Abstract

Introduction

To initiate a shared activity, animals, including humans, often establish mutual engagement with one another (Gómez 1994; Susswein and Racine 2008). Physical proximity is not sufficient to establish mutual engagement; therefore, animals exchange communicative signals to initiate a shared activity (Genty et al. 2020; Heesen et al. 2021). The opening of an interaction is often accompanied by social signals that signify the characteristics of the emerging interaction. Especially when initiating an amicable interaction, indicating one’s friendly stance toward others in advance may be crucial to avoid being misunderstood as having hostile intentions. Studying how animals initiate social interactions by exchanging signals is important to elucidate how they manage complex social interactions. This study focused on the use and function of facial signals expressed by Japanese macaques, Macaca fuscata, when they initiate playful interactions.

Play fighting is a form of social interaction commonly observed in immature mammals (Burghardt 2005), and it involves motor patterns, such as “biting” and “grabbing,” which are seemingly similar to the motor patterns used in serious aggression (Palagi et al. 2016; Špinka et al. 2016). Although these motor patterns are used gently enough to ensure that playmates are not harmed, play fighting can sometimes escalate into overt conflict (Palagi 2018; Iki and Kutsukake 2022). As amicable play interactions can foster social bonds between players (Shimada and Sueur 2018), it is vital for them to avoid unnecessary conflicts. To convey a friendly stance toward their partner and prevent escalation to aggression, animals use a variety of play signals, including facial expressions specific to play contexts (van Hooff 1972; Pellis and Pellis 1996; Palagi et al. 2016).

Studies on play faces and other types of play signals have mainly focused on signals expressed during play sessions rather than those expressed at the start of play (for a few exceptions focused on mutual gaze and gestures at the opening of the play, see Fröhlich et al. 2016; Genty et al. 2020; Heesen et al. 2021). For example, ultrasonic calls by rats (Rattus norvegicus) during play decrease the likelihood of play escalation (Kisko et al. 2015; Burke et al. 2017). In dogs (Canis lupus familiaris), reciprocal bowing during play serves to sustain play (Palagi et al. 2015). Regarding facial signals, although a study of sun bears (Helarctos malayanus) suggested that the frequency and mimicry of play face are not correlated with the duration of the play session (Taylor et al. 2019), other studies have reported that (mimicry of) play face during play prolongs play sessions (chimpanzees: Davila-Ross et al. 2011; western gorillas, Gorilla gorilla gorilla: Waller and Cherry 2012; geladas, Theropithecus gelada: Mancini et al. 2013; Tonkean macaques, Macaca tonkeana: Scopa and Palagi 2016; meerkats, Suricata suricata: Palagi et al. 2019).

Although some studies have shown that the opening of a play session is not necessarily accompanied by facial signals (chimpanzees: Spijkerman et al. 1996; Tibetan macaques, Macaca thibetana: Wright et al. 2018) and that facial signals may serve to maintain an ongoing play session rather than initiate a new one (geladas: Palagi and Mancini 2011), other studies have suggested that animals initiate play fighting with various kinds of play signals (play bows in dogs: Byosiere et al. 2016; play faces in chimpanzees: Hayaki 1985; gestures in chimpanzees: Fröhlich et al. 2016; also see Heesen et al. 2021). Any investigation of the use and function of facial signals expressed before the first playful physical contact between individuals requires comparing play sessions both preceded and not preceded by facial signals. However, no previous studies have conducted this type of comparison. The current study addressed this issue by focusing on dyadic play fighting in juvenile Japanese macaques.

Japanese macaques display facial expressions (i.e., play face, also called relaxed open-mouth display) that are specific to the context of play by opening the mouth in a relaxed way and drawing the corners of the mouth slightly backward (Preuschoft and van Hooff 1995; Petit et al. 2008; Scopa and Palagi 2016). Owing to their morphological and functional similarities, this expression is considered homologous to human laughter (van Hooff 1967, 1972; de Waal 2003; Davila-Ross et al. 2015). This study compared dyadic play fighting sessions preceded by play face with ones not preceded by play face. Our key goal was to establish the situations in which play face is displayed and how play face at the opening of play fighting affects subsequent interactions. Play faces are sometimes expressed bidirectionally by both individuals before the first playful physical contact (see Supplementary Information 1.1 and 1.2). Whether both or only one individual expresses play face in the opening of a play bout might affect how subsequent interactions proceed. Hence, we also addressed whether a functional difference exists between bidirectional and unidirectional play face before the opening of a play fighting bout. Specifically, we tested the following two hypotheses. Note that these hypotheses are not necessarily mutually exclusive while some predictions derived from them are incompatible (i.e., Prediction 1b vs. Prediction 2b).

Materials and methods

Results

We found four models with ΔAICc<2 for the probability of play face expression (see Supplementary Information 2). The ΔAICc of the null model was 16.77. The conditional R² for the model with the lowest AICc was 0.57. The multi-model inference analysis identified the sex combination between players as one of the variables with the strongest effect. Its variable importance reached the maximum value (1.0), and the 95% CI of its standardized coefficient did not overlap with zero (Table ​(Table2),2), indicating that the expression of play face was significantly likelier in a male before initiating play with another male than in a female before initiating play with a male or another female (Fig. 1a). No significant difference was detected in the probability of play face expression between dyads combining male subject players and female partners and male–male dyads (Fig. 1a). These results partially support Prediction 1a. The variable importance of the subject player’s age reached the maximum value, and the 95% CI of its standard coefficient did not overlap with zero (Table ​(Table2),2), indicating that older players would be likelier to express play face (Fig. 1b). Also, the variable importance of the absolute value of the age difference between players reached the maximum value, and the 95% CI of its standardized coefficient did not overlap with zero (Table ​(Table2).2). This result indicated that the probability of play face expression increased as the absolute value of the age difference decreased, reaching its maximum value when the age difference was zero (Prediction 1b supported; Fig. 1c). The results also showed that the rank and age differences between players had no significant effects on the probability of play face expression (Predictions 2a and 2b not supported; Table ​Table22).

Table 2

Details of the model-averaged coefficients

	Sum of weights	Standardized coefficient	Lower 95% CI	Upper 95% CI	p-value
Probability of play face expression
Sex combination (ref: male–male)	1.00
Female–female		−0.719	−1.298	−0.141	0.015
Female–male		−0.496	−0.930	−0.062	0.025
Male–female		−0.200	−0.607	0.206	0.334
Absolute age difference	1.00	−0.584	−1.137	−0.030	0.039
Subject player’s age	1.00	0.656	0.171	1.142	0.008
Age difference	0.22	−0.169	−0.530	0.192	0.359
Rank difference	0.18	−0.127	−0.495	0.240	0.498
Subject player’s rank	0.17	0.139	−0.303	0.581	0.539
Play bout duration
Direction of play face (ref: none)	1.00
Bidirectional		0.345	0.155	0.536	<0.001
Unidirectional		0.121	−0.068	0.310	0.211
Absolute age difference	1.00	−0.269	−0.480	−0.058	0.013
Sex combination (ref: male–male)	0.71
Female–female		0.205	0.018	0.392	0.032
Female–male		−0.036	−0.237	0.165	0.727
Kinship (ref: kin)	0.68
Non-kin		−0.198	−0.418	0.022	0.078
Proportion of the duration of advantage
Sex combination (ref: male–male)	1.00
Female–female		0.291	−0.097	0.679	0.141
Female–male		0.002	−0.383	0.388	0.990
Male–female		0.532	0.150	0.915	0.006
Rank difference	0.38	−0.261	−0.631	0.109	0.167
Kinship (ref: kin)	0.36
Non-kin		−0.288	−0.791	0.215	0.262
Absolute age difference	0.31	−0.274	−0.717	0.170	0.226
Direction of play face (ref: none)	0.21
Bidirectional		0.272	−0.068	0.612	0.117
Unidirectional (partner)		−0.077	−0.339	0.184	0.563
Unidirectional (subject player)		−0.053	−0.306	0.199	0.678
Age difference	0.08	0.033	−0.323	0.389	0.262

Open in a separate window

Sample size: N=226, 113, and 226 for the probability of play face expression, play bout duration, and proportion of the duration of advantage, respectively

Open in a separate window

Fig. 1

The mean probability of the expression of play face according to the a sex combination between a subject player and a partner, b a subject player’s age, and c absolute age difference between players. Error bars represent standard errors. Sample size: N=226. *p<0.05; ns non-significant

We found three models with ΔAICc<2 for the duration of play bouts (see Supplementary Information 2). The ΔAICc of the null model was 15.88. The conditional R² for the model with the lowest AICc was 0.46. The multi-model inference analysis showed that the direction of play face had the strongest effect, as the 95% CI of its standardized coefficient did not overlap with zero, and its variable importance reached the maximum value (Table ​(Table2).2). Post hoc tests showed that play bouts initiated with a bidirectional play face lasted significantly longer than ones not initiated with a play face (Tukey’s HSD: p<0.001) and tended to last longer than ones initiated with a unidirectional play face (Tukey’s HSD: p=0.068), whereas the duration of play bouts initiated with a unidirectional play face was not significantly different from the duration of play bouts initiated with no play face (Table ​(Table3;3; Fig. 2a; Prediction 1c partially supported). The analysis also showed that the play bout duration was significantly affected by the absolute value of the age difference between players (Table ​(Table2),2), indicating a shorter duration of a play bout between pairs with a larger age difference (Fig. 2b).

Table 3

Results of Tukey’s post hoc multiple comparison tests of the effect of direction of play face on play bout duration

	Estimates	SE	Z	p-value
None<bidirectional	−0.876	0.235	−3.720	<0.001
None<unidirectional	−0.347	0.259	−1.339	0.373
Bidirectional>unidirectional	0.529	0.238	2.222	0.068

Open in a separate window

For the Tukey tests, we used the function “emmeans” in the R package emmeans

Open in a separate window

Fig. 2

The mean duration of a play bout according to the a direction of play face and b absolute age difference between players. Error bars represent standard errors. Sample size: N=113. ***p<0.01; †p<0.1; ns non-significant (Tukey’s post hoc test)

We found 12 models with ΔAICc<2 for the proportion of time during which a subject player maintained an advantage (see Supplementary Information 2). The ΔAICc of the null model was 2.62. The conditional R² for the model with the lowest AICc was 0.88. The multi-model inference analysis showed that the sex combination between players was the variable with the strongest effect. Its variable importance reached the maximum value, and the 95% CI of its standardized coefficient did not overlap with zero (Table ​(Table2).2). This result indicates that a male player gained a greater proportion of the advantage over a female partner than over a male partner (Fig. 3). The results also showed that the play face direction had no significant effects (Table ​(Table2;2; Prediction 2c not supported).

Open in a separate window

Fig. 3

The mean proportion of time during which a subject player maintained an advantage according to the sex combination between a subject player and a partner. Error bars represent standard errors. Sample size: N=226. **p<0.01; ns non-significant

Discussion

Overall, our results provided substantial support for Hypothesis 1 (i.e., that play face expression reflects an individual’s motivation for the subsequent interaction), but not for Hypothesis 2 (i.e., that play face is expressed before engaging in a risky situation). We showed that play face expression was likelier in a male juvenile before initiating a bout with another male than in a female juvenile before initiating a bout with another female or a male (Fig. 1a). Although no significant difference was detected in the likelihood of play face expression between a male initiating play with another male and a male initiating play with a female, these results imply that play face expression was likelier in males than in females (Prediction 1a partially supported). The analysis also indicated that Japanese macaque juveniles would be likelier to express play face before initiating play with others closer in age than with others more distant in age (Prediction 1b supported; Fig. 1c). Considering that play frequency is higher in males than in females and that juvenile Japanese macaques prefer to play with individuals of the same age (Mori 1974; Eaton et al. 1986; Glick et al. 1986; Nakamichi 1989), our results suggest that play face before the first playful physical contact is likelier to be expressed by an individual with a higher play frequency (i.e., a male) who is initiating play with a preferable partner (i.e., individuals closer in age). The finding that a play bout initiated with a bidirectional play face lasted longer than a bout not initiated with play face and tended to last longer than a bout initiated with a unidirectional play face (Fig. 2a; Prediction 1c partially supported) also supports this interpretation.

Relevant to Hypothesis 1, it is interesting whether play face functions to invite and/or reengage reluctant partners to play. Our finding that play face expression is likelier in individuals with higher play frequency (i.e., males) and with frequent partners (i.e., same-age peers) suggests that play face is not tactically deployed to recruit reluctant and/or infrequent partners (e.g., females and individuals distant in age) but is automatically expressed by an individual who is engaged in a bout that is about to take place. Also, if play face motivates a reluctant partner to play, it would be expected that bouts initiated with unidirectional play face would last longer than bouts initiated without play face, but we found no such result. To elucidate whether play face in Japanese macaques is an unintentional emotional expression or is deployed tactically, future studies need to examine, for example, whether individuals can adjust their play face expressions in response to the attentional state of the partner or whether individuals tend to use facial expressions repeatedly when they do not get the desired response from the partner (Demuru et al. 2015).

Researchers have hypothesized that play face expression is likely before entering a risky situation to avoid escalation into overt conflict (e.g., Bekoff 1972), but our results did not support this hypothesis (i.e., Hypothesis 2). We predicted that play face expression before the first physical playful contact would be likelier in relatively higher-ranked or older players whose playful attacks can sometimes trigger a negative response (e.g., screaming) in the partner (Iki and Kutsukake 2022) than in lower-ranked or younger partners. However, we found no effect of rank or age difference on the probability of play face expression (Table ​(Table2;2; Predictions 2a and 2b not supported). Also, we found no evidence that the presence of play face before the start of a play bout is related to the proportion of the duration of an advantage during the play bout (Table ​(Table2;2; Prediction 2c not supported). Several studies suggest that play signals have the function of communicating that the aggressive action the sender of the play signal is about to deliver has a benign, not hostile, intent (play face in Hanuman langurs, Semnopithecus entellus: Špinka et al. 2016; play bows in dogs: Bekoff 1995) and function to prevent play escalation (ultrasonic calls in rats: Burke et al. 2017). Although our results are seemingly inconsistent with these previous studies, more detailed research is needed to determine whether play faces in Japanese macaques have these previously proposed functions. One example of how to examine this is to compare the use of play face between play fighting and less competitive social play not involving biting (Demuru et al. 2015), although we focused only on play fighting involving at least one biting. Such a comparison would allow us to test whether play face expression is likelier before starting play fighting, which incurs a higher risk of escalation, than before other kinds of less competitive social play. Furthermore, although we analyzed only play bouts that did not escalate into serious conflict, it is also crucial to directly test whether play face in Japanese macaques functions to prevent play escalation.

Although we did not make any specific prediction relevant to this result, we found that males held a greater proportion of an advantage over female partners in a bout than over male partners (Fig. 3). Some researchers have proposed that juvenile play fighting develops physical strength and skills for real fighting and contributes to dominance acquisition in later life stages (e.g., Hass and Jenni 2010; Pellegrini and Smith 1998; Briffa and Lane 2017; but see Sharpe 2005). Since mature male Japanese macaques leave the natal group and move to other groups (Sprague et al. 1998), their ranking in the natal group as juveniles is inconsistent with their ranking in their new group as adults (Suzuki et al. 1998). Nevertheless, juvenile play in Japanese macaques may be related to their species-specific dominance structure. From this perspective, our finding that juvenile males held a greater advantage over female partners than over male partners may relate to the fact that adult male Japanese macaques typically rank higher than females (Johnson et al. 1982). Also, although Japanese macaque males do not necessarily obtain a higher ranking through aggressive interactions with other males (Suzuki et al. 1998; Takahashi 2002), male–male play fighting may tend to be competitive and, therefore, a male player may have difficulty holding an advantage for long periods of time when playing with another male.

Note that caution must be taken in generalizing the results of this study to other species/groups or other types of play for several reasons. First, the use and function of play signals may differ in species with an egalitarian/tolerant social style that contrasts with the extremely despotic social style of Japanese macaques. Some studies suggest that more egalitarian/tolerant species have more complex communication systems due to the need to manage complicated social relationships (Freeberg et al. 2012; see also Kavanagh et al. 2021). Second, it is also worth noting that our study focused only on dyadic play fighting. Previous studies have suggested that polyadic play fighting sessions are characterized by higher unpredictability (Cordoni et al. 2018), and the probability of play face expression is higher in polyadic sessions than in dyadic sessions (Palagi 2008). While this study did not provide substantial support for the hypothesis that play face expression is likelier before engaging in risky situations, we cannot exclude the possibility that individuals did not necessarily have to use play face because dyadic sessions were sufficiently predictable. Japanese macaque juveniles may need to express play faces when initiating polyadic play fighting characterized by high unpredictability. To examine this possibility, future studies must compare play face expression between dyadic and polyadic sessions. Third, Japanese macaques are known to have intraspecific variations in the level of tolerance (Nakagawa 2010; Kaigaishi et al. 2019). Since our study relies on a limited sample size (N=113 bouts) from a single group, caution should be taken in generalizing our results to other groups of Japanese macaques. Nonetheless, we hope this study will provide a foundation for comparison with other more tolerant groups of Japanese macaques.

In conclusion, whereas previous studies have mainly focused on the play signals used during ongoing play sessions, we have focused on the use and function of facial signals expressed before the start of play fighting. Our results that play face expression at the opening of play in Japanese macaques is likelier in individuals with higher play frequency (i.e., males) and with frequent partners (i.e., same-age peers) imply that this type of facial signal indexes the sender’s motivation for subsequent interactions, but may not function to invite reluctant or infrequent partners.

Supplementary Information

Acknowledgements

Author contributions

The first and second authors conceived and designed the study. The first author gathered the data, performed the statistical analyses, and wrote the article.

Data availability

The data sets analyzed in this study can be accessed at https://bit.ly/3SWklWG

Declarations

Footnotes

References