consistent and studied from early age. For both, there are interesting preliminary correlates with individual aspects of facial behavior. The adaptive value of such facial action predictors of personality and popularity is now becoming a subject of increasing interest in the study of behavioral evolution and new developments are expected from currently ongoing research. Such developments may shed some light on the emotions/social intentions of function debate. This will also be discussed in Section 5.

Mobility, universality, and individuality, communication of action tendency vs emotions, are by no means sufficient to explain all aspects that may have played a role in the evolution of facial behavior but it seems certain that they all evolved intertwined as parts of the solution to many adaptive problems created by complex social scenarios.

Evolution of facial mobility

Darwin (1872) was the first proponent of the evolutionary continuity of facial behavior with a treaty on The Expression of Emotions in Man and Animals where the existence of inborn facial behavior patterns and a similarity of form and function within the facial behaviors of mammals are points of emphasis. Almost a century later, van Hoof (1962) called our attention again upon the fact that facial expressions have a phylogeny and that they did not all begin with communication as their primary function. The original function of many facial actions was feeding, soliciting food, suckling, biting, and chewing. Ritualized, they acquired a signaling function. Further in time, many acquired more variation in form and meaning.

van Hoof (1972) and more recently Preuschoft (1992) focused two facial gestalten “families”—the play face and the silent teeth baring face (generally known as SBT) across species of Macaca, Papio, and Pan, finding both analogies in some cases and homologies in others to human facial expressions as well as an increasing divergence of form and function from Macaca to Pan and from Pan to Homo in both the play face and the SBT (van Hoof 1972). de Waal (1988) conducted a study of the behavior of Pan paniscus in which six major facial behavioral patterns were described. A variety of behavioral studies of the common chimpanzee reported a repertoire of different facial expressions that ranged from 6 to 20 (van Hoof 1962, 1967, 1973, 1981; Goodall and van Lawick 1968; Goodall 1986; Berdecio and Nash 1981; Chevalier-Skolnikoff 1982). Description of bonobo facial behavior was resumed by Gaspar (2001ac) resulting in a 45-gestalten facial ethogram. The same study provided a 52-expression facial ethogram for common chimpanzees Gaspar (2001c). These repertoires were larger than the largest ever published ethological inventory of human (infant) facial expressions (Young and Decarie 1977), which consisted of 42 expressions. This large number of Pan expressions was somewhat expectable because chimpanzees and bonobos were observed in a much wider range of contexts than children, no matter how naturalistic observation settings were.

It was generally admitted that humans are the most expressive of all living species (e.g., Chovil 1997) and that primate facial behavior has evolved toward increasing mobility with humans having the most mobile faces of the Primate order, except for ear movements (e.g., Huber 1930/1972, 1931; Andrew 1963a,b). Primates, if we leave the Lemuroidea out, have certainly more mobile faces than other mammals, but face mobility does not increase in a consistent manner as we approach more recent phylogenetic groups as suggested by Huber (1931). Not only is there substantial variation between the studied captive communities possibly attributable to culture (Gaspar 2001a,c) but group size influences context opportunity thus creating more opportunity for repertoire diversity in large groups. More importantly, different primate groups “favor” different areas of the face (e.g., Sonntag 1924; Preuschoft 2000), for example, baboons and mandrills have an extraordinary mobility of the forehead and scalp whereas humans do not. Chevalier-Skolnikoff (1973) noted that the mobility of the mouth area was notable in nonhuman primates but not that of eyebrows, unlike humans. Of course, physiognomic characteristics such as much less color contrast between eyebrow ridges and forehead for adult chimpanzees and bonobos than for Caucasian people make it harder from a distance to notice brow movements. Also, chimpanzee brow ridges are so prominent and wrinkled that eyebrow movements are harder to detect unless the observer is at short distance. Thirty years ago, when the study of the evolution of facial expression was jumpstarting, researchers had no cameras with the zoom power that today’s video cameras do. Nonetheless, part of the human brow actions was already described by van Hoof (1962, 1967) and Goodall and van Lawick 1968; Goodall 1986). Currently, all brow action units (AUs; isolated movements of face muscles or muscle combinations that contract together) described for humans by Ekman and Friesen (1978) were reported both in chimpanzees and bonobos (see Figs. 1 and 2): AU 4 (brow lowering) was present early on in Goodall and van Lawick’s (1968) reports of the Gombe chimpanzees and the other AUs were reported in a four-colony study of chimpanzees and bonobos using video-recording and still and moving picture AU decoding analysis (Gaspar 1999, 2001ac).
Fig. 1

Bonobo Lucy, from the Columbus colony (Ohio, USA) displaying Lifted Eyebrows (AUs 1+2)
Fig. 2

Bonobo Lucy, from the Columbus colony (Ohio, USA) displaying a frown, eyebrows pulled down and drawn together (AU 4)

van Hoof (1962) introduced the practice of discriminating AUs in the study of nonhuman primate facial behavior followed by other researchers (e.g., Hinde and Rowell 1962; Weigel 1979), the quantification of these AUs or the use of a universal human-non-human primate coding system has only recently given its first steps (Gaspar 2001c, 2004). This explains why efficient comparisons between studies are long overdue. The absolute need to use a universal human-non-human primate coding system was earlier pointed out by Jolly (1985) to overcome many of the problems that have discouraged researchers to tackle the difficult questions of facial behavior. One major problem is indeed cross-species comparison of “expressions” that often are homologous but are not quite exactly composed of the same gestalt, differing sometimes in the intensity or presence of one AU. One such example is the chimpanzee cryface, which is identical to the human cry face but in chimpanzees, it was noted to occur (Jolly 1985) only very briefly in the transition from the whimperface to the bared-teeth screamface (Goodall 1986) that typically occurs in infant chimpanzees’ tantrums. So from a formal perspective, the human cry face is identical to a transition gestalt in chimpanzees, not in the context related with typical intensity and duration facial expression [one exception though, was recently observed in one Arnhem chimpanzee, Amber, an adult female who after the death of her daughter Anthé started producing a cry face not transiting to any other gestalten nor accompanying cry (personal observation, February 2004)].

Another major pillar of the discussion on human-non-human primate expressiveness was the Duchenne smile, an “honest” hedonic smile described by and named after the anatomist Duchenne de Boulogne (1859, 1990), and its important expressive element “lip corners pulled upwards and cheek raise” [AU12+AU6 in Ekman and Friesens’s facial action coding system (FACS)]. According to Huber (1931),1 the smile was performed by the muscle risorius, which he reported as absent in chimpanzees and in many “non-Caucasian” humans. Duchenne (1859, 1990) found the pulling of the lip corners upward is produced by the contraction of the muscle Zygomaticus major, an observation that was confirmed in the twentieth century (Ekman and Friesen 1978). This muscle is shared by humans and chimpanzees (e.g., Huber 1931; Sonntag 1924). Risorius is also shared by both species, but increasing evidence shows that its presence varies interindividually (e.g., Seiler 1973; Lever 1987) before it varies interspecifically.

Aside from this muscular-based discussion on what AUs can and cannot be present in Pan and other Primate groups, behavioral studies have repeatedly mentioned the pulling of lip corners upwards in nonhuman primate facial expressions. Table 1 summarizes those findings in studies that have either identified “smiles” or, more unambiguously, describes variation in smiles that contemplates lip corner lifting. This table suggests that AU 12, although present in Macaca and Gorilla, is restricted to socially tense and aversive situations and only in Pan does it widen its contextual association to cover hedonic contexts, much as in the way it happens in humans. Because Macaca and Gorilla data are scattered, it is not really known if this context generalization and form diversification started earlier in Primate evolution or at least evolved independently in some species.
Table 1

Reports of nonhuman primate “smiles” or grin faces in their most common contexts




Social tension


Aggres. Anger




Hinde and Rowell (1962)








van Hoof (1962)




Play F

de Waal (1987)






Thierry et al. (1989)






Preuschoft (1992)




Kalin (1993)





Maple (1982)


Transition SBT–BZ


Open grin


Play F




Schaller (1963)


Open grin

Open grin


Transition pen grin, BZ


Play F


Chevalier-Skolnikoff (1982)



Play F


Laughing F

van Hoof (1962, 1973)




Play F

Goodall (1986, 1988)




Play F

Gaspar (2001ac)








Play F


Laughing F

CMS Closed-mouth smile (lip corners pulled upwards, no teeth visible), SBT silent bared-teeth face (horizontal closed grin), BZ bozo smile (SBT with fully exposed gums and lip corners pulled upwards), VBT vertical bared-teeth face, GRIN formal subdivisions of a grin are not described, Open grin same as GRIN but with open mouth, ROM relaxed open-mouthed face (relaxed face with open mouth and jaw dropped), Play F play face is a ROM with lip corners retracted and pulled upwards, and Laughing F a play face with the upper teeth exposed

The adaptive value of mobility only makes sense in primates that live in habitats with good up close visibility of conspecifics as Redican (1982) rightly pointed out. Terrestrial primates, more than arboreal primates, have varied up close eye-contact with other group members and in socially complex societies, fine nuances of facial information may prove to be of crucial importance in such interactions as threats, coalition formation, and so forth. It is yet to be found if highly arboreal Hominoids like gibbons are more or less expressive than terrestrial Cercopithecoids like baboons. But if they are more expressive, we may be just dealing with homology because not even the complex society hypothesis is met because gibbons live in small-family units.

The diversity of subtle signals may relate to the architecture of social relations in a species, a highly political primate such as a human or a chimpanzee with diverse and frequent social tensions to manage needs to signal many different social messages and has to be able to decode many clues from interactors to make important social decisions. So, one would expect very aggressive species to display more subtle (therefore diverse) signals in their faces, i.e., to show a higher facial mobility, and to have a better decoding ability than less aggressive species, at least in situations of social tension. An example of this relation in humans is the monitoring of the partner’s face in couples with stable and less conflicting happier couples displaying much less frequent monitoring behaviors of the partner’s face than other subjects in more conflicting less satisfied pairs, or in couples that are just at the beginning of the relationship and know each other less (Noller 1980).

The function of facial behaviors

Although ethological studies, mainly in the seventies, have proposed functions for specific facial behavior patterns, the general function for nonverbal behavior and facial expression is currently under debate. One side of the rope is pulled by the supporters of a perspective that became known among its opposers as the Facial Expression Program (Russel and Fernández-Dolls 1997), which assumes that in at least a few facial expressions, an underlying emotional state is present and the facial action is part of the emotional response (Izard 1971, 1994, 1997). On the other side are the defensors of a recent theory rooted in Behavioral Ecology (Fridlund 1992, 1994, 1997) that states facial expressions occur for an audience, i.e., socially, and to generate certain responses from the receiver. In the 1980s, Hinde (1985a,b) already stated that the function of facial displays was that of informing the receiver about the sender’s future interactive intentions.

The two views, social interaction vs emotion, seem to collide when we consider altogether studies that show the presence of an audience effect (supporting the “social” theory) and the absence of that same effect (supporting the emotion “program”): For instance, in a controlled set of experiments, the faces of individuals reacting to emotion content in films were themselves recorded in social and solitary condition and showed emotional reactions even in the solitary condition (e.g., Ekman et al. 1990; Davidson et al. 1990). There was, however, an audience effect for Japanese subjects but not for Americans, which was interpreted to result from the Japanese more rigid display rules for the public display of emotion.

Alan Fridlund provided an explanation for the occurrence of facial expressions (with emphasis on smile) in private conditions, an “imaginary audience” effect (Fridlund 1991, 1994). According to this theory, subjects in the solitary film watching condition react as if someone else is in the room or as if people on a TV screen could see them back—these are their “imaginary interactors.”

Chovil (1991) reported an increase in the frequency of facial expressions proportional to the increasing visual contact among two interactors as one reported the other a dramatic experience. We should perhaps also consider an “empathy effect” triggering an intensification of the emotional reaction and thus of corresponding facial expression. In testing the theory of the imaginary interactor, it would also be interesting to compare facial behavior when the subject is facing the TV set (or other emotion relevant stimulus) and when he/she cannot possibly have visual access to the stimulus source if it involves a human image. When an audience effect is still there when you have your back facing the TV set, you are not behaving as if the person on TV is real and “manipulable” by your signs but that is expectable if you are facing that person. You may have an automatic display of facial behavior when facing someone else or when believing someone else is in the room with you, but that seems much more unlikely if you are just not in visual contact with the TV actor, which you anyway know is not really there.

As to the audience effect for a real audience condition was studied only regarding adult smiles during bowling (Kraut and Johnston 1979 in Fridlund 1994) and baby expressions during object exploration in the presence of their mothers (Jones et al. 1991). In none of these studies were different types of smiles formally discriminated, i.e., authors took smiles as one expression and there is evidence that formally distinct smiles associate with different social functions and internal states (e.g., Ekman et al. 1988; Ekman and Keltner 1997). In the baby study, babies did smile throughout most of the experiment increasing when the mother looked at the baby and decreasing when she was looking away. This provided good support to the “real audience effect.” Indeed, smiles have multiple functions and causes and certainly some of those are socially relevant. Functions that were not assessed in the above experiment were shown in developmental studies, which found that babies smile during small cognitive achievements like recognizing familiar objects (Schneider 1997). They also do it in the womb, as recently developed 3-D scans reveal (Campbell 2003), providing support to the emotion program for at least some types of smiles. The frequency of smiles increased in the adult bowling player of the other experiment when the player turned and looked at his/her friends, but smiles also occurred without such eye contact when the game was good, albeit less conspicuously. Two factors, emotional and social, seem at play here although the social seems to impose at least when frequency is the measure. Using a different measurement unit, onset and offset timing of Duchenne smiles, Schmidt et al. (2003) found almost no differences between the social and solitary conditions.

So are the two perspectives, the emotion program and the behavioral ecology inspired theories, really in diverging paths or are we just looking at two different sides of the problem? Or rather, can they be discussed not as conflicting opposite explanatory theories but as complementary pieces of the same puzzle?

Going back to Hinde (1985a,b), a facial display would express a conditional action tendency of the sender, i.e., the actor’s facial behavior would depend upon the receiver’s reaction and it would be selected by the behavior elicited in the receiver. Facial expression is then an instrument of social manipulation by means of which the sender attempts to influence the receiver’s conduct (Dawkins and Krebs 1978). The conveyed message can be either true or false.

Thus, a facial expression can occur during the experience of an emotion and can be an honest sign of that emotion and simultaneously or separately serve in a process of negotiation between two individuals (Preuschoft and van Hoof 1997), e.g., an angry face can simultaneously communicate the emotion anger and signal a conditional action tendency of the type “if you do not withdraw or stop what you are doing I will attack you” in the sense of conditional action tendency formerly proposed by Hinde.

There are numerous examples of the emotional, social interaction and double functions in facial expression. For example, the Duchenne smile can signal multiple kinds of messages, emotional (happiness), action tendency (“I will not attack now, I am in a good mood”), and information about the status of the relationship with the interactor (“I am happy in this interaction”). These three messages can be read simultaneously in one and the same interaction with no obvious disadvantage for the sender in any of them. From the sender’s perspective, they are action tendencies related to an internal state but to the sender they can be more than that depending on the context. The disadvantage to the sender can occur in certain situations where the expression is highly unusual, for example, a Duchenne smile in a funeral. This is when control and deception come in (e.g., Ekman and Friesen 1975) and cultural display rules,2 which are all assimilated within our own cultures, come to the rescue (e.g., Ekman and Keltner 1997). Also, the facial action gestalten known as anger universal expressions signal anger and this emotion has a well-known pattern of physiological reactions that includes the facial muscles response (e.g., Ekman et al. 1990; Davidson et al. 1990; Dimberg 1982; Dimberg and Thunberg 1998). These expressions also indicate an action tendency, generally conditional in which a threat is implied; if the receiver does not do what is required or expected by the sender, the later can retaliate. In both cases, there is no doubt that the expression(s) communicates a motivational state or an action readiness state (anger and being ready to attack if conditions are not met).

One can argue that in bluffing and active deceiving, this synthesis of the emotional and social approaches is not validated—sender and receiver have opposite interests. Well, no one ever said all expressions are emotional although they all have social consequences for as long as an interactor is nearby and in that sense they all portray action tendencies and are therefore ultimately social.

This discussion basically resumes an old Behavioral Ecology discussion that started in the 1970s on whether signals are true or manipulated (see Krebs and Dawkins 1984; Krebs and Davies 1993). Manipulation is a distinct concept from that of deceiving inasmuch as manipulation indicates only that the signal is advantageous to the sender and affects the nervous system of the receiver to release a reaction that benefits the former. It does not mean the sender is intentionally deceiving about his/her internal state (although that option is contemplated). Manipulation simply means that the sender exploits the perceptive and psychological predispositions of the receiver (Krebs and Dawkins 1984). In most cases, manipulation involves a message that is not disadvantageous to the receiver. Incidentally, the receiver’s reaction may benefit not only the sender but the receiver as well. Not incidentally, there may even be cooperation in sending and decoding the signal and when that happens, the signal tends to be a more subtle one rather than a conspicuous display because sender and receiver are in tuned; the receiver is a more efficient decoder and the sender saves in signaling costs. These are honest signals for which there are numerous examples in animal communication (see Krebs and Davies 1993).

In human facial communication, at least a few basic emotional expressions are within this definition of honest signaling (however manipulative in the formerly explained sense); children are efficient decoders of expressions from very early on (e.g., Charlesworth and Kreutzer 1973; Philippot and Feldman 1990; Nowicki and Mitchell 1998) and the same emotion-specific neural substrates seem to be activated for both sending and decoding (e.g., Ekman et al. 1983; Ekman and Keltner 1997; Phillips et al. 1997). And, although deceiving occurs abundantly in human facial expression, a significant ability to detect it also occurs (e.g., Ekman and Friesen 1975; Frank and Ekman 1997; Ekman, Friesen and O’Sullivan 1988). It is not so much a matter of “true” or “false” smiles but of all smiles having distinct truthful messages to convey—from the overwhelming noncontrolled happiness to the cultural, social indicator of a coping “miserable smile” (Ekman and Friesen 1975; Ekman, Friesen and O’Sullivan 1988). So, in this sense, unless you lie to yourself, you never lie with facial behavior (Grammer 2004). Actors often comment that when they portray emotions and feelings well, they “become” their characters and they actually feel what the characters feel.

So, the simple fact that facial expressions (spontaneous or controlled) convey information that is likely to elicit a certain response in a receiver speaks in favor of the interactiveintention” conveying message function whenever the individual is at least in visual contact with others. This would be the adaptive value of facial behavior. The message does not need to be intentionally conveyed. Adopting the interactive intention function approach or the closely related action-tendency function proposed by Hinde (1985a,b) leaves the undeniable fact that internal states are always there and that they have physiological correlates with facial actions unshaken (for review, see Zajonc 1985).

However, not all facial expressions are adaptive and the fitness added by a facial behavior is mediated by various factors ranging from intensity to context (for reviews, see Schmidt and Cohn 2001; Patterson 2003). It may be very disadvantageous to leak information about internal states in some occasions and therefore control comes into the scene. However, despite such attempted control, in some cases internal states do show in facial behavior, which can leak undesired information to the sender but can be advantageous to the receiver (Ekman and Friesen 1975, 1982; Ekman, Friesen and O’Sullivan 1988). Different degrees of control over facial actions seem to apply to different internal states and contexts in humans (e.g., Buller and Burgoon 1998; Jakobs et al. 1999).

The adaptiveness of a facial expression also depends on the timing and social context of use. Lakin et al. (2003) review evidence on how the mimicry of nonverbal behavior promotes affiliative relationships and increases after an individual is ostracized.

The functional model of nonverbal behavior (Patterson 1983) proposes that a behavior pattern can serve more than one function even in communication. However, it does not attempt to explain its evolutionary history. In fact, humans and other primates use, in many cases, the same facial action in very distinct acts (Preuschoft and van Hoof 1997; Gaspar 2001ac) and face processing occurs at various levels, e.g., content, personality assessment, age, gender, attractiveness, and many others (e.g., Ekman 1978; Gaspar 1994; Zebrowitz 1997). These other levels of information from facial behavior will be the subject of Section 5.

Facial behavior universals and homology

Early on in the study of facial expression universals, species universality was assumed for nonhuman primate species and thoroughly scrutinized in humans because cultural anthropologists argued that culture could split human facial repertoires into many possibilities with different arbitrary meanings for sometimes the same expression (for review, see Ekman and Keltner 1997) and psychologists wanted to find out which expressions were universal and where did culture actually influence (or even determine) facial behavior. Eibl-Eibesfeldt provided the scientific community with very unequivocal data by documenting many human facial action universals (Eibl-Eibesfeldt 1989; Grammer and Eibl-Eibesfeldt 1990) and the occurrence of some of those facial actions in congenitally blind children in the exact same functions as they were in other people (e.g., Eibl-Eibesfeldt 1970, 1973). These blind children who smiled, laughed, cried, gave angry frowns, and displayed surprised faces in these emotional contexts just as other children of the same age, were, by virtue of their condition, the perfect control for the influence of social learning. Ekman et al. 1987 showed that the phenomenon of universality worked both for coding and decoding these universal facial gestalten. It was settled then that the universality of such a collection of facial gestalten meant that at least for those, there was a biological basis, a genetic coding, and therefore an evolutionary history. Investigating the phylogeny of these human facial expressions was the major force that pushed forward the research on nonhuman primate facial behavior.

Most of the emphasis was on full displays, or gestalten, rather then on isolated facial actions (or AUs such as they are presented in the FACS), but more recently, the componential approach to facial expression has gained strength (e.g., Smith and Scott 1997) as empirical data connects biological meaning more precisely to single actions than to full expressions (which are not discrete behavior units and are thus very hard to isolate in the natural flow of behavior).

Identifying homology in facial behavior involves finding, within a monophyletic group, species continuity both in the formal aspects of the behavior and in its neuromuscular substrate. Formal resemblance of a facial action gestalt is a better indicator of possible homology if it is a complex multiaction gestalt (for review of homologizing criteria for facial expressions, see Preuschoft and van Hoof 1995). Such resemblance between taxa is known today for several such gestalten.

There is evidence of homology in the neuroanatomical substrate of facial expressions in the Hominoidea (Rosenzweig et al. 1996; Sherwood et al. 2003) and in brain areas related to the processing of faces (e.g., Semendeferi et al. 1998; Nimchinsky et al. 1999; Perl et al. 2000; Sherwood et al. 2003). Neurons that are triggered in the presence of specific emotion expressions (and their lateralized location pattern) also seem to have old evolutionary roots and can be found both in humans and macaques (for review, see Hauser 1996).

Also, facial behavior related to emotion seems to be hardwired in the brain, not only in the action process level but also at the reaction and decoding levels. For instance, neuroimaging studies strongly suggest that brain areas involved in the processing of an emotional facial expression (e.g., fear or disgust) are those that are involved in assessing other kinds of stimuli that induce the same type of emotion portrayed in that face and that different emotion expressions also are perceived in distinct neural paths (Phillips et al. 1997).

Looking at formal similarities between species comes after recognizing within species universals—species typical facial behavior. So far, the findings of primate facial expression homologies are for a very narrow range of expressions. This is not surprising though because hard evidence for universality exists only in the well-studied case of human facial expression for seven full-facial gestalten (e.g., Ekman 1973; Ekman et al. 1987; Ekman and Keltner 1997) and for an eyebrow action pattern (Grammer et al. 1988; Eibl-Eibesfeldt 1989), the “eyebrow flash.”

There is remarkable convergence as regards the facial actions and reactions toward some of the universal facial expressions. Particularly those of anger and happiness, there is earlier occurrence (e.g., Charlesworth and Kreutzer 1973; Messinger et al. 1997; Schneider 1997), accuracy of identification (Philippot and Feldman 1990), more stereotypy (e.g., Blurton Jones 1967, 1972), more efficient interpretation (Gaspar 1992, 1994), greater attention (Öhman et al. 2001), and greater cross-species similarity in form and apparent function (Chevalier-Skolnikoff 1982; Gaspar 2001c). Altogether, these approaches strengthen the hypothesis of a biological program for these “basic” expressions. The homology and the identity of function hypothesis require validation.

Research on emotional reactions to the universal emotional facial expressions was reinforcing the biological meaning of those gestalten and the probable coevolution of signal and response (e.g., Dimberg and Thunberg 1998), for instance, reporting stress-related reactions to angry faces even when the stimuli are presented at a subliminal level (e.g., Esteves et al. 1994).

These automated reactions to the universal expressions are transspecific, making a case for their ancient evolutionary roots in the primate order (and possibly further back in some cases): For example, in an experiment where human subjects who had never seen rhesus macaques before had to pick the corresponding contexts for pictures of the macaques facial expressions, humans scored high on correct contexts (Aaron 1984 in Suomi 1996). The rhesus macaques themselves, in a different experiment, displayed adequate behavioral responses to human facial expressions like aversive reactions to an angry human face (for review, see Hauser 1996; Keating 1985).

As of now, specific proposals for homology in primate behavior are the following:
  • Human smile seems to be homologous with the SBT or silent bared-teeth display (van Hoof 1967, 1972; Blurton Jones 1972; Lockard et al. 1977; Preuschoft 1992; Waller and Dunbar 2005). SBT is a grin that mostly (but not always) occurs in silence in macaques and Pan (see Figs. 3, 4 and 5) comprising several variants: with and without pulling up of the lip corners (AU 12), showing or hiding gums, and various transition gradients to open grins.

  • Human laughter seems to be homologous with the nonhuman primate and other mammal’s play face (see Fig. 6) (van Hoof 1967, 1972; Goodall and van Lawick 1968; Chevalier-Skolnikoff 1973, 1982; Panksepp and Burgdorf 1999; Gaspar 2001c; Simonet et al. 2001). It is sometimes referred to as relaxed open-mouthed display (ROM) but a substantial difference was described in several studies (Berdecio and Nash 1981; Goodall 1986; Gaspar 2001c) because ROM is truly a relaxed face with a dropped jaw (and often, the lower lip is also dropped) but without the play face characteristic wherein the lip is withdrawn and the corner of the lip is pulled upward, consequently causing the middle face to wrinkle. However, the two expressions became separated during Hominoid evolution because laughter and play face can occur together but also separately in humans, chimpanzees, and bonobos. The play face occurs most often in human children (Blurton Jones 1972; Lockard et al. 1977) and much more rarely in adults, whereas laughter occurs equally at all ages.

  • Another candidate to homology is the pout face first described by Goodall and van Lawick (1968) and is used by infant chimpanzees mostly to solicit their mother’s attention. It comprises several subtle formal variations accompanying variation in context, however, always has a common denominator, which is the bonding–promoting interaction (Gaspar 2001c). Captive adult bonobos generalized it a bit further and use it in a significant association with sex invitations (Gaspar 2001c). A similar pout (a kiss-looking face) is used by human infants (Blurton Jones 1971) in equally bonding situations.

  • The lip-smacking display seems to be a homology in Anthropoid primates. It consists of a repetitive succession of very fast opening and closing mouth movements and is used in many species in both affiliative and sexual contexts (Napier and Napier 1985). van Hoof (1962) proposed that this display resulted from the ritualization of an intentional movement preceding the lip-smacking that accompanies grooming. Redican (1975) suggested it could be a ritualization of the suckling intention movement shown by all baby primates. Because some species display a teeth-chattering display that resembles both lip-smacking and the SBT display (described above), Preuschoft and van Hooff (1997) suggested it could be a formally and contextually intermediary form between the two. Bonding still collapses all these proposals as a probable primal function after suckling solicitation.
Fig. 3

Transitions in Amber’s “sad” face (silent incomplete chimpanzee cry face) formerly in the Arnhem Colony
Fig. 4

Transitions in Amber’s “sad” face (silent incomplete chimpanzee cry face) formerly in the Arnhem Colony
Fig. 5

Infant chimpanzee Saphira from the Arnhem zoo colony displaying an SBT
Fig. 6

Adolescent chimpanzee Teshua from the Arnhem zoo colony displaying a play face

In spite of all this mostly accepted homologues, some psychologists find the current evidence on the universality of human facial expressions unsatisfactory (e.g., Fridlund 1991, 1994; Russel and Fernández-Dolls 1997). The authors claim that to be accepted as universal, the gestalten must not suffer any cultural variation (both in action and decoding). This demand is in theory ideal but fairly impossible to meet not only in humans but in all the Hominoidea as well. Absolute stereotypy in facial behavior does not happening in humans, chimpanzees, or bonobos where extensive between and within community variation occurs in the frequency, intensity, and sometimes formal variants of facial gestalten (Gaspar 2001a,c). In humans, antistereotypy-known “forces” include the cultural display rules mentioned before and also the rules of decoding (e.g., Biehl et al. 1997; Ekman and Keltner 1997; Matsumoto and Kudoh 1993). Data on cultural variation in gorilla, orangutan, or gibbon facial behavior are still to be collected but there is, in principle, no reason why such variation should not occur.

Idiosyncratic facial behavior and honest cueing: the cases of personality and social attraction

Why go beyond “typical” species or group facial behavior form, intensity, or frequency for a given situation? What sorts of idiosyncrasy in facial behavior have actually been observed and what are the social advantages they may hold? Are there any chances that these idiosyncrasies correspond to strategies that have an impact in individual fitness?

Individualized behavior conveys information about the actor. This information can be honest or dishonest. Honest information provides clues about individual quality that benefit both signaler and receiver. People tend to make inferences about the actor’s personality when viewing pictures of facial behavior (e.g., Ekman 1978; Gaspar 1994; Uleman et al. 1996; Zebrowitz 1997; Krull and Dill 1998;) even when instructed to provide only emotional terms (Gaspar 1994). The role of the portrayed expression, even though intertwined with that of features, seems to be determinant in personality attribution (e.g., Laser 1982; Mueller and Mazur 1997).

A prediction that follows from Dunbar’s (1988, 1993) studies on differently sized primate social societies is that reliable personality cueing can be of advantage in social groups where cheaters can be detected (moderately sized societies) because an individual is seen as accountable. The ability to predict an individual’s personality from small cues is of great advantage to the receiver, on the one hand, because by anticipating the interactor’s behavior he can more appropriately regulate his own behavior and on the other hand, because he can detect cheaters. Yamagishi et al. (2003) provided data supporting that at zero acquaintance, humans detect cheaters better than cooperators based only on photographic face information (taken after cheating or cooperating in a given task), either because they pick up dynamic (action) cues of the personality of the actor or of their previous behavior in the task.

Facial behavior and personality

In humans, the relation between facial behavior (use and decoding of skills as well) and personality traits has generated a great deal of attention, but only a few expressions or aspects of facial behavior and limited contexts were approached. Expressivity could be overall advantageous or just advantageous in certain contexts.

In several studies involving ad lib content attribution to pictures of facial behavior, judges tended to spontaneously infer emotion, situation, and personality of the target individual (e.g., Gaspar 1992, 1994; Krull and Dill 1998; Uleman et al. 1996). But not all expressions are the same in this respect: A happy face elicits more spontaneous attributions of personality than a sad face (Krull and Dill 1998), which is used more to assess the context and emotion. Liu et al. (1992) suggested that this “discrimination” is due to experience and that people would be more likely to attribute personality to more familiar expressions and these are presumably those that are more frequent in the real world. Krull and Dill (1998) also propose that people tend to make attributions to complex states with multiple causes and in which personality is, at the least, as important as the situation.

Borkenau and Liebler (1995) found that facial expression was a major source of personality attribution in an experiment where judges were to make attributions based on the behavior of target individuals during an assorted laboratory task. However, with the exception of “smile,” no specific expressions were mentioned.

It is becoming ever more evident that accurate personality inferences can be made from observations of an unknown individual’s nonverbal behavior (for review, see Zebrowitz 1997). The Big Five personality factors and other personality components can be accurately predicted after very brief viewing of body movements (Grammer 2004). Karl Grammer interprets these findings as evidence of the presence of neural correlates of personality that influence the quality of body movement, thus providing honest signals. This is in accordance with Buss (1988) who predicted that certain psychological dimensions like lethargy/activity would inexorably show in nonverbal behavior in constants such as rhythm, frequency and constancy of head movements, and of change in both body and facial actions.

At the light of the Game Theory, the evolution of individual markers in nonverbal behavior can be seen as an adaptation that confers advantage to those who display (costly) honest signals of their good quality as social partners and of their trustworthiness, enhancing their odds of being chosen as preferred partners in cooperative tasks (Boone and Buck 2003). Others foster associations with these individuals because they can better predict their behavior and will not be exploited by dishonest signals of quality. There is some empirical evidence that there are indeed subtle clues in facial behavior indicating to observers whether their interactor is a cheater or a cooperator (e.g., Ekman and Friesen 1982; Yamagishi et al. 2003).

The case of facial behavior and social attraction

Boone and Buck (2003), while reviewing studies on emotional expression and on cooperation, suggest that emotional expressivity (transparency) can act as a marker for cooperative behavior or trustworthiness and that it is thus adaptive to be socially attracted to someone who is expressive.

Due to the crucial biological meaning of the above basic expressions, socially fit individuals should be able to adequately decode them. Popular children seem to be good decoders (Boyatzis and Satyaprazad 1994; Underwood 1997) and it was recently found that socially anxious adults and other individuals are equally efficient at locating angry faces in the crowd (Öhman et al. 2001).

Expressive3 children were reported to be more popular in kindergarten (Buck 1975; Boyatzis and Satyaprazad 1994; Field and Walden 1982). Popularity in these studies was understood to be synonymous with social attraction and was measured with sociometric techniques, thus, popularity might capture more status-related aspects of social attraction than innate personality correlates that make an individual socially attractive. Preliminary ethological measurements of social attraction, however, point in the same direction showing that popular kindergarten children display certain facial actions (e.g., AU 12) significantly more often than other less popular children (Gaspar 2004, in revision). Preliminary evidence on bonobos and chimpanzees seems to hold the pattern that socially attractive individuals perform a somewhat different facial behavior with expressiveness4 being higher in the most popular individuals (Gaspar 2001c).

It seems that understanding why expressiveness should be advantageous is linked to understanding the advantages of popularity. But, not exclusively. Human females from early childhood to adult age are more expressive than males (e.g., Fujita et al. 1980; Tucker and Friedman 1993) and expressive individuals seem to influence the mood of other individuals and not otherwise (Friedman and Riggio 1981). Unexpressive individuals are often misinterpreted and perceived as boring and uninteresting (Cole 1997).

Dominance and social attraction largely overlap in popularity studies with human children (Babad 2001). A corresponding analysis for nonhuman primates is lacking because social attraction per se was a much neglected subject in primatology, contrary to dominance, which was widely studied, especially as regards formal dominance (after de Waal 1982). In fact, in several primate species, a number of associations were found between facial behavior and formal dominance status with certain facial expressions being more frequent in dominant individuals. (e.g., Chevalier-Skolnikoff 1973; van Hoof 1973; Jacobus and Loy 1981; Preuschoft 1992; Reichler et al. 1998) or differing according to context in those individuals (Gaspar 2001c).

Hubard (2001) found that rejected children more often display angry facial expressions than other children do and popular children also less often display the frown (part of the anger, threat display) than their peers (Murphy and Faulkner 2000).

Conclusions and some remarks on the state of the art

After Darwin’s 1872 insight, The Expression of Emotions in Man and Animals, on the evolutionary roots of human facial behavior, a 60-year silence fell on this matter until a reawakening in the 1960s with the study of both human facial expressions (gestalten or patterns) and of unitary facial actions (componential approach). Since then the study of facial behavior, albeit not always with its biological basis in mind, was developing steadily. Progresses were made in recognizing automatic spontaneous facial actions and also automatic reactions to facial behavior, species continuity in patterns and function and its underlying muscular basis, and neurological correlates of facial action and face processing. Gone are the days when the human/nonhuman animal dichotomy jeopardized the study of human “nature” but the study of homology has, by far, been the most neglected approach of facial behavior.

We are now watching a vigorous reawakening of facial behavior research. In one human ethology, front researchers are paving the way for more time-efficient studies by developing automated systems of facial decoding with computer programs being currently developed for this purpose (e.g., Pentland and Sejnowsky 1992; Yacoob and Davis 1997; Bartlett et al. 1999, 2001; Ohta et al. 2000; Tian et al. 2001; Cohn et al. 2002). These are based on Ekman and Friesen’s (1978) FACS, which is still the major tool with which other researchers manually (and painstakingly so!) code and measure facial action.

In another “human” front, researchers at the Ludwig-Boltzmannn Institute in Vienna are creating human avatars with which they are controlling for physiognomy and dynamic parameters of facial action, which will ultimately clarify the function of facial actions eliminating subjectivity caused by facial features (Grammer 2002, 2004).

From the perception and reaction to facial expression perspective, physiological and cognitive research continues to provide us with phylogenetically relevant information on responses to certain facial actions (e.g., Esteves et al. 1994; Öhman et al. 2001).

In the primatology front, analysis of detailed facial action in Pan is advancing toward a better knowledge of the ontogeny of facial behavior (Chevalier-Skolnikoff 1982; Bard et al. 2001; Bard 2002) of the repertoire of facial actions and ways to measure it (Bard 2004; Gaspar 2004) and the function of individual facial actions and gestalten (Preuschoft and van Hoof 1997; Gaspar 2001c, 2004). As mentioned before, the study of individuality in facial expression is incipient and the study of personality–facial behavior relation is now giving its first steps.

With all these new fronts, it does not look like the study of facial behavior and its evolution will ever hibernate again.


Huber was a German anatomist who performed extensive dissections of nonhuman primate and human faces. His findings on nonhuman primate facial muscles are the most quoted but his report and comments on human racial differences in facial musculature and expression are generally omitted.


Imbedded rules on what can and cannot be shown in the face. These rules also regulate context and actor, i.e., who can and in what circumstances, display what in the face.


Being expressive is, in most studies, synonymous with honest signaler; the actor’s emotions and other states are easily perceived and correctly interpreted by others.


Expressiveness here is in the sense of diversity in the use of distinct facial action gestalten.



The author’s research on facial behavior was financially supported by the EU grants Praxis XXI/BD/9406/96 and POCTI/PSI/57547/2002.

Copyright information

© Springer-Verlag and ISPA 2006