Central Questions for Social Neuroscience Research

Abstract

Over the past decade, interest in social neuroscience has expanded rapidly. This growth raises a fundamental inquiry: what are the central questions that now define the field of social neuroscience? While the answer to this question depends on the appropriate neurobiological level of explanation, ranging from cellular substrates to systems and network-level dynamics, there are commonalities that transcend multiple levels. One longstanding question in the field of social neuroscience concerns whether there is social specificity in the brain. If a neural substrate exhibits social specificity, does it do so at the algorithmic level, the implementational level, or both? Another emerging question concerns how neural systems construct and update internal models of the social world composed of other agents and factors of one's social environment. These internal models of the social world must integrate social information throughout one's lifespan, adding a critical developmental component in building social world models. Relatedly, another core question concerns internal states that dynamically guide social behavior and social cognition. These internal states may have components that are innately driven or acquired by learning in the social world. With powerful neuroscientific tools and behavioral sophistications, the field is gaining a great amount of traction to begin to answer such central questions. This special collection brings together seven selected contributions that reflect cutting-edge and cross-species research in the field of social neuroscience. As editors, we encourage readers to consider the aforementioned and other big-picture questions in social neuroscience as they engage with the manuscripts in this issue.

Understanding the Social Brain

What is the social brain? To answer this question, it is crucial to consider the level of socially specificity which can be found in the nervous system among computational, algorithmic, and implementational levels (Lockwood et al., 2020). Importantly, there could be dissociations between algorithms and implementations that support a social behavioral goal. For example, a socially specific algorithm (e.g., theory of mind or multi-agent reinforcement learning) could be implemented in a neural system that also supports other cognitive operations (e.g., general inference or reinforcement learning). Conversely, there could be a neural implementation (e.g., a cell type or a specific neuromodulator) that is specialized for processing social information.

With social information processing as an opening topic, this special collection begins with a review by Yang and colleagues that proposes a multi-level framework for understanding how the brain processes referential signals in social contexts, such as social cues that direct others’ attention and establish shared understanding (Yang et al., 2026). Integrating evidence from developmental psychology, neuroimaging, computational modeling, and cross-species studies, the authors discuss that referential signals operate across three interacting algorithmic levels: innate, acquired, and deliberative. At the innate level, evolutionarily conserved subcortical and cortical circuits support early emerging sensitivity to biologically salient cues such as eye gaze and biological motion, enabling rapid and automatic social attention from infancy. At the acquired level, social learning gives rise to symbolic and culturally established referential signals, including abstract cues and conventions. At the deliberative level, individuals flexibly generate and interpret referential signals by modeling others’ beliefs and goals using inferential processes in novel or uncertain communicative contexts. The authors also discuss distinct yet overlapping neural substrates supporting each type of algorithm, spanning subcortical pathways, temporal cortices, and prefrontal regions, emphasizing that these systems are not independent but dynamically interact across development and context.

Referential processing in social contexts can become efficient with symbolic computations. In the next piece, Kaneko and colleagues propose a framework for understanding the brain as a mentalizing machine using a symbol grounding mechanism (Kaneko et al., 2026). Mentalizing, the ability to infer others’ beliefs, desires, and intentions, is viewed as a form of hidden state inference under a predictive coding framework, akin to inference processes used by sensory and motor systems. By combining evidence from computational modeling, artificial neural networks, and neuroscience, the authors propose that what uniquely distinguishes mentalizing is not merely the estimation of latent states, but the grounding of those states in one's own subjectively accessible mental experiences. Here, symbol grounding algorithm enables efficient and interpretable social inference without requiring new learning of latent spaces by linking inferred representations of others’ internal states to self-referential mental representations. This work thus proposes a specific algorithm for performing mentalizing operation in which self-mental models are used to simulate others, supported by predictive coding, mirror-like action representations, and perspective-taking. It also discusses potential brain areas implementing these algorithms, including medial prefrontal cortex, temporoparietal junction, superior temporal sulcus, and cerebellar–prefrontal circuits.

Principles Guiding Social State Dynamics

Internal states dynamically regulate social behavior and cognition. In this special collection, Cressy and colleagues describe a unifying and mechanistic framework for social homeostasis, proposing dynamic range plasticity as a key mechanism by which the brain regulates social behavior across changing environments (Cressy et al., 2026). Building upon classical homeostasis models, the authors propose that social systems maintain stability not only through effector responses and set-point shifts but also by flexibly adjusting the range of social experience. Critically, this dynamic range plasticity algorithm is proposed to be implemented via three nodes that can be calibrated by experience—detector, controller, and effector. At the detector level, repeated exposure to predictable social cues alters sensitivity to deviation signals. Within control centers, particularly prefrontal and cingulate circuits, prior experience modulates predictive computations, top-down regulation, and integration of social statistics. Effector systems, including mesolimbic dopamine, stress hormones, and neuromodulatory pathways, show history-dependent remapping of social valuation and action. As a result, dynamic range is shaped by an individual's cumulative social history that regulates effector activation and prevents maladaptive outcomes from repeated set-point shifts. The authors further propose that the width of a dynamic range indexes social tolerance, with broader ranges supporting resilience and flexibility, and narrower ranges leading to hypersensitivity to perturbations. This dynamic range plasticity of social homeostasis can help explain how early caregiving environments and cultural practices affect social tolerance over the lifespan. Moving forward, the dynamic range plasticity hypothesis generates testable predictions for understanding individual differences in social resilience, vulnerability, and psychopathology.

Prosocial and Collective Behaviors

Prosocial behaviors and collective behaviors represent the pinnacles of advanced social cognition. In this special collection, Zhang and colleagues present evidence across humans, nonhuman primates, and rodents to examine the neural mechanisms underlying prosocial behavior toward others experiencing negative states (Zhang et al., 2026). Prosocial behaviors, such as consolation, helping, and targeted care, are distinguished from empathy and emotional contagion, as recognizing or sharing another's distress does not necessarily result in action. The authors review ethologically grounded studies showing that many forms of prosocial behavior are evolutionarily conserved, while more complex helping behaviors display species-specific elaborations. Across the mammalian species examined, the anterior cingulate cortex (ACC) emerges as a critical hub that integrates information about others’ states, encodes vicarious and shared experiences, and motivates prosocial behavior. Human neuroimaging and lesion studies, nonhuman primate single-unit recordings, and rodent causal manipulation work converge to show that ACC encodes others’ rewards and distress, coordinate with limbic structures such as the basolateral amygdala (BLA), and are necessary for acquiring and expressing prosocial preferences. The authors also identify major open questions, including how emotional recognition, emotional sharing, and social motivational processes are causally linked, and call for experimental designs that independently manipulate these components to better understand prosocial behavior and its dysfunction in psychiatric conditions.

Moving beyond dyadic interaction contexts, Williams and colleagues next review emerging work in the field to discuss the neural bases of collective behavior and group interactions across species (Williams et al., 2026). The authors define collective behavior as emergent group-level dynamics arising from local interactions among individuals and emphasize that such behaviors cannot be reduced to single-agent decision-making. Drawing on studies in insects, fish, birds, mammals, nonhuman primates, and humans, the review highlights both conserved principles, such as alignment, conformity, and distributed feedback, and species-specific adaptations linked to social complexity and communication. The authors consider major analytical frameworks for studying collective behavior, including swarm intelligence and agent-based models, along with technical advances enabling high-resolution measurement of group dynamics in naturalistic settings. At the neural level, converging evidence implicates distributed circuits spanning prefrontal cortex, medial temporal lobe, amygdala, and sensory systems in representing social identity, rank, group structure, and collective decision variables. The authors also describe the roles of neuromodulators, such as dopamine, serotonin, oxytocin, and stress hormones, in group cohesion and other collective phenomena.

Neuromodulatory Control of Prosocial Motivation

The last two contributions in this special collection showcase how neuromodulatory systems shape social behavior and cognition. Neuromodulators are at the center of tuning circuits toward goal-appropriate social behaviors and likely play a major role in dynamically routing relevant information across larger brain networks.

Meisner and colleagues investigate how and when local oxytocin signaling in the primate BLA sustains prosocial behavior by modulating neural activity and communication between BLA and the rostral gyrus of ACC (ACCg; Meisner et al., 2025). Using a social reward allocation task combined with focal oxytocin infusion into BLA and simultaneous electrophysiological recordings from BLA and ACCg, the authors demonstrate that oxytocin's effects on prosocial decision-making are state dependent. When monkeys exhibited high baseline prosocial motivation, oxytocin infusion into the BLA prevented the natural decline in prosocial choices and task engagement over time. However, there was no such effect when baseline prosocial motivation was low, indicating that oxytocin processing in BLA amplifies existing social motivation rather than inducing prosociality. Oxytocin selectively enhanced BLA and ACCg activity for prosocial choices, but only during high prosocial states, and stabilized and strengthened bidirectional BLA-ACCg communication, counteracting the degradation of interareal coordination seen in control conditions. The authors propose a model in which oxytocin acts as a state-dependent gain modulator, enhancing socially relevant representations in the amygdala and stabilizing amygdala–prefrontal interactions that support sustained prosocial behavior. More broadly, this work identifies a social state-dependent circuit mechanism in the primate brain underlying oxytocin's context-dependent social effects.

Examining the link between neuromodulation and prosocial motivation in humans, Talbot and colleagues investigate the causal role of dopamine in shaping prosocial motivation (Talbot et al., 2025). Using a within-subject design, the authors asked whether dopaminergic modulation alters willingness to exert physical effort to benefit others in patients with Parkinson's disease (PD) both “on” and “off” dopaminergic medication. In an effort-based decision-making task, participants chose between a low-effort/low-reward option and a high-effort/high-reward option for benefiting either themselves or an anonymous other person. Both PD patients and controls showed an expected self-bias, being less willing to exert effort for others than for themselves. However, when the medication increased the dopaminergic tone in otherwise low-dopamine states, PD patients became more willing to exert effort for others without affecting the willingness to exert effort for self. These findings suggest that dopamine modulates motivation for higher-order social goals or reduces the subjective cost of effort of prosocial actions, thereby decreasing the tendency to disengage from prosocial acts under fatigue. This study provides causal evidence in humans that prosocial motivation is shaped by dopaminergic tone.

Conclusion

Collectively, the contributions in this special collection illustrate how social neuroscience is converging on a set of unifying principles that span computational, algorithmic, and implementational levels of analysis. Across diverse species, methods, and theoretical perspectives, these contributions emphasize that social cognition cannot be understood solely in terms of isolated brain regions or single-agent computations. Instead, social behavior emerges from interactions among internal states, learned models of others and environments, and neuromodulatory systems that dynamically tune neural circuits and route social information according to context, experience, and motivational demands. These contributions highlight the importance of integrative and ethologically grounded frameworks that link development, computation, and circuit-level mechanisms to better understand the behavioral and neural mechanisms of social world modeling.

A recurring theme across the contributions is the need to bridge levels of explanation, relating between neural algorithms and neural implementation under specific social computational goals, and to connect individual cognition with dyadic and group interactions and even societal level phenomena. In doing so, social neuroscience will be better positioned to explain individual differences, social resilience and vulnerability, and the emergence of social dysfunction. We hope that this special issue serves as a catalyst for new conceptual and empirical approaches to understanding the social brain.