Experimental evidence for yawn contagion in orangutans (*Pongo pygmaeus*)

Contagious yawning (CY), the involuntary mirroring of yawns, is a widely observed phenomenon, yet its evolutionary and proximate causes remain debated. While CY's presence in many social species initially suggested a link to sociality, inconsistencies in findings challenge this hypothesis. To investigate this further, the researchers focus on orangutans, the least social great apes, offering a unique perspective on CY's phylogenetic distribution. Understanding whether orangutans exhibit CY and whether this is influenced by factors like familiarity can shed light on the evolutionary trajectory of this behaviour. This study aims to test for the presence of CY in orangutans and investigate the potential influence of familiarity on the likelihood of contagious yawning. This is crucial because orangutans, due to their semi-solitary nature, could clarify the role of sociality in CY's evolution. If CY is found in orangutans, despite their low sociality, it would weaken the sociality hypothesis. Conversely, its absence would support this hypothesis. This investigation contributes to a broader understanding of CY's evolution and the underlying mechanisms driving this complex social phenomenon.

Previous research on CY has yielded mixed results, with some studies demonstrating CY in various social species (e.g., chimpanzees, bonobos, dogs) and others failing to find evidence in other social species (e.g., gorillas, some macaque species). Hypotheses about the proximate mechanisms of CY include the perception-action mechanism (PAM) that suggests CY may be linked to empathy or emotion contagion. This hypothesis posits that CY relies on the same mechanisms as emotion contagion, where individuals automatically synchronize their emotional states. Studies supporting this link have shown increased brain activity in regions associated with social cognition and theory of mind during CY. However, other research has challenged the strong link between CY and empathy, noting inconsistent findings regarding gender biases and responses in individuals with autism spectrum disorder. Another hypothesis suggests social modulation of CY through social closeness, with individuals yawning contagiously more often when interacting with familiar others. However, this hypothesis too lacks universal support, with some studies failing to find a familiarity bias in various social species. The debate on ultimate causes of CY remains similarly inconclusive, with the social coordination hypothesis suggesting CY aids group vigilance and alertness. However, this hypothesis has yet to be rigorously tested, and the thermoregulatory function of yawning is still subject to debate. Phylogenetic comparisons also raise interesting questions, with CY's presence in some but not all great apes complicating the picture of its evolutionary roots.

This study employed an experimental design using eight captive orangutans housed at Apenheul primate park. The orangutans were exposed to three categories of mute videos: familiar conspecifics, unfamiliar conspecifics, and a 3D orangutan avatar. Each video category included yawn and control conditions (neutral facial expressions and body posture). Videos were presented on a 47" TV screen placed in front of the orangutans’ enclosures. Each test session involved two trials, one with yawn videos and one with control videos, counterbalanced across subjects. The order of conditions and video categories was randomized to prevent bias. Each trial included a 30-second primer to attract attention followed by 56 seconds of video stimuli. A 3.5-minute observation period followed each trial. The entire video database was cycled four times (4 blocks). Observers recorded the occurrence of yawns and scratching (as a measure of arousal) during the observation periods. Inter-rater reliability was assessed for 15% of the video recordings, demonstrating good agreement for both yawns and scratching. Hurdle models were used to analyze the data, treating the presence or absence of yawning as a binomial response and the number of yawns (in cases where yawning occurred) as a count response. The models incorporated condition (yawn vs. control) and trigger (familiar, unfamiliar, avatar) as fixed effects, and subject nested in trial as a random effect.

Across 700 trials, the study recorded 83 yawns across 8 orangutans and 289 sessions. A significant effect of condition was observed: yawning was more likely in the yawn condition compared to the control condition (β = 3.45, SE = 1.06, p = 0.001). A significant interaction effect between trigger and condition was found. Orangutans were more likely to yawn in response to yawn videos of familiar (β = 6.62, SE = 1.59, p < 0.001) and unfamiliar (β = 3.45, SE = 1.52, p = 0.023) conspecifics, but not to the avatar. The likelihood of CY was not significantly affected by familiarity once the avatar data was removed from the analysis. No significant difference in scratching behavior was found between the yawn and control conditions, suggesting the stimuli were similarly arousing and ruling out stress as a primary factor.

The findings demonstrate that orangutans exhibit CY in response to videos of conspecifics yawning, regardless of familiarity. This counters the initial assumption that CY is exclusively present in highly social species, as orangutans show limited affiliative tendencies. The absence of a familiarity effect suggests that social modulation of CY might not be a universal feature, challenging hypotheses linking CY to emotional contagion and social closeness through a shared PAM. The lack of response to the avatar suggests that real-life social cues are necessary to trigger CY, potentially due to the 'uncanny valley' effect where almost-realistic representations are perceived as unsettling. Evolutionarily, the presence of CY in orangutans, but not gorillas, raises questions about its phylogenetic origins, suggesting the mechanism underlying CY could have originated in the last common ancestor of great apes or evolved independently in different lineages. The possibility of CY being lost in the gorilla lineage cannot be excluded.

This study provides compelling evidence for CY in orangutans, challenging the existing link between CY and high sociality. Future research should investigate the potential role of sex differences in stimuli and responders, age-related developmental patterns of CY in orangutans, and the influence of social closeness and stimulus realism on CY expression. Further studies with diverse species and improved methodologies will enhance our understanding of the evolution and proximate mechanisms underlying this intriguing behavior.

The study utilized a sample of zoo-housed orangutans, which might influence affiliative behavior and CY expression compared to wild populations. The small sample size and restricted stimulus selection (male orangutans only) limited the ability to analyze potential effects of sex, age, and social closeness in detail. Lack of continuous gaze quantification might have introduced a bias, although efforts were made to maximize attention to the screen. Further research is needed to address these limitations.