Social animals engage in a variety of interactions with their peers that have different functions, including the reinforcement of social bonds through affiliative interactions (Connor et al., 2000) or the learning of important behaviors through play (Kuczaj and Yeater, 2007) or socio-sexual interactions for instance (Herzing, 1996). Social living provides opportunities for prosocial behaviors, where actions from a donor potentially benefit the recipient without preluding benefits to the donor (de Waal, 2008). Prosocial behaviors may manifest in a variety of contexts, such as care giving, affiliation, social teaching, cooperation and sharing (Jean-Loup, 2019), and these are all apparent in cetaceans. The most common form of caregiving is parental care, which in cetaceans is restricted to extensive female-only (i.e. maternal) care in a form or nursing (e.g. Herzing, 1997; Martin and Da Silva, 2018; MacLeod and D’Amico, 2006; New et al., 2013). In cetacean societies, individuals other than the parents may also contribute to the care of the young, in a form of alloparental associations (Weinpress and Herzing, 2015; Simard and Gowans, 2004; Augusto et al., 2017; Whitehead, 1996; Gero et al., 2009). Affiliative associations (i.e., selective social interactions with specific individuals, including spatial proximity and behavioral synchronization) are also characteristic of cetaceans. Synchronous swimming with one or more partners is often described as an affiliative behavior in bottlenose dolphins (Tursiops sp, Connor et al., 2006; Fellner et al., 2013; Clegg et al., 2017). These behaviors may involve tactile interactions (Sakai et al., 2016) and could facilitate social learnings of prey capture (Bender et al., 2009; Guinet and Bouvier, 1995). But probably the most extensively studied prosocial behavior in cetacean is cooperation.

Bottlenose dolphins (Tursiops sp.) in the wild have been recoded to cooperate to capture food or to gain and maintain access to mating partners. A wide range of ingenious cooperative hunting tactics have been observed with bottlenose dolphins in the wild such as mud ring feeding (Ramos et al., 2022), strand-feeding (Jiménez and Alava, 2015), cooperative prey herding (Benoit-Bird and Au, 2009), coordinated leader-wall hunting (Gazda et al., 2005) and even inter-specific cooperation with fishermen (Romeu et al., 2017). The diversity and flexibility of these strategies likely contributed to the successful, wide range distribution of bottlenose dolphins. Cooperation, as part of their mating strategies, has been demonstrated though male-alliance formation. In Sarasota, Florida, unrelated male dolphins form stable alliances; when working together, they can maintain access to a single female dolphin, fend off predators and doing so increase their reproductive fitness compared to males with no partners (Wells, 2014). Male dolphins (Tursiops sp.) in Shark Bay, Australia, engage in multi-level alliances, creating examples of the most complex social network outside humans (Krützen et al., 2003). These alliances can last for decades or even for lifetime (Connor and Krützen, 2015).

Cooperation and coordination have been tested with dolphins under human care, in a synchronous button pressing task (Jaakkola et al., 2018), in a coordinated rope pulling task (Yamamoto et al., 2019), and in the spontaneous cooperative use of a set of cognitive enrichment devices. Cognitive enrichments represent a special category of environmental enrichments. They allow the animals to practice their natural problem-solving skills, control some aspect of their environment and the use of these enrichments correlates with one or more validated measures of their wellbeing (Clark, 2013, 2017, Clark et al., 2013). A set of cognitive enrichment devices were designed to test cooperative problem-solving in dolphins. They were made of PVC pipes and caps with rope handles attached. They contained ice cubes and fish (Mallotus villosus) which could be obtained by the simultaneous pull of the rope handles. Dolphin pairs, trios and quartets successfully manipulated the cognitive enrichment (Kuczaj et al., 2015; Matrai et al., 2021a, 2021b, 2022a). During the use of these devices, aggression decreased, while play and affiliative, social interactions increased significantly (Matrai et al., 2022b). Once the devices were opened, the dolphins could have access and even share their content. In wild cetaceans, food-sharing has only been reported in a handful of studies. Kin-directed food-sharing was recorded in a population of fish-eating killer whales, mostly expressed by adult females towards maternal kin (Orcinus orca, Wright et al., 2016). Prey-transfer was recorded between adult and calf marine tucuxi dolphin (Sotalia fluviatilis), whereby the mother captured a fish, swam side-by-side with the calf and transferred the fish to the calf’s mouth (Spinelli et al., 2008). In a case study, two adult common bottlenose dolphins (Tursiops truncatus) in Costa Rica were documented sharing a fish; the adults were a female dolphin (who was accompanied by a calf) and a male dolphin (Fedorowicz et al., 2003). A group of rough-toothed dolphins (Steno bredanensis) were observed to carry a large mahi-mahi (Coryphaena hippurus) and pass it between each other (Brower and Curtsinger, 1979). In a more recent study, five rough-toothed dolphins repeatedly passed an unidentified species of jack (Caranx spp.). A dolphin took pieces of flesh off the fish and dropped it, and then another dolphin picked it up and performed a similar action (Ramos et al., 2020).

The proximate mechanisms responsible for these prosocial behaviors may differ based on the donor’s motivation. It can be driven by higher-level social skills such as empathy and altruism, but just as well by lower-level mechanisms, including self-rewarding for the donor, and simpler neurobiological mechanisms. Finally, neuropeptides may be sufficient to elicit these behaviors (Jean-Loup, 2019). While empathy and the neurobiological mechanisms are understudied in non-humans, neuropeptides have been reported to promote prosocial behaviors. Oxytocin was found to facilitate prosocial behavior and prevent social voidance in rats and mice (Lukas et al., 2011). Oxytocin also enhanced grooming and food-sharing in bats (Desmodus rotundus, Carter and Wilkinson, 2015). During cognitive bias testing with dolphins under human care, affiliative behavior correlated with more positive emotional states (Clegg et al., 2017). Altruism in cetaceans has long been suggested as a driver, however, controlled tests are lacking. The most intriguing potential altruistic events were recorded between different species in the wild. A humpback whale (Megaptera novaeangliae) was recorded to interfere with mammal-eating killer whales (Orcinus orca) to help pinnipeds and fish species (Pitman et al., 2017). In Hawaii, humpback whales (Megaptera novaeangliae) were observed engaging in unusual, possibly cooperative play behavior with bottlenose dolphins (Tursiops truncatus, Deakos et al., 2010). In addition, Sakai et al. (2016) and her colleagues monitored a wild Indo-Pacific bottlenose dolphin (Tursiops aduncus) that adopted and cared for a genetically distant neonate calf. Finally, wild dolphins that participated in a provisioning program offered fish and cephalopods to staff members multiple times. This “gift-giving” behavior occurred as part of a sequence such as playing as well as a discrete behavior (Holmes and Neil, 2012).

While food-sharing has been frequently documented in a range of species, there is only limited documentation in cetaceans, which is surprising considering the social complexity and richness of cooperative interactions recorded in the taxon. Investigating food-sharing under controlled conditions allows us to better understand the socio-cognitive abilities, emergence and maintenance of the social structure and social dynamics of wild dolphins. While the maintaining mechanism of prosocial behaviors is debated, their presence in non-human species have frequently been reported and has intrigued researchers for over half a century. Controlled tests in the aquarium helps us understand the complex social interactions of wild dolphins. Research in aquariums has included a test of spontaneous use of water fountains in a prosocial vs selfish manner and Giving Assistance Task in which a dolphin (Tursiops truncatus) could provide shower to another or do nothing. In both tasks the dolphins more frequently chose the prosocial option (Nakahara et al., 2017). However, when dolphins (Tursiops truncatus) were tested in an instrumental helping task, transferring a token or do nothing, they did not spontaneously help their partners (Pérez-Manrique and Gomila, 2019). In another study dolphins (Tursiops spp.) were tested with dynamic stimuli, suggesting that the dolphins were able to recognize prosocial patterns, and also directed high-arousal behaviors toward these displays (Johnson et al., 2018). Finally, when dolphins (Tursiops truncatus) dyads were involved in a Prosocial Choice Tests where they offered the a prosocial (both subject and recipient rewarded), selfish (only subject rewarded) and null options (none rewarded), Lalot et al. (2021) found prosocial and reciprocal tendencies.

In this study, we tested a specifically designed cognitive enrichment that allowed dolphins to engage in prosocial food-sharing actions. The operation of the device required one dolphin’s action of opening and keeping the device open (actor) until another individual retrieved its content (recipient). We monitored the following responses during the test: trial outcomes (number of fish retrieved), the roles (actor or recipient) assumed by the dolphins and any device-related behaviors of the dolphins. We hypothesized that sociability and social dynamics are reflected in the dolphins’ device-related actions. If the dolphins were to engage in prosocial food-sharing actions rather than each individual randomly interacting with the device, we expected to observe higher frequency or longer opening (tugging) events in the presence of other individual(s) than when acting alone.

Our experiment provided evidence for the first successful use of a cognitive enrichment device designed to allow prosocial food-sharing in dolphins. All five male but none of the female dolphins participated in the food-sharing, whereby one dolphin pulled the device open so that the other could retrieve and consume the fish. Although the male group size was smaller than the female group, male dolphins showed a higher level of participation in this task and demonstrated role separation. Males showed a higher level of tugging in the presence of other dolphin(s). In females, tugging showed no significant difference in the presence or absence of other dolphins. These sex differences were revealed by both frequency and duration data analysis. Finally, the involvement of the different male dolphins in the overlapped tugging events highly correlated with their food-sharing success. This correlation suggests an association between the two and confirms the prosocial use of this device.

Cognitive enrichments have gained popularity over the past decades due to their benefits in terms of welfare and science, as have been repeatedly confirmed (Morimura, 2006; Puppe et al., 2007; Manteuffel et al., 2009; Meyer et al., 2010; Clark, 2013; Clark et al., 2013; Zebunke et al., 2013; Clark, 2017; Matrai et al., 2022b). The design of cognitive enrichments for cetaceans mainly focused on problem-solving (Clark et al., 2013) and cooperation (Kuczaj et al., 2015; Matrai et al., 2020; 2021a; 2021b; 2022a). Dolphins (Tursiops aduncus) successfully learnt to manipulate these novel enrichments without receiving any specific training to operate them and invented a novel cooperative behavior referred to as cooperative play (Matrai et al., 2021a, 2021b, 2022a). However, in our previous studies, the cooperative manipulation of the devices allowed all collaborators to benefit directly from the action. Once the dolphins opened the device, its content became available to all participants. The dolphins shared the content of the devices, rendering the cooperative action mutualistic. Cetaceans in the wild might engage in cooperative actions in which the imminent benefits to the participants may be unequal, such as during group hunting (Gazda et al., 2005) or during alloparental associations (Gero et al., 2009). Thus, in this study, we aimed to test a cognitive enrichment device that was specifically designed for cooperative use where one dolphin, the actor, would temporarily forfeit access to the fish reinforcement and simultaneously provide access for another individual, the recipient, in a prosocial fashion. The nature of the dolphins’ actions was determined by their device-related behavior over the session time. The tugging events in the male group were significantly longer in the presence of other dolphins, suggesting prosocial tendencies.

Our findings, therefore, corroborate the prosocial nature of the food-sharing in males; only males (but not females) increased their tugging (both in frequency and duration) if the other dolphins were in the vicinity, independently from their distance of the device. In contrast, females only tugged somewhat longer (the difference was not statistically significant), while the other dolphin(s) were further away from the device and tended to reduce tugging (not statistically significantly) when the other(s) was(were) as close or closer to the device than themselves suggesting selfish, rather than cooperative, interest. Moreover, most of the tugging in the female group was recorded with the three subadult dolphins (Ally, Ginger and Halo) and most of the observations were also recorded by the three of them watching their mothers or each other. Observational learning plays an important role in dolphins’ lives (Yeater and Kuczaj, 2010; Kuczaj et al., 2012). For example, dolphin calves were documented to learn specific foraging strategies by observing their mothers (Boran and Heimlich, 1999; Sargeant and Mann, 2009). In another study on social learning in dolphins, Kuczaj et al. (2012) found that dolphin calves were selective in choosing who to observe and/or imitate. They were more often recorded to watch the play behavior of other calves than adults (including both related and non-related individuals). Thus, the observation events of the female group in our study were more likely to reflect on the subadults’ predisposition towards watching others as part of learning skills rather than their interest towards prosocial food-sharing. The difference in the potential intentionality of the two groups was further supported by the strong correlation between all overlapped tugging events in males but not in females.

Both fish retrieval and device-related actions showed a marked difference towards the male group. Only males obtained fish and were more likely to interact with the device in the presence of other males. The difference in cooperative actions between the male and the female group is in line with our previous findings; males were also more likely to open the mutual cooperative enrichment devices than females (Matrai et al., 2021a). Moreover, while females were not at all, males were recorded to engage in cooperative play up to 90% of the testing time (Matrai et al., 2021b). When participation in enrichment sessions was investigated between the sexes, Eskelinen et al. (2015) also reported a greater level of participation in males. In addition, when a novel visual enrichment was introduced, males were reported to have a more extended response (Winship and Eskelinen, 2018). In other studies, however, females were reported to exhibit a higher rate of interactions with novel enrichments (Lopes et al., 2016). These findings suggest a difference in inter-sexual predisposition towards enrichments, which may partially explain the difference in the dolphins’ device-related behaviors. However, the sharp contrast in the two groups’ tendency to cooperate (with complete absence of cooperative play in females, Matrai et al., 2021a) indicates that the difference lies not in whether to play or not to play but in how to play. Studies have shown that the presence of a younger individual may influence the mother’s behavior, as the juveniles tend to absorb more attention (Clay and de Waal, 2013), hence decreasing affiliative behaviors among adult females (Fultz et al., 2022) and altering the social structure of the group (Kirk and Wascher, 2018). However, in our study the three subadult dolphins were all older (five- and seven-year-old), independent dolphins. In addition, using of the cooperative enrichments, we documented interactions between all individuals in the group (unpublished data). Thus, we do not believe that the presence of the younger animals reduced the adult females’ device-related behaviors. A possible functional explanation is that these sex differences may reflect on the differences in the natural alliance formation tendencies of the two sexes (Connor et al., 2000; Connor, 2007). Association between females in the wild have been reported with other females with dependent calves and was described as a loosely connected network (Whitehead and Mann, 2000; Frere et al., 2010) rather than long-term alliances that were observed in males (Connor et al., 2000; Connor, 2007). Thus, the consistently recurring sex differences in various experimental test paradigms corroborate our current and previous findings and are in line with observations in the wild.

Role separation could be investigated with the active engagement in prosocial food-sharing of all five male dolphins. Of the five males, only one, Domino, switched roles from actor to recipient, while Angelo remained recipient, Anson and Toto stayed actors. The observed role separation could be due to differences in individual preferences for a given action (pulling the handle or taking the fish) or due to the coordinated division of ‘labor’. While in Step 1, differences in skills could also contribute to role separation (it is possible that not all five males learnt to engage with the device in both roles), Step 2 and 3 proved that experience was not a likely influencing factor. In Step 2, all five males demonstrated their abilities to participate as actors and four of them engaged in both roles. However, in Step 3, with a single device again, the dolphins reverted to nearly exclusively participating in one role only. Moreover, the strong correlations between the fish retrieval and the frequency and duration of overlapped tugging events suggest that role separation was consistent and maintained over time. Role separation in cooperative actions in the wild was documented during group hunting; Gazda et al. (2005) reported division of labor in cooperative hunting by a single dolphin acting as a ‘driver’ herding the fish towards the rest of the group, the ‘non-drivers’. The mud ring feeding strategy provides another example of role specialization in cooperating hunting. The ‘ring-maker’ individual circles around the fish, beating his tail to create the mud ring, while the other group members assist by herding the fish (Engleby and Powell, 2019). Role specialization has also been reported during travelling; Lewis et al. (2011) observed that certain dolphins of the group consistently lead the group during travelling (Lewis et al., 2011). Role separation in cooperative hunting and leadership was observed to be persistent. The drivers of the two groups did not change during the observation period (Gazda et al., 2005). The dolphins (Tursiops aduncus) in the Lower Florida Keys all spend little time leading group movements, but a few individuals assumed consistent leadership (Lewis et al., 2011). While the role of fidelity in cooperative hunting is not yet fully understood, Lewis et al. (2013) found a positive correlation between leader-follower associations and relatedness.

Food-sharing, when associated with real fitness consequences, can be considered altruistic. In our experiment, the actor dolphin did not have access to fish, only the recipient. However, the associated fitness costs/benefits in terms of resources are arguably negligible. Therefore, future studies should investigate this possible explanation by testing food-sharing when its associated costs and benefits are more expressed. The proportion of food-sharing between kin and non-kin was near equal. Although are relatively low sample size in terms of related vs non-related dyads did not allow to test the role of relatedness, the similar number of fish retrieved make relatedness an unlikely explanation behind the observed prosocial behavior. Domino (father) and Angelo (son) were the only pair in which both dolphins took both roles and opened the device for each other. Their actions were mainly separated over the three testing steps and represented a relatively small percentage of the fish they each retrieved. Thus, it is difficult to determine if the dolphins expressed reciprocity. The frequency and duration of overlapped tugging events for each other showed a slight preference over other dolphins; however, there was a similar trend observed with the remaining three dolphins, as they also tugged for Angelo and Domino the most. Therefore, these patterns reflect more likely on the preferred roles of Domino and Angelo.

Age is also unlikely to strongly affect the dolphins’ role preference, as the two recipients, Angelo and Domino, were the youngest and the oldest animals in the group. While personality traits were not investigated in this study, it may provide an explanation for role preference. For instance, as Domino was the father of Angelo, he might have passed on certain personality traits to his son, and this may explain their disposition towards the recipient role. In line with this, dolphins under human care have been reported to have distinct and well-defined personalities (Highfill and Kuczaj, 2007, 2010). Furthermore, differences in personality have been reported to contribute to dyadic bond formation (Moreno, 2017) and influence the dolphins’ actions in a cooperative task (Bagley et al., 2020). Finally, avoidance of punishment and manipulative strategies, especially in social organizations that include dominance hierarchy (Cant and Johnstone, 2006; Clutton-Brock and Parker, 1995), have been observed as other potential drivers of apparently altruistic behaviors. In contrast to this explanation, while we acknowledge that it cannot be ruled out, we observed no aggressive behaviors (chasing, biting, or pushing) during the experiment. In fact, in a long-term (three years) evaluation of the use of the cognitive enrichments, also including the food-sharing device of this study, the dolphins were found to be more affiliative with each other in non-research periods of the days when research sessions were conducted (Matrai et al., 2022b).

We acknowledge the limitations of our study, specifically, the involvement of only two same-sex groups and no mix-sex combination, and the relatively low sample size that does not allow us to exclude all alternative explanations behind the observed patterns of prosocial food-sharing. Given the limited literature on the topic, we argue that our findings provide a basis for future studies, investigating if sex differences recorded in our study would be corroborated in other captive groups and whether males would cooperate with females if they had them as available partners. Prosocial actions have been observed in wild societies, however, the proximate mechanism of these actions is not always clear. While cetaceans possess the cognitive capacity and social structure favorable for prosocial actions, controlled tests are lacking. Cognitive enrichment studies with controlled group settings and repeated trials can offer insights and bring scientists closer to understanding prosocial actions in the wild.

In conclusion, our study has successfully demonstrated the first application of a prosocial cognitive enrichment in dolphins under human care, using a novel food-sharing device. Both males and females engaged in device-related actions, but it was only in males that prosocial food-sharing was notably observed. These preliminary findings, particularly the unique role separation observed, provide the foundation for further investigations into the long-term nature, function, and emergence of socio-cognitive skills and cooperation and may contribute to our understanding of social behaviors and dynamics in cetaceans.