Animal Behaviour

A series of field studies in chimpanzees have shown that the examination of communicative behaviour (e.g. alarm calling) could be particularly fruitful to help uncover mind-reading abilities in non-human animals. In this study, we sought to extend communication-based protocols to one species of Cercopithecids. Specifically, we looked at whether social gazing behaviour of LHoests monkeys (Allochrocebus lhoesti) is sensitive to the informational state of conspecifics using an original experimental design, thereby focusing on their ability to represent conspecifics rather than human experimenters mental states. We presented a group of 11 zoo-housed monkeys with a box that contained either an appetitive (mealworms), aversive (raptor stuffed toy) or neutral stimulus (wood bark chips). The discoverer (the first individual that inspected the content of the box) emitted more social gazes in the aversive condition than in the neutral and appetitive conditions. Besides, social gazing was related to the informational state of the conspecifics in the appetitive and neutral conditions, with more social gazes directed towards informed individuals (who had inspected the box) rather than uninformed ones (who had not inspected the box yet). Additional examination of the behaviour of the discoverers revealed that they were likely addressing social gazes to individuals who were in close proximity to them, suggesting that they preferentially gaze at informed conspecifics because the latter remained in proximity. Our study calls for a more widespread assessment of mind-reading capacities in primates, to further reveal the evolutionary history of traits that were thought to be uniquely human until fairly recently.

Social learning can facilitate information spread within groups and is generally assumed to increase learning efficiency in animals. Here, we asked how individual learning is affected by skill level of a demonstrator present during learning. We predicted that both task-naive and task-experienced individuals benefit from a task-experienced, conspecific demonstrator. We used the all-female clonal Amazon molly (Poecilia formosa) for our experiments. We further propose a model framework which lets us directly relate each research question to a single model parameter. First, following a classical conditioning paradigm over five days, half the fish were trained to find food inside a cylinder, while the others were provided with food randomly dispersed in their tank. As a result trained individuals visited the cylinder with a higher likelihood than randomly fed individuals and showed consistent individual differences in their learning performance. In a second step, we allowed these fish to observe a conspecific while we continued (for those trained) or started (for naives) individual training. We found that trained individuals did not benefit from a partner, regardless of the partners proficiency, but showed higher average performances compared to naives. Naive individuals showed a decrease in learning performance when paired with experienced partners but not when paired with other naive ones. Our results suggest that Amazon mollies learn a foraging task individually and differ consistently in their individual learning ability. Further, social learning depends on own, as well as the demonstrators skill level in a way that observing informed conspecifics may hamper own learning. HighlightsO_LINaturally clonal fish learn operant conditioning tasks C_LIO_LIIndividuals show consistent differences in learning abilities C_LIO_LIExperienced social partners are not helpful during learning C_LI

Animals use learning to adaptively adjust their behaviour to conditions taking advantage of previous experiences. While individual learning is advantageous, it includes risks and costs that can be overcome by learning from others. Despite its relevance, the ability to learn from observing others behaviour was tested in only a limited number of birds species, of which only a few feeds mainly on plants. There is still much to be learn what kind of information birds are interested and capable of gathering from conspecifics in the foraging context. We set out to test the social learning skills of a granivorous gregarious species of cardueline finch, the Serin (Serinus serinus), because they likely follow conspecifics cues to forage. We used an observer-demonstrator paradigm where observers were given the opportunity to learn to reach hidden food from observation of demonstrators that were previously trained to perform the task. Almost half of the birds tested were capable of learning (40%) from a conspecific in a colour-food association experiment, and learners were able to remember the association after fifteen days. Also, almost half of the birds tested for this revealed to be capable of reversal learning. The ability to learn was not influenced by sex or age, of both demonstrators and learners, but learners had longer wings than non-learners. We investigated whether personality could explain the differences in learning ability of these birds. We assessed individual boldness and sociability through the novel object test and mirror personality tests respectively. Although we found repeatability those personality traits we found no association with social learning ability.

Innovativeness and social learning are the pillars of cultural evolution. While the role of social tolerance in social learning has long been acknowledged, its impact on innovativeness remains poorly understood. Here, we test six groups of captive white-faced capuchins (Cebus capucinus) and brown capuchins (Sapajus apella) to explore and compare the relationship between social tolerance and problem-solving propensities. White-faced capuchins are renowned for their rich repertoire of social behaviours, whereas brown capuchins are known for their diverse foraging repertoire, resulting in a broad range of culturally transmitted behaviours in both species. We performed a co-feeding experiment to assess species differences in social tolerance and a set of open diffusion novel food puzzles to test innovativeness. We measured associations for each possible dyad within each group during both experiments and compared the resulting social networks between conditions. We also measured neophobia in an independent experiment presenting groups with three novel objects. Overall, during both experiments, brown capuchins were more tolerant, and proportionally more individuals spent their time within the testing areas at any given moment. We also found that networks across experimental contexts were more similar in brown capuchins compared to white-faced capuchins. Moreover, although differences in approaching and exploring food puzzles were marginal, the success rate was higher in brown capuchins than in white-faced capuchins. Finally, neither eigenvector centrality within the networks nor neophobia could explain individual problem-solving success. Age was the strongest predictor for successfully extracting a food item. The results indicate that species-level differences in social tolerance may contribute to divergent innovation patterns, whereas individual-level variation within species may have only a minor influence. We conclude that the higher social tolerance observed in brown capuchins during feeding and foraging contexts can contribute to their broader repertoire of foraging behaviours and foraging traditions. Data availabilityAll data needed to evaluate the conclusions are present at OSF (https://doi.org/10.17605/OSF.IO/BY5MJ). Declaration of InterestThe authors declare no competing interests. Author ContributionsSandro Sehner: Writing - original draft, review & editing, Conceptualization, Methodology, Formal analysis, Data curation, Funding acquisition. Claudia Fichtel: Writing - review & editing. Peter Kappeler: Writing - review & editing. Helene Meunier: Writing - review & editing, Conceptualization, Methodology.

How individual differences translate into group outcomes is a timely and debated issue. Recent studies, especially in social arthropods and fish, focus on diversity of personality traits. These studies suggest that the phenotypic group assortment by personality type of an animal group, including the presence of "keystone" individuals, leads to group-level personality differences and can strongly impact both group and individual outcomes. However, little attention has been given to the variation of a given trait within a group. Theory predicts that phenotypic homogeneity rather than heterogeneity yields the optimal group performance, especially in an anti-predatory context, but the experimental support includes mainly morphological traits, e.g. body size or colouration. Here, we focus on personality and group level differences in a highly gregarious bird, the European starling (Sturnus vulgaris). We investigated how different degrees of within-flock variation in exploration affect flock outcomes in exploratory behaviour and in escape response after a frightening stimulus. First, we established consistent individual differences in exploration. Then, flocks of 4 birds were formed to obtain gradual differences in mean and in variation of exploration scores among flocks. Flocks underwent an exploration test and a perturbation test. More exploratory individuals entered the test arena earlier, tended to start flying more rapidly and also stopped flying sooner upon frightening. Flocks with a more homogeneous distribution with respect to this personality trait were significantly faster to enter the test room, but no effect on the escape response emerged. The mean exploration tendency of the flock or the maximum exploration tendency of purported key individuals within groups did not play a notable role in explaining such differences in group performance. Our results indicate that it is not the individual that predicts/drives the flock outcome, but rather a group feature, namely phenotypic variation within the group.

Spontaneous eye blinking is a ubiquitous behavior in animals including humans necessary for lubricating the ocular surface and preventing dryness. Beyond this functional role, eye blinking also provides a window into an animals cognitive state and attention allocation. Here, we investigated in 13 female Barbary macaques the modulation of spontaneous eye blinking during two naturally occurring activities differing in their attentional demand (resting and allo-grooming) and additionally assess the influence of social relevance of the interaction on attention allocation. Eye blink rates were significantly lower during grooming compared to resting, suggesting increased attention during this cognitively more demanding task. Dominance rank difference and affiliative relationship strength between the groomer and groomee did not additionally influence eye blink rate. Blinking was timed to coincide with ingestion events during grooming, which may serve as explicit attentional breakpoints. By systematically timing blinks with periods of decreased visual demand, macaques effectively minimized information loss during the non-visual phase of the grooming process. Our study provides insights into the regulation of spontaneous eye blinking in nonhuman animals using a non-invasive tool for the study of visual attention and cognitive load.

AO_SCPLOWBSTRACTC_SCPLOWThe formation of culture in animal societies, including humans, relies on the social transmission of information amongst individuals. This spread depends upon the transmission of social information, or social learning, between individuals. However, not all information spreads. To better understand how constraints at the individual-, dyad- and group-level might influence the formation of culture, we experimentally introduced four innovations (novel behaviours) across three troops of wild chacma baboons (Papio ursinus). At the individual-level, different phenotypic traits constrained individuals use of social information about the innovations. At the dyad-level, we found evidence for social reinforcement and directed social learning affecting who learnt and from whom. Group-level characteristics also limited the diffusion of information, which spread more slowly through social networks that showed less mixing across age classes. Nevertheless, despite these multi-level limitations, the four innovations quickly spread through all the social groups in which they were tested, suggesting that the formation of animal cultures can be surprisingly resilient to constraints on information transmission.

Animals can acquire information through individual learning or by copying others. Simulations suggest that social learning is expected to lead to better rewards, but experimental studies confirming this remain scarce. We tested how a well-known form of social learning in ants, tandem running, affects individual foraging success of Temnothroax nylanderi foragers in controlled laboratory experiments. We manipulated the number and the variability of food sources and assessed the foraging choices of ants searching individually (i.e. scouts) or using social learning (i.e. recruits). We found that social learners indeed discovered better food sources than individual learners, but only in rich environments. However, social learners collected less food ([~]60% fewer foraging trips) than scouts during our trials. Interestingly, individual learners improved their success over time by switching food sources more frequently than social learners. These experimental findings highlight that the relative value of social and individual learning in an ant society depend on the foraging environment and show different temporal dynamics. The ability of individual learners to exploit profitable food sources through a strategy of food source switching, while avoiding the opportunity costs of social learning can help explain why many social insects, especially those living in small colonies, do not use communication in foraging.

ObjectivesMost primate species live in groups, and temporal and spatial coordination of activities of individuals is essential for maintaining group cohesion, and there is still considerable debate to which degree social organization, the extent of despotism, and resource distribution shape group coordination processes. As different baboon species exhibit considerable variation in all of these factors, they constitute an excellent test case to resolve this debate.\n\nMaterials and MethodsWe analyzed group departures and progressions of Guinea baboons, Papio papio, in the Niokolo Koba National Park in Senegal. Guinea baboons live in a multi-level society with strong male bonds and a lack of a clear dominance hierarchy between males.\n\nResultsTwo-thirds of departures were initiated by adult males, and one third by adult females. Although initiators were more likely to signal than followers, signaling did not affect the initiation success. During group progression, males that were not affiliated with females were predominantly found in the front, while affiliated males, females and young were observed more frequently closer to the center of the group, and no preferences for rear positions. Overall, affiliated subjects were more likely to depart and travel together.\n\nDiscussionThe group departures in Guinea baboons differed strikingly from the elaborate negotiation behaviors among male hamadryas baboons, Papio hamadryas. We did not observe that specific individuals dominated the group coordination. Neither social organization, variation in despotism, nor resource distribution alone explain variation between species. Future studies should test whether specific combinations of factors promote the occurrence of negotiation processes.\n\nResearch HighlightsO_LIIn wild Guinea baboons, both adult males and females initiated group departures\nC_LIO_LIInitiators signaled during departures, but this did not affect initiation success\nC_LIO_LISolitary males were predominantly found at the front during group progression\nC_LI

Communication networks are widespread across species, permitting information flow and facilitating social connections across space and time. In birds, communication networks are well studied in territorial species with long-range songs connecting individuals across space, where unintended listeners extract information from others signalling interactions. Yet acoustic signals also play important social roles at short range, forming communication networks that connect individuals within larger social units. Wild zebra finches (Taeniopygia castanotis) provide a unique model system to examine such communication networks in a non-territorial species with multiple males singing at shared locations in close temporal proximity. Zebra finches breed in loose colonies in the Australian arid zone, typically stay near their partner, move around in pairs or small groups, and gather at social hotspots, thus forming dynamic, potentially multi-level, societies. Here, we used the individually distinctive male song, recorded over 1,835 song bouts at two breeding sub-colonies and three social hotspots to quantify singing activity. We identified 163 males based on spectrographic similarities, verified by a deep learning model (BirdNET), and constructed communication networks based on temporal singing proximity at shared locations, social hotspots and the breeding colony. We reveal higher singing activity at social hotspots, with no dawn song and singing peaking at different times of day between breeding sites and social hotspots. Communication networks, with distinct males singing in close temporal proximity, were apparent in both contexts, with larger networks at hotspots. These networks were not strongly nested, yet many individuals met repeatedly at either the hotspot or breeding sites and some even maintained associations across breeding colonies and hotspots, providing building blocks for a multi-level society. These networks may facilitate synchronized foraging and breeding as adaptations to a harsh and unpredictable environment. Additionally, our approach offers a novel road map for widening the understanding of communication networks.

Herring gulls are one of the few species that thrive in anthropogenic landscapes and their familiarity with people makes them an excellent target for studies of inter-species social cognition. Urban gulls pay attention to human behaviour in food-related contexts and we set out to investigate whether such cues can be used to redirect a gulls attention to potential food items in their environment. Herring gulls were given free choice of two differently coloured anthropogenic food items in the presence of a demonstrator, who was either sitting still or pretending to eat food from an item that matched one of the secondary food items. We found that a demonstrator mimicking eating significantly increased the likelihood of an approach or peck. Furthermore, 95% of pecks were directed towards the secondary food item which colour-matched the demonstrators food item. The results show situation-dependent attentional modulation in gulls, whereby gulls are able to use human cues for stimulus enhancement and foraging decisions. Given the relatively recent history of urbanisation in herring gulls, this cross-species social information transfer is likely to be a by-product of the cognitive flexibility inherent in kleptoparasitic species.

A substantial gap in dominance hierarchies research is understanding how such relationships change over time. Previous studies have generally used a single interaction matrix, assuming that dominance is temporally consistent. Yet food sources can be scarce and temporary, leading to changes in species frequency and assemblage composition. Hummingbird assemblages fluctuate as these tiny birds track seasonal floral resources, with individual movements affecting species composition. Artificial feeders, which may foster a greater number of aggressive interactions, provide a stable environment for studying dominance, unlike ephemeral and shifting wild resources. To assess the temporal stability of dominance relationships, we documented aggressive interactions in a hummingbird assemblage over three months on the western slope of the eastern Cordillera of Colombian Andes. We tracked local changes in species composition and visit frequency, recording interactions among 11 hummingbird species using high-speed cameras at three time scales: hour, day, and month. Dominance relationships and their certainty were analyzed through Perc analyses and interaction networks. Over three months of recording, and perusing over 27 hours of non continuous highspeed videos, we registered 3,907 aggressive interactions and 12,629 visits. Our results revealed substantial temporal variation in dominance relationships, species composition, and visit frequency, even within the constant food source that feeders provide. Dominance hierarchies were generally intransitive, with large species dominating, but also with smaller species occasionally winning interactions. We found high uncertainty in dominance relationships within shorter intervals with fewer interactions (e.g., hourly); and that rank changes between dominant species across months and days were linked to shifts in feeder visits. We recommend that future studies incorporate temporal variation, certainty analyses, and interaction networks to determine whether dominance relationships are stable enough to form reliable hierarchies and to explore the ecological significance of reversals in dominance and associated determinants. HighlightsO_LIDominance fluctuates across temporal scales, with monthly, daily, and hourly variations. C_LIO_LIHierarchies become less stable at finer temporal scales. C_LIO_LIBody size and feeder visitation frequency shape dominance interactions. C_LIO_LISmaller species initiate more aggressive interactions but lose more frequently. C_LIO_LIRelationships among dominant species are less predictable than among subordinates. C_LI

Inter-individual recognition is crucial for stable social relationships and it is frequently mediated through vocal signatures. In socially complex species, recognition may additionally require additional levels corresponding to other layers of social organisation such as the pair, family, social group or colony. Additional vocal signatures may encode these different levels of social organisations for recognition. We investigated this hypothesis in the calls of the rook (Corvus frugilegus), a highly social corvid. Rooks form large breeding colonies where multiple pairs nest in clusters. We recorded the calls of five colonies located in France and in Great Britain, including both wild and captive colonies. To exclude variations due to different call types, we focused on the loud nest call produced exclusively by nesting females during the breeding season. We compared the acoustic distance of calls from each individual and between individuals at various levels of nest proximity, i.e. from the same nest cluster, from different nest clusters, from colonies within the same country, and from colonies in different countries. The only vocal signatures we found were at the individual level, but not at the nest cluster or colony level. This suggests a lack of vocal convergence in this species, at least for the nest call, which may be important for pair recognition in large colonies. Further studies should now evaluate if types of calls other than the nest call better carry vocal signatures as markers of different layers of sociality in this species, or if vocal divergence is a more general vocal phenomenon. In that case, applying new methods of monitoring vocal signatures in wild individuals should help understand the cognitive, social and environmental mechanisms underlying this vocal singularisation. 1. Significance statementInter-individual recognition is crucial for social relationships in animals, and is often mediated by individual-specific acoustic characteristics in vocalisations, called a vocal signature. High levels of social organisations, such as a social group of familiar conspecifics or a breeding colony, may likewise be signalled by vocal signatures shared by multiple individuals. We used machine-learning techniques to investigate vocal signatures at multiple social levels in the nest call of brooding female rooks, a corvid species that breeds colonially but lives year-round in social groups. We find evidence of a strong individual vocal signature, but no common vocal signature even in females that nest close together, or in the same colony. A strong individual vocal signature may be a potent tool to monitor populations in this species with minimal disturbance and minimal material, especially as corvids are frequently targeted by human-fauna conflicts in continental Europe.

Animals are exposed to different visual stimuli that influence how they perceive and interact with their environment. Visual information such as shape and colour can help the animal detect, discriminate and make appropriate behavioural decisions for mate selection, communication, camouflage, and foraging. In all major vertebrate groups, it has been shown that certain species can discriminate and prefer certain colours and that colours may increase the response to a stimulus. However, because colour is often studied together with other potentially confounding factors, it is still unclear to what extent colour discrimination plays a crucial role in the perception of and attention towards biologically relevant and irrelevant stimuli. To address these questions in reptiles, we assessed the response of three gecko species Correlophus ciliatus, Eublepharis macularius, and Phelsuma laticauda to familiar and novel 2D images in colour or grayscale. We found that while all species responded more often to the novel than to the familiar images, colour information did not influence object discrimination. We also found that the duration of interaction with images was significantly longer for the diurnal species, P. laticauda, than for the two nocturnal species, but this was independent from colouration. Finally, no differences among sexes were observed within or across species. Our results indicate that geckos discriminate between 2D images of different content independent of colouration, suggesting that colouration does not increase detectability or intensity of the response. These results are essential for uncovering which visual stimuli produce a response in animals and furthering our understanding of how animals use colouration and colour vision.

Studying vocal coordination in a cooperative setting provides insights into how animals process social information, coordinate their actions, and make decisions in complex social settings. It can also reveal how animals use vocal signals to navigate complex social interactions that might not be apparent in solitary or non-cooperative contexts. To our knowledge, experimental evidence for intentional vocal communication in controlled experiments has been demonstrated only in dolphins. Here, we investigate whether peach-fronted conures (Eupsittula aurea) can use vocal communication to coordinate behaviour in a cooperative task. We chose parrots due to their complex social systems, their vocal learning and mimicry skills, and their advanced cognitive abilities. We tested four individuals pairwise in all combinations in the loose string paradigm, which requires synchronization of the two individuals to retrieve a food reward by simultaneously pulling a string. To verify that the parrots used vocal communication to solve the task, we tested them first under three control conditions where visual information and simple behaviour sufficed to solve the tasks. We then compared their vocal behaviour in the control conditions with their vocal behaviour in a task without visual information. During performance of the latter task, the birds demonstrated significantly higher call rates and variability and specific associations between call types and trial outcome compared to the control conditions. After the performance of the task, the birds used vocal convergence more often after failed trials, potentially as a reconciliation mechanism. Our results indicate that peach-fronted conures can use vocalizations to coordinate the solving of a cooperative task, and that vocal convergence may serve as a bonding mechanism following unsuccessful cooperative efforts.

Underlying social learning and other important aspects of successful adaptation to social life is social awareness, where individuals are required to pay attention and respond flexibly to others in their environment. We tested the influence of social context (alone, affiliate, non-affiliate, heterospecific) on behavioural interactions (manipulation, caching, head & body out of sight i.e. barrier use) with food and objects during development at fledging (1-2 months), juvenile (3-8 months) and sub-adult (14-18 months old) in 10 carrion crows (Corvus corone; C. cornix) and nine common ravens (C. corax). These species are closely related, generalist corvids and subjects were all hand-reared and housed in highly comparable conditions. Both species will routinely cache, i.e. hide food and other items for later recovery, and engage in cache-pilfering (stealing) strategies. They will interact and play with objects, potentially as part of developing social bonding and/or physical skills. We found that corvid behaviour was influenced by social context, with birds showing higher frequency of head & body out of sight behaviour while others were observing than when alone, suggesting they have an awareness of others presence and respond by using barriers when interacting with items. There were no differences based on observer identity, supporting an interpretation of item interaction and play potentially driving development of physical skills in this setting. There were developmental effects, including increased manipulation and use of barriers as juveniles as well as increased caching with age. Ravens generally cached more than crows. Objects were manipulated more frequently than food, while barriers were used more with food, indicating that food was more likely to be actively hidden while objects may promote open play. We discuss our findings in relation to other social and developmental influences on behaviour and the wider ramifications for identifying the drivers of play in animals.

Knowledge about the environment is fundamentally important to move, find resources and forage efficiently. This information can either be acquired through individual exploration (personal information) or from other group members (social information). We experimentally assessed the use of social information and its influence on foraging efficiency in sheep, Ovis aries. Naive individuals paired with an informed partner that knew the food patch location, found the patch significantly faster compared to naive individuals paired with another naive individual. Similarly, they spent a significantly lower proportion of time exploring areas away from the food patch. We further found that the previous outcome of using social information (success = access to feed vs failure = no access to feed) had no impact and sheep continued to use social information in subsequent foraging trials and foraged similarly efficient. Our results suggest, naive sheep that are unfamiliar with resource locations, forage more efficiently when informed individuals are present compared to when all individuals are naive. If informed individuals play a similar role in larger groups, new management practices could be developed to improve foraging efficiency when sheep are moved to new paddocks or in paddocks with heterogenous and dynamic resource distribution.

When facing diseases, strategies such as social distancing have proven effective in preventing disease spread. Cleaning interactions are social interactions where cleaner organisms remove ectoparasites from clients. These interactions can be potential vectors of infectious diseases in reef fishes. Here, we aimed to determine if cleaner wrasses (Labroides dimidiatus) employ social distancing and other behavioural tactics to prevent disease spread within the context of cleaning interactions. To this end, we used bacterial lipopolysaccharides (LPS) immunostimulation to induce "sickness" behaviour in cleaners. After LPS injection, we performed four behavioural trials: social preference (from cleaner and client perspectives), social interaction, and bystander cooperativeness tests. In the social preference test, LPS-injected cleaners showed lower activity rates but no change in their social preference. However, some clients preferred "sick" cleaners. Contrarily, during the social interaction test, cleaners reduced the number of interactions, despite clients willingness to interact. For the bystander test, LPS-injected cleaners were unable to adjust their behaviour to be more cooperative. Reduced activity levels revealed lethargy, which may lead to passive self-isolation, allowing for some social distancing of healthy individuals. This suggests that immune-stimulated cleaners undergo changes in their behaviour which can be viewed as a passive social distancing preventing disease spread.

AbstractSocial companions can aid animals in discovering novel resources, with socially connected individuals often locating food more quickly. However, increased sociality also heightens the risk of harassment and disease. Animals would thus benefit from flexibly switching between social and solitary states in response to resource fluctuations. Whether wild animals do this strategically and successfully remains unclear. We experimentally studied groups of wild guppies (Poecilia reticulata) in a resource-poor environment to test if they adjust their social state after recent experience with relatively unpredictable and short-lived novel food patches. Using individually marked fish, we found that groups exposed to ephemeral food patches significantly increased their time spent social--doubling it--compared to control conditions. Groups with stronger foraging motivation showed greater increases in sociality, and individuals in such groups were more successful in locating food patches thereby realizing greater food intake. In a second field experiment, we show that wild guppy groups similarly increased sociality in response to food odours, suggesting strategic adjustments of sociality rather than mere satiation-driven changes in activity budgets. Our findings thus provide causal evidence that animals can be active engineers, rather than passive recipients, of social foraging benefits, highlighting the adaptive nature of fission-fusion dynamics.

The social environment provides both opportunities and challenges for foragers. How multiple components of sociality and foraging interact simultaneously is poorly understood, while experimental manipulation of sociality within wild foraging groups is rare. We tested whether production learning, scrounging, and foraging rate are associated with individual social network metrics and manipulated social stability in mixed species flocks of wild birds. First, individuals were randomly allowed access to, and learned to forage primarily from, one of five feeders. Thereafter, we performed two reversals, manipulating social connections by either assigning birds to a new feeder along with all others assigned to their previous feeder (stable treatment) or reassigning each bird individually (unstable treatment). Most of the effects observed were context dependent but in general we found that: i) Learning was slower in the stable treatment, for individuals with higher weighted degree, and in smaller flocks; ii) Scrounging was higher in the stable treatment, increased with weighted degree, but decreased with flock size; and iii) Foraging rate was predicted only by weighted degree. Our findings demonstrate the complex and dynamic nature of the relationship between sociality and foraging and suggest that selection for group living is more nuanced than usually recognized.