Cooperation: Behavioral manifestations and neural mechanisms.

Skip to main content An official website of the United States government Here's how you know Here's how you know Official websites use .gov A .gov website belongs to an official government organization in the United States. Secure .gov websites use HTTPS A lock ( Lock Locked padlock icon ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. Search Log in Dashboard Publications Account settings Log out Search… Search NCBI Primary site navigation Search Logged in as: Dashboard Publications Account settings Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Curr Opin Neurobiol . Author manuscript; available in PMC: 2026 Apr 12. Published in final edited form as: Curr Opin Neurobiol. 2026 Feb 19;97:103177. doi: 10.1016/j.conb.2026.103177 Search in PMC Search in PubMed View in NLM Catalog Add to search Cooperation: behavioral manifestations and neural mechanisms Mengping Jiang Mengping Jiang 1 Department of Neurobiology and Department of Biological Chemistry, University of California, Los Angeles, CA 90095, USA Find articles by Mengping Jiang 1 , Steve W C Chang Steve W C Chang 2 Department of Psychology, Department of Neuroscience, Kavli Institute for Neuroscience, and Wu Tsai Institute, Yale University, New Haven, CT 06520, USA Find articles by Steve W C Chang 2 , Weizhe Hong Weizhe Hong 1 Department of Neurobiology and Department of Biological Chemistry, University of California, Los Angeles, CA 90095, USA Find articles by Weizhe Hong 1, * Author information Article notes Copyright and License information 1 Department of Neurobiology and Department of Biological Chemistry, University of California, Los Angeles, CA 90095, USA 2 Department of Psychology, Department of Neuroscience, Kavli Institute for Neuroscience, and Wu Tsai Institute, Yale University, New Haven, CT 06520, USA * Correspondence: [email protected] Issue date 2026 Apr. PMC Copyright notice PMCID: PMC13070002  NIHMSID: NIHMS2156301  PMID: 41719608 The publisher's version of this article is available at Curr Opin Neurobiol Abstract Cooperation is one of the defining features of our species. Our ability to work together in groups was pivotal to our evolutionary history and is crucial for maintaining current societies. Despite its significance, the questions of what drives cooperation, how it emerges, and what determines its success or failure remain largely unresolved. In recent years, a growing body of research across species has begun to address these fundamental questions using diverse behavioral and neural approaches. In this review, we discuss the key features of cooperative behavior and summarize recent advances in understanding its behavioral and neural basis. We highlight the critical roles of partner information and flexible behavioral strategies in facilitating cooperation and provide an integrated overview of neural mechanisms spanning the entire cooperative process, from recognizing and monitoring partners to employing adaptive behavioral strategies, making cooperative decisions, executing joint actions, and evaluating outcomes. The concept of cooperation Cooperation is the process in which two or more individuals work together to achieve shared goals [ 1 , 2 ]. It was pivotal to our evolutionary history and is crucial for maintaining current societies [ 3 , 4 ]. In behavioral ecology, cooperation primarily encompasses mutualism (mutualistic cooperation) and reciprocity (reciprocal cooperation), which are typically distinct from altruism (altruistic behavior) ( Fig. 1 ). These concepts are distinguished by the timing of benefits (immediate or delayed) and whether there is a net cost to the actor. Mutualistic cooperation occurs when a behavior produces immediate positive effects for both actor and recipient, typically requiring real-time coordination and joint decision-making [ 2 , 5 – 7 ]. By contrast, reciprocity involves providing benefits to another individual with the expectation of future return, implicating delayed mutual exchange [ 1 , 8 – 10 ]. Altruism refers to voluntary, often unilateral, actions that benefit others and promote social bonding even in the absence of direct personal gain [ 11 – 15 ] ( Box 1 ). In game theory, cooperation is studied as a strategic interaction in which individuals balance self-interest with collective benefit under conditions of interdependence [ 16 , 17 ] ( Box 1 ). Fig. 1. Spectrum of cooperation. Open in a new tab Mutualism (mutualistic cooperation) involves simultaneous actions by both individuals that produce immediate, shared benefits. Reciprocity (reciprocal cooperation) occurs when one individual initially helps the other, and the partner later returns the favor, yielding delayed mutual benefits. Altruism (altruistic behavior) represents a unidirectional act in which one individual incurs a cost to benefit another without an immediate or guaranteed return. Imitation, mimicry, and copying behavior are alternative forms of behavioral coordination that does not involve cooperation. Box 1. Defining and distinguishing behaviors related to cooperation. Reciprocity is a form of social exchange in which one individual provides a benefit to another with the expectation that the favor will be returned in the future. Unlike mutualistic cooperation, where both partners gain simultaneously, reciprocity involves a temporal delay: the initial actor may incur a cost or gain less benefit at the time of action, while the partner benefits immediately, and the roles reverse later. This mechanism is thought to stabilize cooperation in evolutionary contexts by discouraging free-riding and promoting repeated interactions among partners. In neuroscience and psychology, reciprocity is often studied through iterated game-theory tasks (e.g., the Prisoner’s Dilemma or Trust Game), which reveal how memory, reputation, and social norms shape the expectation of future return. Altruism refers to voluntary actions intended to benefit another individual, such as helping, sharing, comforting, or providing resources. Unlike mutualistic cooperation, which requires both partners to act together to achieve a joint outcome, prosocial behavior can be unilateral: one agent acts in a way that benefits the other, regardless of whether they themselves gain a direct reward. In evolutionary terms, prosociality is thought to promote group cohesion and social bonding, and it can occur even when the actor incurs a personal cost. In experimental animals, prosocial behavior has been studied using paradigms such as food-sharing tasks, door-opening tasks, and consolation behaviors (e.g., helping trapped conspecifics in rats or allogrooming stressed partners). Imitation, mimicry, and copying behavior refer to replicating another individual’s actions, gestures, postures, or behavioral choices. In ethology and neuroscience, these behaviors are distinguished from true cooperation in that they do not necessarily serve a shared goal or require real-time coordination between partners. Copying involves reproducing another’s behavior, such as choosing the same lever, object, or path, often occur mechanistically through reinforcement or attention to salient social cues, without requiring understanding of the partner’s intentions or the causal structure underlying their actions. Imitation represents a more sophisticated form of social learning where the observer intentionally reproduces an action to achieve a similar outcome or learn a specific skill from the demonstrator, implying partial goal understanding. A related phenomenon, mimicry, involves automatic or low-level replication of observable behavior driven by perceptual or motor resonance, such as matching a partner’s movement, or facial expression, typically occurring without conscious intent. Unlike cooperation, which relies on predicting and adjusting to partner’s actions, these observational learning serve primarily as adaptive shortcuts that allow individuals to exploit others’ successful strategies without engaging in trial-and-error learning. Game theory is a mathematical framework for analyzing situations in which the outcome of one individual’s decision depends not only on their own actions but also on the actions of others. It provides tools to study strategic interactions, where agents (or “players”) choose among possible actions (or “strategies”) with associated payoffs that reflect costs and benefits. In social behavior research, game theory has been widely applied to understand cooperation, competition, reciprocity, and negotiation, using paradigms such as the Prisoner’s Dilemma, Dictator Game, Ultimatum Game, and Stag Hunt. These models allow researchers to formalize concepts like trust, fairness, and mutual benefit, and to predict under what conditions cooperation emerges or breaks down. Importantly, game theory bridges biology, psychology, economics, and neuroscience by linking abstract models of decision-making to measurable behavior and neural processes. Mutualistic cooperation—what it is and what it is not Although mutualistic cooperation involves the instantaneous benefits shared by two or more participants, it can sometimes be confounded with coincidental coordination, imitation, mimicry, or copying behavior that incidentally yield mutual benefits ( Fig. 1 , Box 1 ). Even when a behavioral task provides an opportunity for two individuals to obtain rewards through joint action, this does not necessarily mean they are using cooperative rather than alternative strategies for obtaining rewards. Recent studies have provided a clearer operational framework to distinguish genuine cooperation from alternative, non-cooperative behavioral strategies ( Fig. 2 ) [ 5 – 7 ]. Specifically, three key criteria help establish true cooperation: Fig. 2. Summary of behavioral paradigms and neural studies of mutualistic cooperation. Open in a new tab (A) Representative images showing cooperative behavior paradigms in macaques, marmosets, and rodents (mice and rats) with brain regions implicated in cooperative behavior. (B) Schematic illustrating the normal cooperative condition and analytical controls. In the normal cooperative condition, both animals must coordinate their actions to receive a mutual reward. Analytical controls involve comparing observed cooperative decision sequences with shuffled data, where the trial order of one animal’s behavioral sequence is randomly permuted to estimate chance-level action coordination. (C) Experimental controls used to distinguish true cooperation from alternative explanations. The cooperative condition without social communication uses an opaque divider to prevent visual access between animals. The unilaterally cooperative condition allows one animal to receive reward independently while the other requires coordination. The non-cooperative condition provides individual rewards regardless of partner behavior. (D) Summary table of prior studies examining cooperative behavior across species. Panels in B and C are adapted from Jiang et al . 2025 Science. (1). Coordination must exceed coincidental co-occurrence of actions: Two or more individuals may show coordinated behavior without cooperating. For example, two people entering the same elevator each morning display coordinated timing but pursue independent goals (arriving to work on time) based on shared routines rather than partner-specific information. Similarly, individuals may show coincidental coordination when their actions independently converge onto a similar timing after a common environmental cue. To test for genuine cooperation, researchers can use two complementary approaches. First, computationally shuffling behavioral sequences reveals whether temporal co-occurrence exceeds chance levels ( Fig. 2B ); if individuals’ actions are still aligned when their action sequences are temporally shuffled, coordination was likely coincidental [ 5 , 7 ]. Second, experimentally preventing partner information exchange (using barriers or separate chambers) tests whether coordination depends on real-time social interaction (cooperative condition without social communication; Fig. 2C ) [ 5 , 7 ]. If individuals continue to succeed without access to partner cues, they are not cooperating but rather responding independently to shared environmental contingencies. (2). Each and all participants’ behavior must be mutually contingent, not merely sequential. Cooperation requires reciprocal engagement where each and all participants must actively monitor and adapt to each other’s actions in real time. This differs from asymmetric “following” behavior, where one participant initiates an action and the second participant responds based solely on that cue, without the first participant adjusting to the second’s behavior. Such sequential interactions may appear cooperative but lack the bidirectional contingency characteristic of true cooperation. Cooperation also differs from imitation or behavioral copying, where one individual observes and reproduces another’s specific actions without requiring feedback between partners. To test for mutual contingency, one can examine whether role reversal disrupts performance. Leader-follower relationship may emerge during cooperation; however, if the interaction primarily relies on a fixed leader-follower structure, where one individual consistently initiates and the other consistently responds, this suggests sequential following rather than reciprocal coordination. In contrast, genuine cooperation should be robust to role switching, where both individuals actively monitor and adapt to their partner rather than merely following a predetermined or conditioned behavioral sequence [ 5 ]. (3). Decisions must involve active partner coordination, not merely passive responses to social cues. Even if both individuals’ behaviors are mutually contingent, a participant may still make decisions based on an unintended presence or action of the other participant without actively coordinating with them and responding to other’s actions. For instance, if an individual makes a decision based on another participant’s presence at a particular location—but does not require mutual adjustment of actions in response to each other’s actions—this represents social cue-dependent decision-making rather than cooperative decision-making . While both individuals may receive benefits, each participant makes an independent decision triggered by social cues rather than mutually adjusting self-behavior in response to other’s actions. To test whether coordination depends on active partner information versus passive social cues, one can introduce a control that mimic passive social presence. For example, replacing the cooperative partner with a recorded playback, a predictable mechanical stimulus, or a visible but non-cooperative conspecific that provides similar social cues (unilaterally cooperative condition; Fig. 2C ) can reveal whether the individual adjusts behavior based on dynamic partner actions or simply responds to the presence of social stimuli [ 5 – 7 ]. If participants still receive reward when the partner’s behavior becomes non-contingent or is replaced by a static cue, the individual is likely relying on passive social cues rather than engaging in genuine cooperative behavior. Cooperative behavior across species Cooperation in wild animals and humans Cooperation is ubiquitous across biological systems, evident among viruses, bacteria, plants, animals, and humans. In the wild, cooperative interactions may occur both within and between species. Notable examples include cooperative hunting [ 18 , 19 ] and collective defense [ 20 ]. Compared with non-human species, human cooperation is not only more cognitively sophisticated but also occurs on a vastly larger scale [ 3 , 4 ], encompassing cultural, technological, and scientific enterprises such as the construction of museums, the Olympic Games, and the Human Genome Project. Indeed, cooperation forms the cornerstone of modern human society. Cooperation in animal models There has been a long history of studying cooperation using human-designed tasks that allow quantitative analysis of its specific components. A simple yet well-established paradigm is the classic rope-pulling task, in which two individuals must pull a rope simultaneously to obtain a reward. This paradigm has been used in a wide range of species, including chimpanzees [ 21 ], macaques [ 22 ], elephants [ 23 , 24 ], dolphins [ 25 ], and birds [ 26 , 27 ], all of which can complete the task. These studies in social animals provide valuable opportunities to quantify behavioral patterns during cooperation, but they remain limited in their ability to probe the underlying neural mechanisms of mutualistic cooperation. Recent advances using temporal based coordination began to fill this gap [ 5 , 6 , 28 , 29 ], offering tractable systems for mechanistic investigation. This coordination-based task has been examined across multiple species, ranging from non-human primates to rodents, including macaques [ 28 ], marmosets [ 7 , 29 ], tree shrews [ 6 ], rats [ 6 , 10 , 30 – 34 ], and mice [ 5 , 35 – 37 ]. These paradigms typically require two individuals to coordinate their actions within a defined time window, such as simultaneously nose-poking, pressing buttons, pulling levers, performing coordinated movements, or jointly remaining on a platform, to obtain mutual benefits. To facilitate learning, a curriculum training strategy is often implemented, in which task difficulty is gradually increased, for example, by progressively shortening the time window for coordinated actions [ 5 – 7 ]. In a mutualistic cooperation study in mice, two subjects were trained to coordinate their actions to obtain shared rewards by simultaneously nose-poking within a narrow time window [ 5 ]. In this study, multiple control conditions were implemented to demonstrate that mutualistic rewards arise from active coordination rather than coincidental or cue-dependent behaviors [ 5 ]. First, computational shuffling of behavioral sequences revealed that successful coordination exceeded chance levels, indicating animals were not simply adopting timing-based strategies independent of partner information. Second, blocking visual communication with an opaque divider caused cooperative performance to drop substantially, confirming that partner information is essential. Third, conditions where animals received individual rewards regardless of coordination—either both animals independently (non-cooperation) or only one requiring coordination (unilateral cooperation)—failed to elicit cooperative behavior above chance, ruling out behavioral mimicry or following of social cues. A study of mutualistic cooperation in freely moving marmosets requiring synchronous lever pressing has also found that social vision is critical: cooperation was significantly impaired both by an opaque divider between animals and when marmosets coordinated with a fully visible automated lever not controlled by the partner [ 7 ]. Together, these controls establish that genuine mutualistic cooperation requires active coordination involving partner information perception, mutual benefits, and bidirectional behavioral adjustments—distinguishing it from timing-based coincidence, social cue-dependent decision-making, or behavioral copying. Animal models with social deficits, such as those carrying mutations in Shank3 mutations associated with autism spectrum disorder [ 36 ], exhibit disrupted coordination during cooperative tasks, including obtaining fewer rewards, exhibiting reduced cooperative efficiency, and displaying increased latency to initiate trials. These behavioral alterations provide a mechanistic link between impairments in cooperative behavior and underlying neurodevelopmental dysfunction. Behavioral components of cooperation Partner information in cooperation Successful cooperation requires flexible coordination between participants. Participants need to dynamically adjust their behavior according to partner’s status and anticipate forthcoming actions [ 5 ]. In rodents, the importance of partner-related information has been systematically examined by disrupting or removing access to such cues [ 5 , 7 , 31 ]. Physically approaching the other participant allows an individual to monitor partner’s state more closely [ 5 , 6 ]. When visual contact between partners is blocked, cooperative performance decreases sharply or is abolished almost completely. To acquire partner-related information, visually oriented primates such as macaques and marmosets rely heavily on social gaze [ 7 , 28 ]. For example, freely moving marmoset dyads adjust the level of reliance on a social gaze dependent cooperative strategy depending on the partner identity [ 7 ]. Behavioral strategies facilitating cooperation To achieve efficient cooperation, animals develop diverse behavioral strategies that enable them to monitor their partner’s behavior and coordinate their actions. In addition to approach and social attention, waiting represents a more action coordination mechanism, in which animals actively wait for their partner when the partner is absent or distant. For example, chimpanzees, elephants, Barbary macaques and peach-fronted conures wait to pull a rope until their partner is present and stop pulling when the partner becomes inattentive [ 21 – 24 , 26 ]. Kea parrots can wait for a partner when cooperation is required to obtain food [ 38 ]. Rodents such as rats and mice also display waiting behavior [ 5 , 35 ]. Moreover, in the mutual cooperation task, mice engage in direct interaction by mutually poking their noses toward each other to coordinate their actions [ 5 ]. The increase of this mutual nose-to-nose interaction is accompanied with the increase of cooperation performance. Intriguingly, mice refine this mutual interaction as learning progresses, adopting a strategy that allows both animals to maintain virtual and physical contact with their partner while simultaneously positioning themselves efficient transition to cooperative action [ 5 ]. In primates, social gaze serves as a key strategy to gather partner information, analogous to the approach behavior seen in rodent cooperation. Monkeys are more likely to cooperate after viewing social cues, indicating that visual fixation on a partner’s face or actions is a critical driver of cooperative learning [ 28 ]. Marmoset dyads use two distinct types of social monitoring based on social gaze, and the temporal precision between social gaze and partner’s lever pulls is enhanced during successful cooperation, suggesting the use of social gaze to couple partner’s behaviors [ 7 ]. In addition to the gaze-dependent strategy, marmosets employ a rhythmic strategy, coordinating their actions based on internal timing and task structure rather than visual cues alone. Leader-follower structure Within the framework of mutualistic cooperation, leader-follower patterns are likely to emerge as an adaptive strategy to enhance coordination [ 6 , 34 , 39 ]. For instance, a more proficient cooperator who quickly learns the task rules may take the lead to guide a less experienced partner toward successful outcomes. Moreover, because perfect symmetry in coordination is rarely achievable, the establishment of a stable leader-follower structure enables individuals to adopt complementary roles, thereby minimizing conflict and confusion in complex cooperative contexts. However, the formation of a stable leader-follower structure needs to be carefully distinguished from a fixed leader-follower relationship that reflects sequential following behavior as opposed to true cooperation. Social relationships in cooperation Social relationships, such as kinship, familiarity, social status, and sex composition, profoundly influence cooperative interactions. Familiarity and kinship enhance cooperative performance, and strong social bonds facilitate the maintenance of cooperative interactions [ 7 ]. People preferentially cooperate with partners who have demonstrated cooperative tendencies in previous interactions [ 40 ]. Interestingly, mixed-sex marmoset pairs (female-male) often outperform same-sex male pairs, suggesting that sex composition functionally shapes cooperative dynamics [ 7 ]. Social status represents a natural hierarchical structure in many species and is distinct from the concept of leadership. In marmosets, dominant individuals can strategically follow the actions of subordinates, suggesting that dominant animals adopt adaptive roles to facilitate cooperative problem solving [ 7 ]. Moreover, marmosets with different dominance statuses exhibit distinct cooperative strategies: dominant individuals generally engage more actively and display higher overall social attention so that they can work with the subordinates to facilitate successful cooperation [ 7 ]. Neural mechanisms of cooperative behavior Mutualistic cooperation requires neural networks involved in social memory, partner recognition, decision-making, action coordination, and reward evaluation. By combining behavioral paradigms with genetic, circuit, and computational approaches, researchers have begun to identify the neuronal populations that encode and causally regulate these behavioral processes. We highlight recent advances, primarily from non-human primates and rodent studies, that elucidate the neural mechanisms underlying core building blocks of cooperation ( Figs. 2 and 3 ). Fig. 3. A model of cooperative process between two individuals. Open in a new tab (A) During cooperation, individuals continuously monitor their partner’s state, actively adjust their own behavior and decision making based on partner’s behavior or state (e.g., whether to hold or proceed), and execute task-related actions such as poking or pressing. During this process, individuals may use multiple behavioral strategies, including approach, waiting, interaction, and gaze, to facilitate social coordination. Successful coordination leads to shared benefits, which may reinforce the motivational drive to sustain cooperation over time. Cooperation involves recognition of partner’s identities, social memory (past interactions), as well as additional social factors such as kinship, familiarity, social status, and sex. (B) Schematic illustrating in vivo calcium imaging in mice using a miniature microendoscope (Miniscope) in the anterior cingulate cortex (ACC). (C) Correlation between the fraction of neurons encoding partner position and cooperative poke ratio in individual animals. (D) Principal component (PC) projections of ACC population activity associated with approach, waiting, and interaction behaviors, showing distinct neural trajectories for each behavioral strategy. (E) Performance of decoders in classifying proceed versus hold decisions. (F) Correlation between decoder performance for classifying correct versus miss pokes and animals’ cooperative performance. Animals with better neural discrimination of trial outcomes exhibited higher cooperative performance. Panels in B-F are adapted from Jiang et al . 2025 Science . Recognition and perception of partner While identity, familiarity, social status and sex composition all influence cooperative performance, recognizing and differentiating others is a fundamental prerequisite for initiating and maintaining cooperative relationships [ 7 , 41 , 42 ]. In human cooperation, the right dorsolateral prefrontal cortex (dlPFC) shows increased activity when individuals encounter non-cooperative partners [ 40 ]. In non-human primates, partner recognition recruits brain regions specialized for encoding facial and identity information [ 43 – 45 ]. In rodents, neurons in the medial amygdala, the ventromedial hypothalamus, and the medial prefrontal cortex respond differentially to sex- and species-specific social stimuli [ 46 – 49 ], supporting categorical social recognition. The amygdala is also involved in social status perception in both non-human primates [ 50 , 51 ] and rodents [ 52 ]. Monitor and infer partner state The ability to actively monitor a partner’s state and anticipate their intentions or future actions is critical for successful cooperation. A recent study in mice demonstrates that the ACC encodes partner’s position and their relative distance to target locations relevant for cooperative behavior [ 5 ]. Notably, this partner-state encoding becomes stronger during key cooperative moments, and individuals with more robust partner representations exhibit better cooperative performance, suggesting that the ability to track and process partner-related information predicts cooperative capacity ( Fig. 3B – C ). Beyond representations of other’s positions, studies in primates have identified critical roles of the middle superior temporal sulcus (STS), amygdala and medial frontal cortex (MFC) played in inferring others’ internal states and supporting adaptive social coordination. Specifically, neurons encoding predicted actions of a partner are more prevalent in the middle STS than in the anterior cingulate gyrus (ACCg) [ 17 ]. Neurons in the dorsal anterior cingulate cortex (dACC) encode other’s unknown actions and can predict forthcoming choices [ 16 ]. Moreover, neurons in the amygdala accurately predict partner’s decisions, enabling individuals to reconstruct other’s internal states [ 53 ]. Within the MFC, neurons in the convexity preferentially encode other’s actions and errors, whereas sulcus neurons are more closely associated detecting others’ errors and guiding subsequent action correction [ 54 , 55 ]. In addition, observing a partner increases spiking coordination between the midlevel visual cortex (V4) and dlPFC [ 28 ], and evaluating partner’s reward engages a specific communication between ACC and the amygdala [ 56 ]. Together, these findings indicate that partner-dependent cooperative behaviors recruit a distributed network of neural nodes involved in social monitoring and inference. Behavioral strategies To effectively monitor partner information and position themselves for flexible coordination, animals use several preparatory behavioral strategies, including approach, waiting, interaction, and social gaze. The ACC and PFC play central roles in organizing these preparatory behaviors [ 5 , 28 ], with distinct neuronal subpopulations selectively encoding specific behavioral components. In mice, separate ensembles of ACC neurons encode approach, waiting, and interaction behaviors ( Fig. 3D ), and the strength of these neural representations is positively correlated with cooperative performance [ 5 ]. In macaques, both V4 and dlPFC encode social gaze-related activity—whereas most V4 neurons respond exclusively to social gaze, dlPFC neurons encode both social gaze and action-related variables [ 28 ], suggesting integrative processing of visual and motor information during social coordination. Moreover, interactive social gaze variables are widely represented in the prefrontal-amygdala circuits of primates, supporting an integrative role of social gaze in these brain areas central to social cognition [ 57 ]. Social decision-making Decision-making occurs in the brief moment preceding action execution, when participants make an internal choice to proceed with or withhold the cooperative action based on other’s actions. Cortical regions such as the PFC, ACC, and amygdala are implicated in this process [ 5 , 28 , 58 ]. In mice, ACC neurons can differentiate the decisions to hold or to execute the cooperative action ( Fig. 3E ), and the fraction of neurons encoding these decision-making processes is positively correlated with cooperative performance, suggesting that a stronger representation of social decision-making process is associated with cooperative efficiency [ 5 ]. Moreover, in freely moving marmoset pairs, the dmPFC carries out a gaze-dependent social evidence accumulation process to guide cooperative decisions [ 58 ], illustrating one specific manner by which social monitoring directly guides cooperative decision-making. Action execution and outcome Animals can distinguish whether their actions are performed individually or jointly, indicating strong context dependence in motor representations. The prelimbic cortex (PRL) and nucleus accumbens (NAc) exhibited increased delta/theta activity, whereas the BLA showed decreased delta/theta activity when rats stayed together on a platform, compared to when they jumped onto it individually [ 33 , 34 ]. At the single neuron level, neurons in the rat OFC and PRL respond differently to the same actions depending on whether they occur during individual or cooperative performance [ 6 ], suggesting that these regions encode the contextual state of an action rather than the action itself. In mice, the ACC shows a similar encoding pattern, differentiating actions that lead to successful or unsuccessful cooperation and signaling whether partner-related information is available during action execution ( Fig. 3F ) [ 5 ]. Importantly, inhibition of ACC activity in mice disrupts the coordination of actions between partners without affecting the general ability to execute actions, underscoring its specific role in mediating cooperative coordination rather than basic motor output [ 5 ]. Beyond action execution, the OFC, NAc, and caudate nucleus exhibit increased activity following mutual cooperation in human [ 59 , 60 ]. These regions process outcome-related signals in a context-dependent manner. In primates, OFC neurons predominantly encode rewards received by oneself, whereas the ACCg encodes reward outcomes not only for oneself but also for the partner. In contrast, neurons in the anterior cingulate sulcus (ACCs) preferentially respond to rewards delivered to other [ 61 ]. The amygdala also encodes reward information for both self and other [ 53 , 62 ], and specific communication patterns between ACCg and amygdala are associated with prosocial preference [ 63 ] and vicarious reward [ 56 ]. Within the dmPFC, distinct populations of self-type and other-type neurons modulate their activity according to the reward magnitude received by oneself or a social partner, respectively [ 64 ]. These studies underscore the importance of the prefrontal cortex in processing action and reward outcomes for self and other that are important for the reciprocal aspects of cooperation. Future direction: understanding cooperation across scales, contexts, and systems. While rodent and primate studies highlight key regions such as the OFC, PRL, dlPFC, ACC, and amygdala in encoding social and cooperative information, brain-wide dynamics linking these regions remain elusive. Large-scale, high-density recording techniques such as Neuropixels probes will allow simultaneous monitoring of distributed neural activity to reveal how sensory, motor, and social signals are integrated for coordinated action. Beyond dyadic interactions, real-world cooperation often involves groups, yet how neural and computational mechanisms scale from pairs to collectives remains largely unknown. Understanding how multiple individuals cooperate to achieve share goals will illuminate the neural basis of collective behavior and inform strategies for improving group decision-making. Finally, balancing cooperation and competition represents a fundamental social dilemma. Identifying these mechanisms will advance our understanding of social decision-making in complex social settings. Artificial intelligence as a model system for cooperation Recent advances in artificial intelligence (AI), particularly multi-agent reinforcement learning (MARL), have opened new avenues for understanding cooperative behavior among AI and for investigating the computational principles underlying cooperation [ 5 , 65 ]. By training artificial agents with recurrent neural networks to cooperate in simulated environments that mirror biological paradigms used for mice, researchers revealed striking parallels between artificial and biological systems [ 5 ]. Like their biological counterparts in mice, artificial agents develop behavioral strategies such as waiting and active coordination based on partner information, and their neural networks encode both self and partner positions, with stronger partner representations correlating with better cooperative performance [ 5 ]. A defining feature of the multi-agent environment is full access to each agent’s neural network, which allows for precise manipulations. Indeed, selective ablation of task-relevant neurons in artificial networks disrupts specific aspects of cooperation, demonstrating functionally distinct subpopulations that causally regulate cooperative behavior [ 5 ]. This convergence between biological and artificial systems not only reveals key computational principles of cooperation but also positions AI as a tractable platform for testing mechanistic hypotheses that remain technically challenging in animal models, while simultaneously informing the design of more sophisticated collaborative AI architectures for real-world applications. Exploring cooperation between humans or animals and AI agents also represents a promising frontier, where adaptive AI partners can advance human-AI collaboration in domains such as education and companionship. Concluding remarks Over the past few decades, research on cooperation has progressed from qualitative behavioral observations to quantitative, mechanistic investigations that span species, scales, and methodologies. These advances have begun to establish an integrative framework that links behavior, neural circuitry, and computational principles underlying cooperative interactions. Moving forward, the integration of innovative behavioral paradigms with next-generation, brain-wide neural-recordings and computational approaches promises to uncover how cooperation emerges from the coordinated activity of distributed neural networks. Such efforts will ultimately deepen our understanding of the mechanisms that enable cooperative behavior across species, help improve human lives, and develop better social AI. Impaired cooperation is a hallmark of several mental conditions, including autism spectrum disorder and schizophrenia. Understanding the underlying neural mechanisms will elucidate the causes of disrupted social behavior and inform translational approaches for diagnosis and treatment. Acknowledgment This work was supported in part by NIH grants R01 NS113124, R01 MH130941, RF1 NS132912, and R01 MH132736 (to W.H.); a Packard Fellowship in Science and Engineering (to W.H.); a Vallee Scholar Award (to W.H.); a Mallinckrodt Scholar Award (to W.H.). References 1. Brosnan SF, Waal FBM de: A proximate perspective on reciprocal altruism. Hum Nat 2002, 13:129–152. [ DOI ] [ PubMed ] [ Google Scholar ] 2. Gokcekus S, Cole EF, Sheldon BC, Firth JA: Exploring the causes and consequences of cooperative behaviour in wild animal populations using a social network approach. Biol Rev 2021, 96:2355–2372. [ DOI ] [ PubMed ] [ Google Scholar ] 3. Melis AP, Semmann D: How is human cooperation different? Philos Trans R Soc B: Biol Sci 2010, 365:2663–2674. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 4. Pennisi E: How Did Cooperative Behavior Evolve? Science 2005, 309:93–93. [ DOI ] [ PubMed ] [ Google Scholar ] 5. Jiang M, Gu L, Ma M, Li Q, Kao JC, Hong W: Neural basis of cooperative behavior in biological and artificial intelligence systems. Science 2025, doi: 10.1126/science.adw8151. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 6. Jiang M, Wang M, Shi Q, Wei L, Lin Y, Wu D, Liu B, Nie X, Qiao H, Xu L, et al. : Evolution and neural representation of mammalian cooperative behavior. Cell Reports 2021, 37:110029. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Meisner OC, Shi W, Nair A, Nandy G, Jadi MP, Nandy AS, Chang SWC: Diverse and flexible strategies enable successful cooperation in marmoset dyads. Curr Biol 2025, 35:4509–4521.e5. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Wilkinson GS: Reciprocal food sharing in the vampire bat. Nature 1984, 308:181–184. [ Google Scholar ] 9. Wood RI, Kim JY, Li GR: Cooperation in rats playing the iterated Prisoner’s Dilemma game. Anim Behav 2016, 114:27–35. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 10. Wang M, Shi Q, Shao Y, Jiang M, Fu A, Huang Y, Wang Q, Wei L, Zhang Z, Xu J, et al. : Oxytocin-mediated empathy internally facilitates cooperative behaviors in rats. Sci Bull 2025, 70:3397–3408. [ DOI ] [ PubMed ] [ Google Scholar ] 11. Wu YE, Hong W: Neural basis of prosocial behavior. Trends Neurosci 2022, 45:749–762. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 12. Wu YE, Dang J, Kingsbury L, Zhang M, Sun F, Hu RK, Hong W: Neural control of affiliative touch in prosocial interaction. Nature 2021, 599:262–267. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 13. Zhang M, Wu YE, Jiang M, Hong W: Cortical regulation of helping behaviour towards others in pain. Nature 2024, 626:136–144. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 14. Sun F, Wu YE, Hong W: A neural basis for prosocial behavior toward unresponsive individuals. Sci (N York, NY) 2025, 387:eadq2679. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 15. Scheggia D, Greca FL, Maltese F, Chiacchierini G, Italia M, Molent C, Bernardi F, Coccia G, Carrano N, Zianni E, et al. : Reciprocal cortico-amygdala connections regulate prosocial and selfish choices in mice. Nat Neurosci 2022, 25:1505–1518. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 16. Haroush K, Williams ZM: Neuronal Prediction of Opponent’s Behavior during Cooperative Social Interchange in Primates. Cell 2015, 160:1233–1245. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Ong WS, Madlon-Kay S, Platt ML: Neuronal correlates of strategic cooperation in monkeys. Nat Neurosci 2021, 24:116–128. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Muro C, Escobedo R, Spector L, Coppinger RP: Wolf-pack (Canis lupus) hunting strategies emerge from simple rules in computational simulations. Behav Process 2011, 88:192–197. [ DOI ] [ PubMed ] [ Google Scholar ] 19. Boesch C, Boesch H: Hunting behavior of wild chimpanzees in the Taï National Park. Am J Phys Anthr 1989, 78:547–573. [ DOI ] [ PubMed ] [ Google Scholar ] 20. Mausbach J, Goncalves IB, Heistermann M, Ganswindt A, Manser MB: Meerkat close calling patterns are linked to sex, social category, season and wind, but not fecal glucocorticoid metabolite concentrations. PLoS ONE 2017, 12:e0175371. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. Chalmeau R: Do chimpanzees cooperate in a learning task? Primates 1994, 35:385–392. [ Google Scholar ] 22. Molesti S, Majolo B: Cooperation in wild Barbary macaques: factors affecting free partner choice. Anim Cogn 2016, 19:133–146. [ DOI ] [ PubMed ] [ Google Scholar ] 23. Li L-L, Plotnik JM, Xia S-W, Meaux E, Quan R-C: Cooperating elephants mitigate competition until the stakes get too high. PLoS Biol 2021, 19:e3001391. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Plotnik JM, Lair R, Suphachoksahakun W, Waal FBM de: Elephants know when they need a helping trunk in a cooperative task. Proc Natl Acad Sci 2011, 108:5116–5121. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 25. Jaakkola K, Guarino E, Donegan K, King SL: Bottlenose dolphins can understand their partner’s role in a cooperative task. Proc R Soc B: Biol Sci 2018, 285:20180948. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 26. Ortiz ST, Castro AC, Balsby TJS, Larsen ON: Problem-solving in a cooperative task in peach-fronted conures (Eupsittula aurea). Anim Cogn 2020, 23:265–275. [ DOI ] [ PubMed ] [ Google Scholar ] 27. Péron F, Rat-Fischer L, Lalot M, Nagle L, Bovet D: Cooperative problem solving in African grey parrots (Psittacus erithacus). Anim Cogn 2011, 14:545–553. [ DOI ] [ PubMed ] [ Google Scholar ] 28. Franch M, Yellapantula S, Parajuli A, Kharas N, Wright A, Aazhang B, Dragoi V: Visuo-frontal interactions during social learning in freely moving macaques. Nature 2024, 627:174–181. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 29. Meisner OC, Shi W, Fagan NA, Greenwood J, Jadi MP, Nandy AS, Chang SW: Development of a Marmoset Apparatus for Automated Pulling to study cooperative behaviors. eLife 2024, 13:RP97088. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 30. Schuster R: Cooperative coordination as a social behavior: Experiments with an animal model. Hum Nat 2002, 13:47–83. [ DOI ] [ PubMed ] [ Google Scholar ] 31. Łopuch S, Popik P: Cooperative Behavior of Laboratory Rats (Rattus norvegicus) in an Instrumental Task. J Comp Psychol 2011, 125:250–253. [ DOI ] [ PubMed ] [ Google Scholar ] 32. Avital A, Aga-Mizrachi S, Zubedat S: Evidence for social cooperation in rodents by automated maze. Sci Rep-uk 2016, 6:29517. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 33. Conde-Moro AR, Rocha-Almeida F, Sánchez-Campusano R, Delgado-García JM, Gruart A: The activity of the prelimbic cortex in rats is enhanced during the cooperative acquisition of an instrumental learning task. Prog Neurobiol 2019, 183:101692. [ DOI ] [ PubMed ] [ Google Scholar ] 34. Conde-Moro AR, Rocha-Almeida F, Gebara E, Delgado-García JM, Sandi C, Gruart A: Involvement of prelimbic cortex neurons and related circuits in the acquisition of a cooperative learning by pairs of rats. Cogn Neurodynamics 2024, 18:2637–2658. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 35. Zhang K-M, Shen Y, Jia C-H, Wang H, Bi G-Q, Lau P-M: A new paradigm of learned cooperation reveals extensive social coordination and specific cortical activation in mice. Mol Brain 2023, 16:40. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 36. Han KA, Yoon TH, Shin J, Um JW, Ko J: Differentially altered social dominance- and cooperative-like behaviors in Shank2- and Shank3-mutant mice. Mol Autism 2020, 11:87. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 37. Feng W, Zhang Y, Wang Z, Wang T, Pang Y, Zou Y, Huang H, Sheng C, Xiao M: A water-reward task assay for evaluating mouse mutualistic cooperative behavior. bioRxiv 2021, doi: 10.1101/2021.02.06.430037. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 38. Heaney M, Gray RD, Taylor AH: Keas Perform Similarly to Chimpanzees and Elephants when Solving Collaborative Tasks. PLoS ONE 2017, 12:e0169799. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 39. Cheng Y, Chen Y, Kwak M, Kempner RP, Singha R, Winslow J, Liu R, Khan U, Spangler T, Khan A, et al. : Asymmetric Social Representations in the Prefrontal Cortex for Cooperative Behavior. bioRxiv 2025, doi: 10.1101/2025.08.27.672249. [ DOI ] [ Google Scholar ] 40. Suzuki S, Niki K, Fujisaki S, Akiyama E: Neural basis of conditional cooperation. Soc Cogn Affect Neurosci 2011, 6:338–347. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 41. Tremblay S, Sharika KM, Platt ML: Social Decision-Making and the Brain: A Comparative Perspective. Trends Cogn Sci 2017, 21:265–276. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 42. Gangopadhyay P, Chawla M, Monte OD, Chang SWC: Prefrontal–amygdala circuits in social decision-making. Nat Neurosci 2021, 24:5–18. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 43. Tsao DY, Freiwald WA, Tootell RBH, Livingstone MS: A cortical region consisting entirely of face-selective cells. Science 2006, 311:670 674. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 44. Freiwald WA, Tsao DY: Functional Compartmentalization and Viewpoint Generalization Within the Macaque Face-Processing System. Science 2010, 330:845–851. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 45. Chang L, Tsao DY: The Code for Facial Identity in the Primate Brain. Cell 2017, 169:1013–1028.e14. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 46. Chen P, Hong W: Neural Circuit Mechanisms of Social Behavior. Neuron 2018, 98:16–30. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 47. Bergan JF, Ben-Shaul Y, Dulac C: Sex-specific processing of social cues in the medial amygdala. eLife 2014, 3:e02743. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 48. Remedios R, Kennedy A, Zelikowsky M, Grewe BF, Schnitzer MJ, Anderson DJ: Social behaviour shapes hypothalamic neural ensemble representations of conspecific sex. Nature 2017, 550:388–392. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 49. Kingsbury L, Huang S, Raam T, Ye LS, Wei D, Hu RK, Ye L, Hong W: Cortical Representations of Conspecific Sex Shape Social Behavior. Neuron 2020, 107:941–953.e7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 50. Noonan MP, Sallet J, Mars RB, Neubert FX, O’Reilly JX, Andersson JL, Mitchell AS, Bell AH, Miller KL, Rushworth MFS: A Neural Circuit Covarying with Social Hierarchy in Macaques. PLoS Biol 2014, 12:e1001940. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 51. Munuera J, Rigotti M, Salzman CD: Shared neural coding for social hierarchy and reward value in primate amygdala. Nat Neurosci 2018, 21:415–423. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 52. So N, Franks B, Lim S, Curley JP: A Social Network Approach Reveals Associations between Mouse Social Dominance and Brain Gene Expression. Plos One 2015, 10:e0134509. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 53. Grabenhorst F, Báez-Mendoza R, Genest W, Deco G, Schultz W: Primate Amygdala Neurons Simulate Decision Processes of Social Partners. Cell 2019, 177:986–998.e15. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 54. Yoshida K, Saito N, Iriki A, Isoda M: Representation of Others’ Action by Neurons in Monkey Medial Frontal Cortex. Curr Biol 2011, 21:249–253. [ DOI ] [ PubMed ] [ Google Scholar ] 55. Yoshida K, Saito N, Iriki A, Isoda M: Social error monitoring in macaque frontal cortex. Nat Neurosci 2012, 15:1307–1312. [ DOI ] [ PubMed ] [ Google Scholar ] 56. Putnam PT, Chu C-CJ, Fagan NA, Monte OD, Chang SWC: Dissociation of vicarious and experienced rewards by coupling frequency within the same neural pathway. Neuron 2023, 111:2513–2522.e4. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 57. Monte OD, Fan S, Fagan NA, Chu C-CJ, Zhou MB, Putnam PT, Nair AR, Chang SWC: Widespread implementations of interactive social gaze neurons in the primate prefrontal-amygdala networks. Neuron 2022, 110:2183–2197.e7. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 58. Shi W, Meisner OC, Jadi MP, Chang SWC, Nandy AS: Canonical decision computations underlie behavioral and neural signatures of cooperation in primates. bioRxiv 2025, doi: 10.1101/2025.10.22.683999. [ DOI ] [ Google Scholar ] 59. Rilling JK, Gutman DA, Zeh TR, Pagnoni G, Berns GS, Kilts CD: A Neural Basis for Social Cooperation. Neuron 2002, 35:395–405. [ DOI ] [ PubMed ] [ Google Scholar ] 60. Decety J, Jackson PL, Sommerville JA, Chaminade T, Meltzoff AN: The neural bases of cooperation and competition: an fMRI investigation. NeuroImage 2004, 23:744–751. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 61. Chang SWC, Gariépy J-F, Platt ML: Neuronal reference frames for social decisions in primate frontal cortex. Nat Neurosci 2013, 16:243–250. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 62. Chang SWC, Fagan NA, Toda K, Utevsky AV, Pearson JM, Platt ML: Neural mechanisms of social decision-making in the primate amygdala. Proc National Acad Sci 2015, 112:16012–16017. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 63. Monte OD, Chu CCJ, Fagan NA, Chang SWC: Specialized medial prefrontal–amygdala coordination in other-regarding decision preference. Nat Neurosci 2020, 23:565–574. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 64. Noritake A, Ninomiya T, Isoda M: Social reward monitoring and valuation in the macaque brain. Nat Neurosci 2018, 21:1452–1462. [ DOI ] [ PubMed ] [ Google Scholar ] 65. Ning Z, Xie L: A survey on multi-agent reinforcement learning and its application. J Autom Intell 2024, 3:73–91. [ Google Scholar ] ACTIONS View on publisher site PDF (872.1 KB) Cite Collections Permalink PERMALINK Copy RESOURCES Similar articles Cited by other articles Links to NCBI Databases Cite Copy Download .nbib .nbib Format: AMA APA MLA NLM Add to Collections Create a new collection Add to an existing collection Name your collection * Choose a collection Unable to load your collection due to an error Please try again Add Cancel Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top