Cleaner fish are sensitive to what their partners can and cannot see

Theory of mind (ToM), the ability to understand others' mental states, has primarily been studied in primates and other large-brained species. However, the hypothesis that ecological pressures for strategic deception might drive ToM abilities in less closely related species has gained traction. This study uses the cleaner wrasse (Labroides dimidiatus) as a model organism to test this hypothesis within the context of their cooperative cleaning mutualism. Cleaner wrasses exhibit strategic deception and punishment, providing a natural setting to investigate whether they understand what their partners can and cannot see. The mutualistic relationship between cleaner fish and client fish creates a competitive environment where strategic behavior, such as cheating and detection of cheating, is crucial for survival. This interdependence may drive the evolution of sophisticated social cognitive strategies, including perspective-taking abilities. Understanding whether ToM-like abilities exist in cleaner fish would challenge the widely held belief that larger brains are a prerequisite for these capacities, and potentially provide insights into the evolutionary pressures driving the development of ToM.

Previous research on ToM has largely focused on primates, demonstrating their ability to represent what others perceive and know. Studies have utilized competitive contexts, such as food acquisition, to test primates' understanding of others' visual access. For instance, rhesus monkeys and chimpanzees have shown the ability to infer what a competitor can or cannot see, and they even strategically hide information from competitors to gain an advantage. However, evidence for such abilities in other, less closely related species has been limited. The existing literature suggests that ToM abilities might be associated with larger brains; however, the current study challenges this view. Previous work on cleaner wrasses has demonstrated their capacity for behavioral flexibility and strategic responses to client behavior, but their ToM capabilities remained largely unexplored. This study aims to fill this gap in the literature.

The study employed two experiments using 12 pairs of cleaner wrasses. In Study 1, a transparent or opaque partition separated male and female cleaners. Females were presented with a plate containing preferred and unpreferred food items. The researchers measured the number of unpreferred items eaten by the female in the presence or absence of the male's visual access, and subsequently measured male punishment behavior. In Study 2, a more complex setup involved opaque and transparent barriers offering the female a choice of where to eat, again with manipulation of the male's visual access. The number of unpreferred items eaten, male punishment, and the female's choice of barrier were recorded. The study used Generalized Linear Mixed Models (GLMMs) to analyze the data, considering factors like condition (male visible/invisible), block (in Study 2), and individual fish as random effects. The primary measures were the number of unpreferred food items consumed (indicating cheating) and the occurrence of male punishment. Training was implemented to habituate the fish to the experimental setup and food items. Multiple observations were made for each pair of fish, and careful measures were taken to ensure reliable data collection through methods such as blind coding and video review. Statistical analyses were conducted using R.

Study 1 revealed that females consumed significantly fewer unpreferred food items (i.e., cheated less) when the male could see them compared to when he could not. This suggests that females are sensitive to the male's visual access and adjust their behavior accordingly. Furthermore, male punishment behavior was negatively correlated with female cooperation (the number of unpreferred items eaten), regardless of whether the male could see the female. Study 2, with its more complex setup, demonstrated an initial pattern consistent with Study 1's findings: females preferred to feed behind the opaque barrier when the male was visible. However, this pattern reversed in later trials. Male punishment in Study 2 was higher when females chose to feed behind the opaque barrier, regardless of the male's visual access. This implies that male punishment may be influenced by factors beyond simple cheating detection, perhaps involving other behavioral cues associated with the female's choices.

The findings strongly support the hypothesis that cleaner wrasses possess ToM-like capacities. Their ability to adjust their cheating behavior based on whether their partners can see them mirrors similar behaviors observed in primates. The observation that male punishment is not solely based on visual access to the female's cheating suggests that more complex social dynamics and cues might influence the male's behavior. The reversal of the pattern in Study 2’s later trials might indicate learning or adaptation by the females, further highlighting their behavioral flexibility. This research challenges the assumption that ToM is limited to species with large brains, suggesting that ecological pressures can drive the evolution of sophisticated social cognition in unexpected taxa. The results contribute to our understanding of the evolutionary pressures and mechanisms underlying ToM.

This study provides compelling evidence that cleaner wrasses demonstrate theory of mind-like abilities, challenging the assumption that such cognitive capacities are exclusive to large-brained species. The findings highlight the importance of ecological pressures in shaping cognitive evolution. Future research should explore the specific cues used by both males and females in their interactions, investigate the role of learning and experience in the development of these abilities, and examine ToM-like behaviors in other species facing similar ecological pressures.

The correlational nature of the findings between female strategic cheating and male punishment does not fully establish causality. It is also important to note that the experimental setup in Study 1 did not allow for complete control over visual access in the female's perspective. The study's use of artificial client fish might not perfectly reflect natural interactions. The sample size, while sufficient, could be increased in future studies to enhance statistical power.