Decoding function in primate facial expressions: re-evaluating broader contexts and terminology

In this paper, we reflected on the role of pioneering studies in shaping research on nonhuman primate facial communication, as well as on how their terminology for describing certain expression types may have imposed limitations over time and, in some cases, misdirected research in the field. We then examined the extent to which more recent nonhuman primate studies have applied novel methods to reveal the complexity inherent in their facial morphology, while studies of their functions have not kept pace. Drawing on the current state of research on nonhuman primate facial communication, we concluded that important opportunities exist to investigate the likely complexity of these functions, representing a fundamental step toward reconstructing the evolution of nonverbal communication in humans and their closest relatives.

Pioneering observational studies on nonhuman primates (hereafter ‘primates’) initially identified distinct types of facial expressions, as part of establishing repertoires of overall primate behaviour (e.g., van Hooff, 1973; Schaller, 1963; Andrew, 1963). These types of expressions were classified based on their most characteristic morphological features as well as the broader behavioural contexts in which they occur (e.g., Redican, 1975; van Hooff, 1962; Andrew, 1965). This classification is often still used nowadays in primate research and the approach of linking types of expressions to broader behavioural contexts has provided important insights for better understanding primate facial communication. However, with the increasing use of advanced technologies in research (such as FACS; Facial Action Coding System: Ekman and Friesen, 1978, 2002) adapted for nonhuman species and the expanding access to primates in both captive and wild settings, it is becoming increasingly clear that a deeper understanding of primate facial expressions requires going beyond interpretations of function that are based only on broad behavioural contexts. Here we propose that examining specific facial morphologies within precisely defined interactional situations is likely to yield insights that analyses on broader context alone may overlook.

The primate and human face, with its complex musculature, is likely to play an important role in this nonverbal communication (Waller et al., 2024; Parr, 2011; Schmidt and Cohn, 2001). The adoption of in-depth facial coding, such as via FACS-based methods (Vick et al., 2007; Ekman and Friesen, 1978, 2002), though not exclusively (Davila-Ross and Dezecache, 2021), has contributed to the view that primate facial expressions may be not only the externalisations of emotion, but also communicative signals used in social interactions (Waller et al., 2017; Clark et al., 2020; Davila-Ross and Palagi, 2022; Fridlund and Russell, 2006). Given that most primates are highly social, effective signalling is essential for maintaining bonds, coordinating behaviour, and accessing resources (Silk and Kappeler, 2017; Burrows, 2008). In the present work, we therefore reflected on how pioneering observational studies have helped shape research on primate facial communication and how more recent methods revealed a complexity in primate facial morphology. We also discussed how research on the functional aspects of facial expressions has yet to fully keep pace with the complexity found in form, and the ways in which terminology of facial expressions may impose limitations and, in some cases, misdirected research in the field. Building on the advances in facial communication research, we highlighted current opportunities to explore the likely complex functions of these expressions, a key step in reconstructing the evolution of nonverbal communication in humans and their closest relatives.

Pioneering observations: primate facial expressions in broader contexts

Early observational studies of primates identified distinct facial expressions often as part of efforts to catalogue broader behavioural repertoires (e.g., van Hooff, 1973; Schaller, 1963; Andrew, 1963). These expressions were typically classified by their form and, at times, their inferred function (e.g., Redican, 1975; van Hooff, 1962; Andrew, 1965). Classification based mainly on form focused on the most characteristic morphological features by using either a descriptive approach (e.g., protruding lips face and silent bared-teeth (SBT) display) or by leaning on similarities to human expressions in form (grimace and yawn). Classification that additionally addressed potential functions was based more on the broader behavioural contexts in which they occur (e.g., threat face and play face).

While such classifications were highly important for providing a foundation for understanding primate facial communication, they also gave the impression that primate facial expression types have a relatively straightforward function. A well-documented example is the SBT display, which has historically been strongly (although not exclusively) associated with submissive behaviour (Bolwig, 1964; Andrew, 1963; van Hooff, 1967; Lockard et al., 1977). However, over the years, it was found that this expression may notably differ in its production with dependence on the social environment. While the SBT display may function as a submissive signal from lower-ranking individuals to higher-ranking individuals in despotic primate groups with strict hierarchies and low post-conflict reconciliation, in more tolerant groups it is used by individuals of all ranks across a broader range of contexts beyond submission, including affiliative contexts (see Preuschoft and van Hooff, 1995, 1997; Thierry, 2007).

This pattern was found for macaques of despotic species such as rhesus macaques (Macaca mulatta) and pig-tailed macaques (Macaca nemestrina) (de Waal and Luttrell, 1985; Bobbitt et al., 1964; Demaria and Thierry, 2001), compared to more tolerant species such as crested macaques (Macaca nigra) and moor macaques (Macaca maurus), where SBT displays are produced by individuals of all ranks and occurs across multiple contexts (Duboscq et al., 2013; Petit and Thierry, 1992; Preuschoft and van Hooff, 1995, 1997; Thierry et al., 2000). Similar patterns were also observed in mandrills (Mandrillus sphinx: Bout and Thierry, 2005; Laidre and Yorzinski, 2005; Otovic et al., 2014) and tufted capuchins (Cepus apella: Carosi and Visalberghi, 2002; Visalberghi et al., 2006). Similarly, in chimpanzees (Pan troglodytes) and the more tolerant bonobos (Pan paniscus), the SBT displays were found to occur in multiple contexts, where they cannot be strictly classified as merely submissive signals (van Hooff, 1972; Waller and Dunbar, 2005; Palagi et al., 2020; de Waal, 1988; Vlaeyen et al., 2022).

Facial expression classifications from pioneering observational studies are still widely used in primate research (see Davila-Ross and Dezecache, 2021). This can introduce bias in further studies, particularly when broader behavioural context is used to classify expression types. For instance, the term “play face” ties the expression closely to the context of play, where the function was arguably addressed as implying “this is play” or “this is just play” (e.g., van Hooff, 1972; for mammal communication, see Pellis and Pellis, 1996), but this terminology may constrain research by overlooking the potential of related open-mouth expressions outside play as well as the potential of related closed-mouth expressions within play. We, therefore, propose decoupling expression terminology from their typical contexts in which they occur in order to enable a more objective assessment of function.

For play faces, we, thus, suggest using descriptive terms, such as open-mouth faces and laugh faces (see Davila Ross et al., 2008, 2015; Davila-Ross and Dezecache, 2021). In his seminal work, Jan van Hooff (1972) reported that these expressions in a chimpanzee colony at Burgers’ Zoo lacked upper-teeth exposure, which was the reason why he did not hypothesise them to be precursors of human smiles. In the past twenty years, however, research on various chimpanzee colonies, has documented that these expressions of play reveal the upper teeth (e.g., Davila Ross et al., 2008, 2015; Crepaldi et al., 2024; Austry et al., 2026; Palagi et al., 2019; Cordoni and Palagi, 2011). In humans, teeth exposure in smiles is culturally mediated, with different patterns observed between Western and Eastern societies (Liu et al., 2010). These differences in the chimpanzee findings emphasise the value of replicating studies across diverse captive and wild primate populations in order to identify variation in facial expression types and to treat putatively prototypical forms with caution.

In-depth measuring of primate facial morphology: towards complexity

With the widespread use of digital camcorders and video-coding programmes, especially since the last twenty-five years, primate research on in-depth coding of subtle behavioural changes, such as facial movements, increased substantially and, consequently, has enhanced our understanding of primate facial expressions. A major advance in detailed facial movement coding in primates was the development of ChimpFACS (Chimpanzee Facial Action Coding System: Vick et al., 2007), adapted from the human-focused FACS (Facial Action Coding System: Ekman and Friesen, 1978, 2002; based on Hjortsjo, 1970). ChimpFACS is an anatomically based tool with which one can identify chimpanzee facial muscle movements homologous to those in humans (Vick et al., 2007). This method allows researchers to conduct detailed and systematic analyses of chimpanzee facial expressions in a way that is comparable to human facial expressions (Vick et al., 2007). Since ChimpFACS, the FACS-based approach has been adapted for use in several non-human species (e.g., Caeiro et al., 2013; Waller et al., 2012; Correia-Caeiro et al., 2025; Mahmoud et al., 2025).

An important advantage is that FACS-based methods enable cross-species comparisons. Because FACS-based methods lean on homologous facial movements, such comparisons can support phylogenetic reconstructions using the principle of maximum parsimony (Saitou and Imanishi, 1989), as commonly done by geneticists and evolutionary biologists, but here applied to behavioural data. Using ChimpFACS, it was, for instance, found that laughing chimpanzees produced multiple facial movements, including raised and pulled back lip corners and raised upper lips and cheeks (Davila-Ross et al., 2015), that have been identified for human laugh expressions coded with FACS (Drack et al., 2009; Ekman et al., 1990; Ruch, 1993; Shiota et al., 2003). Consequently, a shared ancestry for hominid laugh faces was inferred (Davila-Ross et al., 2015). Furthermore, cross-species comparisons can be insightful when examining the same facial muscle activations across closely related species, for instance by comparing macaque species (Rincon et al., 2023; Julle-Daniere et al., 2015) when using MaqFACS (FACS-based method for macaques: Parr et al., 2010).

These FACS-based tools have enabled more fine-grained investigations of primate facial morphology, which revealed a wide array of facial variants for expression types, such as for laugh faces (Davila-Ross et al., 2015; Waller et al., 2015), SBT displays (Clark et al., 2020; Pellis et al., 2011; Fantoni et al., 2025) and yawn faces (Loprete et al., 2025). Thus, such complexity has become evident in some of the traditionally defined facial expression types, which were originally identified when this level of variation could not be captured. Accordingly, their graded facial morphs may overlap with previously identified expression types, especially when they already show similarities in their prototypical forms. A classic example of such overlap, in which not only morphological similarity but also potential evolutionary relatedness across the expression types have been proposed (Davila-Ross and Dezecache, 2021), is the laugh face and the SBT display.

Indeed, the laugh faces vary in the degree of mouth opening and extent of upper- and lower-teeth exposure (Austry et al., 2026; Davila-Ross et al., 2015; Crepaldi et al., 2024; Waller and Cherry, 2012; Palagi et al., 2019; Cordoni et al., 2025; Parr et al., 2007); the SBT display in monkeys and apes also may vary in the degree of mouth opening and extent of upper- and lower-teeth exposure, although it has historically been characterised by upper-teeth exposure (e.g., de Waal, 1988; Parr et al., 2007; Carosi and Visalberghi, 2002; Pellis et al., 2011; Fantoni et al., 2025). While their main difference in form arguably lies in that the SBT may also involve a closed mouth, as already noticed early on in primate facial expression research (e.g., van Hooff, 1972), there is clearly overlap between the open-mouth variants of these two types of expressions.

This example illustrates how historical labelling practices may have shaped subsequent research on facial expressions, further underscoring the need for critical re-evaluation. A speculative consideration further highlights this issue. If the chimpanzee colony observed in van Hooff’s (1970) seminal work at Burgers’ Zoo (see Pioneering Observations) had exhibited upper-teeth exposure during laughter and play — as has been documented in most other chimpanzee colonies since then (for an overview, see Davila-Ross and Dezecache, 2021) — it seems plausible that the SBT displays would have been more closely linked to laugh faces than to closed-mouth SBT displays, based on their facial morphology. Consequently, Davila-Ross and Dezecache (2021) have addressed that there is a possibility of a closer evolutionary link between the laugh face and the SBT display. In our view, this is even more likely if these types of expressions exhibit greater functional overlap than has traditionally been assumed (see Preuschoft and van Hooff, 1995).

It is also important to note that FACS and other in-depth coding methods capture facial movements rather than emotions or their functions. Accordingly, interpreting facial behaviour requires attention not only to broader behavioural contexts but also to specific situations, particularly within social environments. However, caution is needed when identifying emotional or communicative value from these behavioural events alone (Fridlund and Russell, 2006; Fridlund, 2014), similarly to how it can be difficult to disentangle internal states from observable behaviour in humans (Namba et al., 2020; Day et al., 2023).

Decoding function in primate facial expressions and future directions

As a consequence, primate facial expressions include more nuanced structural components than previously assumed. Their findings have also raised new questions — such as the functional significance of specific morphological variants — many of which remain largely unresolved.

The refined understanding of primate facial expression morphology and complexity has, however, prompted only in a few studies a re-evaluation of their general functions. For example, a FACS-based study by Clark et al. (2020) found that SBT constellations in crested macaques differed across copulative, playful, and submissive contexts. These macaques also directed stronger facial movements to similarly matched partners and in riskier social interactions (Clark et al., 2020, 2022), likely enhancing signal salience. Additionally, their jaw stretch, which further widens the mouth opening, was produced predominantly in playful contexts (Clark et al., 2020). However, the SBT displays in monkeys remain far from being fully understood. For instance, Fantoni et al. (2025) did not find significant relationships between SBT mouth opening and function in Tonkean macaques (Macaca tonkeana), when examining playful, peaceful, or agonistic contexts. Moreover, in more despotic groups, where clear communication may help avoid injury, SBT displays tend to be more morphologically distinct from other open-mouth expressions, such as laugh faces, whereas in more tolerant groups they more closely resemble them (Rincon et al., 2023; Preuschoft and van Hooff, 1995, 1997; Thierry et al., 1989).

However, less detailed approaches to measuring facial morphology than FACS-based ones can also provide valuable insights, often focusing more directly on visible facial stimuli (Waller and Cherry, 2012; Parr et al., 2005). For example, activation of the upper lip muscles (Action Unit 10 in FACS) can be subtle and does not always need to reveal the upper teeth, a feature known to be linked to roughness in social interactions (Palagi, 2006, 2008; van Lawick-Goodall, 1968). Approaches that code such visual features that could trigger responses in conspecifics, like upper tooth row exposure (Palagi et al., 2007; Waller and Cherry, 2012), can therefore yield important information beyond what FACS alone captures. For example, a recent study on face-to-face communication in social play showed that great apes tend to replicate laugh faces of their playmates when the upper teeth are covered, but not when they are visible (Austry et al., 2026). These findings suggest that the “less risky” variant with covered teeth may help individuals to better tune in with others or predict their behaviour, potentially promoting social cohesion, coordination, and skill development (Austry et al., 2026; Diana and Kret, 2025; Sandars and Clay, 2025).

It is important to note that human studies have often examined facial expression functions while accounting for social partner characteristics such as sex, age, and social status (e.g., LaFrance et al., 2003; Mehu, 2011; Mehu and Dunbar, 2008a, 2008b; Robertson and Russell, 2016; Hecht and LaFrance, 1998; Ketelaar et al., 2012; Hess and Bourgeois, 2010). They showed, for instance, that spontaneous versus deliberate smiles may vary with the age and sex of both senders and receivers, inferring functional distinctions (Mehu and Dunbar, 2008a, 2008b). To our knowledge, there is only one study that has used such a focussed approach on facial expression form in relation to social partner characteristics to determine function in primate research (Crepaldi et al., 2024).

Using a FACS-based method to study laugh faces and accounting for such social partner characteristics, Crepaldi et al. (2024) found that both orangutans and chimpanzees modified their facial morphology depending on their playmate characteristics, particularly age. Specifically, the chimpanzees of this study exposed more teeth when playing with older (and therefore larger and stronger) social partners, whereas the orangutans did so when playing with younger (smaller and weaker) individuals, see Figure 1 (Crepaldi et al., 2024). The findings suggest that teeth exposure during play may function as a submissive signal in the chimpanzees but as an affiliative signal in the orangutans (Crepaldi et al., 2024). We, therefore, propose that analysing specific facial morphologies with social partner characteristics in primate research may reveal insights that are missed when relying on broader context alone.

Based on the aforementioned findings on laugh face as well as SBT display morphology in relation to function (e.g., Austry et al., 2026; Crepaldi et al., 2024; Clark et al., 2020; Fantoni et al., 2025), it appears that the often-discussed relationship between form and function is less rigid than generally assumed. Specifically, it appears there is no one-to-one rule in that a particular morphological feature addresses a specific meaning and that this relationship varies between interacting individuals and across groups. For instance, teeth exposure — found across differently labelled expressions — may serve different functions, even within the same type of facial expression. This pattern may extend to other facial movements. We therefore recommend that primate facial communication research examine function in close relation to in-depth facial morphology, considering not only different expression types, species, and groups, but also social circumstances, including social partner characteristics. This approach most likely represents an important step toward reconstructing the evolution of nonverbal communication in humans and their closest relatives.