Dopaminergic mechanisms of dynamical social specialization

Solié, C.; Nicolson, A.; Justo, R.; Layadi, Y.; Morin, B.; Batifol, C.; Reynolds, L. M.; Le Borgne, T.; Fayad, S. L.; Gulmez, A.; Rodriguez Quevedo, Y.; Allegret-Vautrot, J.; Centene Guglielmi, G.; de Chaumont, F.; Didienne, S.; Debray, N.; Hardelin, J.-P.; Girard, B.; Mourot, A.; Naudé, J.; Viollet, C.; Marti, F.; Delord, B.; Faure, Ph.

doi:10.1038/s41586-026-10301-4

Download PDF

Article
Open access
Published: 01 April 2026

Dopaminergic mechanisms of dynamical social specialization

Nature (2026) Cite this article

36k Accesses
65 Altmetric
Metrics details

Subjects

Abstract

Social organization and division of labour are fundamental to animal societies^1,2,3, yet how these structures emerge from individual interactions and are shaped by neuromodulation remains unclear. Here, using behavioural tracking in a semi-natural environment, neural recordings and computational models that integrate reinforcement learning and social condition, we show that triads of isogenic mice develop specialized roles spontaneously while solving a foraging task under social constraints. Notably, despite minor intra-sex differences in behaviour when mice were tested alone, male triads formed stable worker–scrounger relationships driven by competition, whereas female triads adopted uniform, cooperative strategies. These sex-divergent roles were shaped by dopaminergic activity in the ventral tegmental area. Model analysis revealed how intra-sex and inter-sex parameter differences in resource exploitation, combined with contingent social interactions, drive behavioural specialization and division of labour. Most notably, it highlighted how contingency, amplified by competition, magnifies individual differences and shapes social profiles. The plastic, adaptive nature of social organization was apparent when sex mixing or reintroducing experienced individuals into naive groups reshaped role distribution. Furthermore, dopaminergic manipulations confirmed this plasticity, reshaping roles and altering group structure. Our findings support a multi-scale feedback loop whereby social context shapes neural states, which in turn reinforce behavioural specialization and stabilize social structures.

Convergent evolution of a conserved molecular network underlies parenting and sociality

Article 04 November 2025

Automatic mapping of multiplexed social receptive fields by deep learning and GPU-accelerated 3D videography

Article Open access 01 February 2022

Spontaneous behaviour is structured by reinforcement without explicit reward

Article Open access 18 January 2023

Main

Social animals interact and coordinate their behaviours to form social organizations—structured patterns of relationships and interactions that stabilize into collective forms (for example, division of labour and norms) and specialized roles^3,4,5,6,7. A key challenge in behavioural science is deciphering how individual differences in behaviour emerge, how they contribute to collective actions and how they are linked to neural activity. While the broad effects of social organization on individual behaviour and specialization are well documented^8,9,10,11,12, the underlying cognitive and neurophysiological mechanisms remain largely unexplored, mainly because of the difficulty of studying such mechanisms in controlled, artificial social settings that also allow neurobiological investigation.

In social groups, the production of and access to shared resources drive strategies such as competition or cooperation^2,8,13. The producer–scrounger game illustrates this dynamic: some individuals produce resources, others exploit them^1,2,14,15. Thus, foraging strategies span from independent food acquisition to exploiting others’ efforts, balancing effort, risk and reward under ecological and social constraints¹³. Evolutionary game theory¹⁶ predicts stable equilibria between such strategies, but typically assumes that behaviours are predetermined and stable¹⁷, overlooking the internal mechanisms that guide an individual’s behaviour, including the neural and cognitive processes that grant adaptative, learning-based flexible strategies^17,18,19. Central to this adaptability is dopaminergic signalling, which reinforces rewarded actions by signalling when reward is larger than expected^20,21,22,23, including social reward^24,25. A second key process is the exploration–exploitation trade-off^26,27,28,29, whereby individuals must choose between exploiting known options and exploring alternatives, a process in which dopamine has also been implicated^{30,31,32,33,34,35}.

Here we hypothesize that social foraging strategies, including producer–scrounger dynamics, emerge flexibly through socially constrained reinforcement learning. In particular, we suggest that dopaminergic signalling shapes variability in learning and decision policies, thereby contributing to behavioural specialization. As social organizations forms, individuals adjust actions on the basis of expected payoffs, which in turn modify neural activity and learning rules, ultimately stabilizing social roles. Recent advances in continuous animal tracking and behaviour quantification in semi-naturalistic settings^{36,37,38,39,40} enable detailed measurements of how individuals interact and adapt their strategies over time. Using these approaches, we examined how foraging specialization emerges in small groups (n = 3) of isogenic mice housed in a controlled semi-natural environment.

Two foraging strategies in lone mice

We first assessed mice foraging behaviour and underlying neural mechanisms in a lone context. Female or male mice (n = 62) were placed alone in a 50 cm × 50 cm multi-compartment environment and tracked continuously for 5 days and 4 nights using the Live Mouse Tracker system³⁶. This setup includes a lever on one side and a food dispenser with a beam on the opposite side, enabling mice to obtain pellets by pressing the lever and consume them at any time (Fig. 1a). Mice displayed principally nocturnal foraging, with lever presses and nose pokes in the food dispenser occurring primarily during the dark cycle, as expected (Fig. 1b). Over time, the number of lever presses (#LP) increased and the number of nose pokes decreased, indicating that mice had learned the association between lever press and food retrieval (Fig. 1c, left, and Extended Data Fig. 1a). Accordingly, trajectory durations between lever press and food dispenser showed a skewed distribution, with 54% having a duration of less than 6 s, which we defined as a complete sequence (CS; Fig. 1c, right). The percentage of complete sequences (%CS = 100 × number of complete sequences divided by #LP) increased over sessions, stabilizing at 35%, suggesting an instrumental use of lever presses (Fig. 1c and Extended Data Fig. 1a).

**Fig. 1: Different reward-seeking strategies emerge in a non-social semi-natural environment.**

To explore the neural bases of this association, we used in vivo fibre photometry in DAT-Cre mice expressing GCaMP7c in dopaminergic neurons of the ventral tegmental area (VTA) (Fig. 1d). A phasic increase in VTA dopaminergic neuron activity was observed at the time of food delivery, regardless of the sequence completeness or of learning, consistent with reward signalling (Fig. 1d). By contrast, we found a specific phasic increase in VTA dopaminergic neuron activity at the time of lever press, but only after learning and for complete sequences, consistent with the prediction of food reward by lever press²⁰. No such increase at the lever press time was observed for incomplete sequences, regardless of whether the mouse moved afterwards. This indicates that the increase characterizing complete sequences was not related to the movement of the mouse (Extended Data Fig. 1b,c). Additionally, we found a correlation between VTA dopaminergic activity at lever press and the time taken to reach the food magazine (Extended Data Fig. 1c, bottom). These findings further suggest that lever presses associated with complete sequences acquired an instrumental status during learning, reflected in reward prediction-like VTA dopaminergic activity.

We next analysed individual decision-making strategies using archetypal analysis^32,38,40—that is, expressing individuals as a weighted combination of archetypal strategies—a dimensionality reduction method rooted in the concept of optimality with respect to multiple evolutionary constraints⁴¹ (Methods). Archetypes based on #LP and %CS from day 1 to day 4 identified two distinct profiles: Achievers and Storers (n = 31 each; Fig. 1e,f and Extended Data Fig. 1d,e). The two groups pressed the lever equally often, but differed in how they collected rewards. Storers frequently left pellets unretrieved, whereas Achievers completed a higher proportion of sequences (Fig. 1e–g and Extended Data Fig. 1f). Thus, both engaged the task, but differed in the immediacy of reward retrieval. Female mice were about twice as likely as male mice to be classified as Storers and clustered closer to this archetype in the behavioural space (Fig. 1h,i). Post-experiment body mass differed between sexes but not between archetypes (Fig. 1j). Finally, spontaneous VTA dopaminergic neuron activity, recorded post-experiment in anaesthetized males, showed higher firing frequency in Achievers than in Storers (Fig. 1k and Extended Data Fig. 1g), suggesting a link between VTA dopaminergic activity and foraging strategies^32,37,38.

Sex-dependent differences between triads

We next explored the foraging strategies of mice (n = 195) in a social context, by housing triads of either all-male or all-female task-naive mice for 8 days and 7 nights (Fig. 2a,b). Lever pressing was predominantly nocturnal (Fig. 2b). Male and female mice performed a similar number of lever presses, but females achieved significantly fewer complete sequences (Fig. 2c and Extended Data Fig. 2a), reflecting higher latencies between lever press and food retrieval, and more consecutive lever presses before consumption (Extended Data Fig. 2b). Archetypal analysis of eight behavioural features (Fig. 2d)—%LP intra-cage (#LP per mouse divided by the total #LP in the cage on that day) and %CS over the past 3 days, plus the number of food pellet gains and losses due to conspecifics (Methods)—revealed a triangular behavioural space with Storer, Worker and Scrounger archetypes (Fig. 2d and Extended Data Fig. 2c,d). Storers pressed frequently but completed few sequences, leaving pellets temporarily in the food dispenser (thus delaying reward retrieval), a pattern consistent with the Storer archetype in the lone context. Despite leaving pellets available to others, Storers rarely engaged in competition, resulting in low numbers of food losses and gains (Fig. 2e). By contrast, Workers performed frequent complete sequences and produced food that was often consumed by others, whereas Scroungers rarely pressed the lever and primarily consumed food earned by cage mates (Fig. 2e,f). Workers and Scroungers showed clear signs of competition, with opposite food gain and loss patterns (Fig. 2e) and lever press strategies (Fig. 2f and Extended Data Fig. 2e). Spatial analyses revealed that the lever press is a salient event for Workers and especially Scroungers, who oriented towards the feeder zone following conspecific lever presses, whereas Storers showed no such modulation (Extended Data Fig. 3a,b). These roles emerged rapidly, with clear distinctions already on day 1 that remained stable thereafter (Extended Data Fig. 3c–e). Indeed, within hours on day 1, Workers showed frequent complete sequences, Scroungers pressed little, and Storers pressed often but rarely completed sequences (Extended Data Fig. 3c).

**Fig. 2: Distinct sex-dependent strategies emerge within groups.**

Distance to archetypes differed between sexes, with females being closer to the Storer archetype and males being closer to the Scrounger (Fig. 2g) and Worker archetypes. This sex bias was evident both at the individual and group levels: 86% of females were classified as Storers, whereas 87% of males belonged to the Worker or Scrounger archetypes (Fig. 2h). Correspondingly, triad composition deviated from chance, with 75% of female triads composed solely of Storers and 76% of male triads combining Workers and Scroungers (Methods; Fig. 2i), suggesting that intrinsic mechanism and social interactions constrain admissible group configurations: for instance, triads of three Scroungers were very rare. Workers and Scroungers did not differ in final body mass, and the lower final mass in Storers (Fig. 2j) simply reflects the overrepresentation of females, which are lighter overall, in this group. Tube tests, performed before and after the microsociety experiment (Extended Data Fig. 4a), revealed that dominance hierarchies and foraging roles evolved independently, with no consistent relationship between dominance rank and archetypal distance in either sex (Extended Data Fig. 4b). Thus, although hierarchies and foraging roles evolve in groups, they appear to reflect independent dimensions of social organization.

Dopaminergic correlates of social roles

To investigate the neural underpinnings of social roles, we recorded VTA dopaminergic activity during the task with fibre photometry in DAT-Cre mice (n = 19; Fig. 3a,b). All mice showed dopamine transients at food delivery, but the responses at lever presses differed with sex and profile (Fig. 3c,d). Analysis of the results on day 1 of the experiment showed that dopamine transients were already detectable at food delivery in both sexes (Extended Data Fig. 5a), whereas responses related to lever press were significant only in males (Extended Data Fig. 5b).

In females (Fig. 3c), dopaminergic activity increased at food delivery and at their own lever presses. In males (Fig. 3c–d and Extended Data Fig. 5c), Workers responded to their own lever presses, consistent with reward prediction, whereas Scroungers responded to conspecific lever presses, suggesting that for Scroungers, the actions of others became predictive of reward. Notably, the responses of Scroungers were state-dependent: dopamine transients occurred only when mice were oriented towards the lever or dispenser (that is, focused), but not when engaged elsewhere (unfocused) (Fig. 3e and Extended Data Fig. 3a,b). To relate dopaminergic responses to the behavioural space, we analysed dopaminergic peak activity in each mouse as a function of its distance to the three archetypes. Dopaminergic responses to own lever presses were stronger near the Worker and Storer vertices than near the Scrounger one (Fig. 3f and Extended Data Fig. 5d–f), whereas responses to conspecific lever presses peaked near the Scrounger vertex (Fig. 3g and Extended Data Fig. 5e–g). These role-dependent activity patterns cross-validate behavioural archetypes and show that VTA dopaminergic dynamics reflect individual roles within the group. Post-task electrophysiological recordings of VTA dopaminergic neurons in anaesthetized mice (Extended Data Fig. 5h) revealed further sex- and role-dependent effects in firing rate (Fig. 3h,i) and bursting activity (Extended Data Fig. 5i,j). Although baseline dopaminergic activity (that is, before the microsociety task) was comparable between sexes, group housing increased firing in males but not in females, leading to a post-task difference (Fig. 3h and Extended Data Fig. 5h). At the individual level, dopaminergic tone was associated with behavioural specialization: mice closer to the Storer archetype showed lower firing rates than those aligned with Worker or Scrounger profiles (Fig. 3i and Extended Data Fig. 5j). These findings suggest that social experience induces sex-dependent plastic changes in dopaminergic activity, and that dopaminergic tone is associated with the roles adopted by individuals within the microsociety.

Explore–exploit trade-off shapes roles

To unravel the mechanisms of specialization, we developed a Q-learning model (Methods) in which ‘e-mice’ learned to find food in an environment (Fig. 4a) with six spatial states, including a lever press and a food dispenser. Over time, state transition Q-values formed a spatial gradient towards the food dispenser. Specialization could arise from the interaction of intra- or inter-sex individual parameter variability, social constraints (lever press accessibility or food availability) or contingency (individual random choices). We first assessed how behavioural parameters accounted for specialization in the lone (one e-mouse) and social (three e-mice) conditions. Learning rate (α) and temporal discount factor (γ) parameters had moderate effects, whereas the inverse temperature (β), controlling the exploitation–exploration trade-off (high-valued options versus random decisions), was critical (Extended Data Fig. 6a–e). In the lone context, varying β values (Fig. 4a,b) reproduced behavioural patterns observed in mice (Fig. 4c). A lower β favoured exploration (with shallower gradients (Extended Data Fig. 6a) and erratic trajectories, resulting in a lower %CS) (Extended Data Fig. 6b,c), resembling Storers, whereas higher β promoted exploitation (Extended Data Fig. 6b,c), as in Achievers. Thus, introducing sex-dependent β variability, with males exhibiting higher β values, replicated sex differences observed in isolated mice (Fig. 4d). When moving from individuals to groups, even with identical β values within a group, e-triads with high β values (β ≥ 1) favouring exploitation displayed distinct gradients and specialization into Workers and Scroungers, with individual roles determined by contingent dynamics (Fig. 4e and Extended Data Fig. 7a,b), whereas e-triads with low β values (β ≤ 1) favouring exploration displayed no specialization (Extended Data Fig. 6d). Consequently, sex-dependent β distributions closely matched both the archetypal distributions and triad compositions observed experimentally (Fig. 4f,g). In mixed groups (two low β and one high β), all profiles emerged: high-β individuals mainly became Scroungers, whereas low-β individuals became Workers or Storers (Extended Data Fig. 7c–f). This shows that role adoption depends not only on individual β values but also on group composition. For example, low-β females alone do not produce Workers (Extended Data Fig. 7a), yet the same β values in females produce Workers when two low-β females are combined with a high-β male (Extended Data Fig. 7e).

**Fig. 4: Reinforcement learning model and emergent strategies.**

Competition and contingency shape roles

We assessed the necessary conditions, in terms of decision parameters, for social behavioural specialization to occur in a mathematically tractable reduced model version (Methods) with two e-mice and two locations (lever and food dispenser). Differentiation arose through symmetry breaking at a supercritical pitchfork bifurcation at β = β_bifurc (Fig. 4h). Below β_bifurc (female e-mice), the single fixed point for Q-values towards D (q_D) (and other probabilities; Extended Data Fig. 8a-c) corresponded to Storers. Above β_bifurc (male e-mice), two stable fixed points branches emerged, corresponding to Workers and Scroungers (bistability; Fig. 4h and Extended Data Fig. 8a–c).

Qualitative analysis⁴² revealed the causal mechanism underlying symmetry breaking and behavioural bistability in male e-mice (Fig. 4i–k and Extended Data Fig. 8d,e). Initially, contingency (individual random choices) determined which e-mouse first accessed the dispenser. Positive feedback between reinforcement learning and action selection (softmax) processes led one e-mouse (Scrounger) to predominantly occupy the dispenser area, reducing food access for the other e-mouse and increasing its lever press probability (Worker). Competition for shared resources thus drives specialization in groups of male e-mice, with higher β values intensifying this effect by promoting greater resource exploitation. By contrast, female e-mice maintained uniform behaviours, owing to lower levels of exploitation and competition. Thus, β emerged as a critical sex-dependent control parameter accounting for female uniformity versus male specialization, where specialization arose mainly from task contingency.

These theoretical results allowed us to make three predictions. First, introducing a high-β mouse (male) into a group of low-β mice (females) should disrupt female uniformity, leading some to adopt distinct roles, and increase the propensity of the single male to become a Scrounger (Extended Data Fig. 7d–f). Second, placing a naive individual into a group with a pre-established Scrounger should lead to specialization and adoption of complementary roles through competitive dynamics. Third, manipulating β value biologically via dopaminergic activity should shift behavioural strategies^30,31,32,34: decreasing β should reduce competition and promote Storers, whereas increasing β should increase competition and promote specialization.

Triad composition and history shape roles

To test whether interacting with high β individual drives differentiation in low-β mice, we formed mixed-sex triads composed of one male (high β) and two females (low β) (Fig. 5a). The model predicted that this would eliminate uniform Storer strategies and promote specialization, with the male adopting a Scrounger role (Fig. 4k and Extended Data Fig. 7c–f). In such triads, the male performed fewer lever presses than females, but unlike in sex-matched triads, there was no sex difference in %CS (Extended Data Fig. 9b). Although both sexes were equally distant from the Storer archetype, males were significantly closer to the Scrounger archetype (Extended Data Fig. 9c). The overall distribution was characterized by a marked reduction of Storers (Fig. 5a) and showed Worker–Scrounger combinations in most triads (Extended Data Fig. 9d), supporting the hypothesis that behavioural specialization emerges from the set of β values within the group, beyond individual values.

We next assessed how stable a Scrounger role is in dynamic social contexts by retaining one male Scrounger from a male triad and housing it with two naive males during a second week of group living (Fig. 5b). By the end of the second week, the previously specialized Scrounger had significantly increased its #LP (Extended Data Fig. 9e) and its position in archetypal space had shifted away from the Scrounger pole in 85% of cases (11 out of 13; Extended Data Fig. 9g), adopting either a Worker (7 out of 13) or a Storer (4 out of 13) profile. This suggests that behavioural specialization is neither fixed nor intrinsic, but emerges dynamically with a β-like decision parameter setting predispositions that are resolved through social contingency and interaction history.

Pre-task dopaminergic modulation shifts roles

VTA dopaminergic activity has been linked to exploration–exploitation parameters^{30,31,32,33,34}, suggesting that dopaminergic tone could bias behavioural specialization. At the end of the microsociety experiment, males showed higher spontaneous VTA dopaminergic activity than females (Fig. 3h,i); however, this difference is likely to reflect social experience during group living rather than a pre-existing signature. We therefore reasoned that altering dopaminergic tone could reshape group strategies by controlling the tendency towards social uniformity or specialization. We tested this by increasing baseline VTA dopaminergic activity in males or lowering it in females.

We used a pre-task stimulation protocol^43,44 to test whether the dopaminergic tone at the onset of social interactions biases subsequent role adoption through its effect on the exploration–exploitation balance. In male DAT-Cre mice⁴³, we selectively increased basal VTA dopaminergic activity by using excitatory channelrhodopsin (ChR2) (Fig. 5c and Extended Data Fig. 10h,i) before the start of the microsociety experiment, rather than during the experiment, to avoid confounding effects of affecting dopaminergic activity during the lever presses. Electrophysiological recordings confirmed that this stimulation reliably increased VTA dopaminergic neuron excitability for at least 6 h post-stimulation (Extended Data Fig. 10a–d). Three conditions were tested: (1) YFP-only mice (control); (2) triads with one ChR2-stimulated mice (1ChR2); and (3) triads with three ChR2-stimulated mice (3ChR2) (Fig. 5c). Archetypal analysis revealed a redistribution of behavioural profiles, with the emergence of Storers specifically in 3ChR2 triads (Fig. 5d). The distance to the Storer archetype was significantly reduced in the 3ChR2 groups compared to controls (Extended Data Fig. 9j), and triad composition shifted to more Storers (Extended Data Fig. 9k). In 1ChR2 triads, this redistribution of archetypal profiles was less pronounced, but it extended to non-stimulated members, suggesting that altering dopaminergic tone in a single mouse can also reshape group dynamics.

The complementary prediction in females was tested by reducing VTA dopaminergic activity with inhibitory DREADDs (hM4Di) activated by clozapine N-oxide (CNO) 24 h before the microsociety experiment and during the first day of the experiment (Fig. 5e and Extended Data Fig. 9l,m). We compared female triads composed of: (1) three mCherry mice (control); (2) one hM4Di-expressing mouse (1hM4Di); or (3) three hM4Di-expressing mice (3hM4Di). Compared with control triads, 3hM4Di females showed a marked shift in behavioural space, with male-typical Scroungers strategies emerging (Fig. 5f). Distance to the Scrounger archetype decreased (Extended Data Fig. 9n), and several triads developed Worker–Scrounger–Scrounger configurations dominated by Scroungers (Extended Data Fig. 9o). By contrast, the 1hM4Di condition had no detectable behavioural effect.

Together, these results confirm that manipulating dopaminergic tone biases the emergence of specific behavioural strategies in a sex-dependent manner. This highlights the complex interplay between VTA dopaminergic neurophysiology and the social environment in shaping individual strategies, providing a neural basis for the adaptive specialization in mouse microsocieties.

Discussion

This study highlights the importance of exploring social organization through the lens of division of labour to gain deeper insights into behavioural specialization in animal microsocieties. Evolutionary approaches conceptualize behavioural specialization as the outcome of stable strategies^16,45, leading to fixed, genetically determined social roles. By contrast, our findings reveal a dynamic, context-dependent process, in which roles emerge through learning and repeated social interactions rather than reflecting intrinsic traits. As a result, behavioural roles can be reshaped when group composition changes^17,19. Moreover, social norms, conceptualized here as implicit rules governing group behaviour, emerge rapidly from individual history and the underlying structure of dynamical interaction, guiding both individual roles and collective dynamics.

A key contribution of this mouse microsociety paradigm is the demonstration that an individual cognitive parameter (β) (the exploration–exploitation trade-off), operating near to a bifurcation, is sufficient to constrain and shape emergent social organization. Our models and experiments suggest that β reflects dopamine-dependent control of decision variability, thereby linking VTA activity, learning dynamics and social organization. Social competition increases tonic dopaminergic activity of VTA neurons in males and promotes Worker–Scrounger specialization, whereas optogenetic stimulation of these dopaminergic neurons leads to the opposite outcome, favouring Storer-like behaviours. These findings indicate that the β value is not simply determined by the overall firing rate of these dopaminergic neurons, but by interactions between distinct dopaminergic components: tonic, non-specific activity (as during optogenetic activation) and phasic, event-locked responses amplified by competition. The β value may also depend on the salience or signal-to-noise ratio⁴⁶ of phasic dopaminergic signals, explaining why competition may enhance dopamine salience and behavioural differentiation, whereas optogenetic activation may blur this distinction. Consistent with this possibility, early (day 1) photometry recordings revealed robust food-related dopamine transients in both sexes, but a clearer lever press-related phasic response in males than in females (Extended Data Fig. 5a,b), suggesting that differences in dopamine salience at the time of lever press may already exist before role differentiation unfolds. Increasing VTA dopaminergic tonic baseline activity with optogenetic stimulation may reduce the salience of phasic, task-related responses, lowering β and favouring less differentiated behaviour, and consequently shifts males towards female-like Storer profiles, whereas decreasing activity restores Worker–Scrounger specialization. Although this mechanism explains our findings, it is likely to be only one among several processes through which dopaminergic tone can influence role adoption. Indeed, changes in tonic activity of VTA dopaminergic neurons could also affect arousal, stress reactivity, motivation or social drive—additional routes through which β, as an effective behavioural parameter, may be shifted.

Consistent with this, neural recordings showed that Workers display VTA responses to their own lever presses, whereas Scroungers respond to conspecific presses, indicating that phasic dopaminergic activity is event-specific and role-dependent. These responses scaled with the position in the archetypal space, supporting their relevance to behavioural specialization. In addition, by manipulating tonic dopaminergic activity before social engagement, we further demonstrated that baseline neuromodulatory state biases group composition. Although our manipulations targeted VTA neurons globally and dopamine can also influence arousal or motivation, the effects were reproducible and consistent between groups. Fibre photometry and electrophysiology together show that both phasic signals and long-lasting plasticity contribute to the emergence and stabilization of specialization. Together, our results indicate that behavioural specialization arises not only as a response to social constraint, but also through a feedback loop in which social context modulates dopaminergic activity, which in turn consolidates and stabilizes social roles.

Many studies have documented sex effects on social behaviours^{47,48,49,50,51}. A notable aspect of our findings is the pronounced sex-dependent divergence in social organization between all-male and all-female mouse triads, despite only minor sex differences when mice were tested alone. Male triads exhibited distinct and stable divisions of labour, with roles such as Workers and Scroungers, whereas female groups displayed remarkable uniformity, predominantly adopting Storer strategies. These differences highlight how sex-dependent cognitive⁵² and neurophysiological differences interact with social environments to shape collective dynamics. In males, high β values promote competition and role specialization, whereas lower β values in females favour uniformity. Nonetheless, β differences alone cannot explain why entire groups of female mice adopt the same strategy. This uniformity points to group-level mechanisms—reduced competition and reinforcement of homogeneity—that collectively stabilize Storer profiles. Consistent with recent findings⁴⁹, social competition amplifies behavioural divergence in males but has weaker effects in females, highlighting sex-dependent adaptations to social dynamics. Although dominance is associated with greater reward motivation and risk taking⁵³, and dopamine signalling contributes to such social hierarchy^54,55,56,57, we show that specialized roles in resource acquisition can emerge independently of social rank, pointing to partly orthogonal dopaminergic mechanisms underlying social hierarchy and behavioural specialization.

Early life contingency has a crucial role in shaping behavioural and social outcomes. Small initial differences, whether due to chance or experience, can be amplified through feedback loops, producing divergent trajectories^{12,49,58,59,60}. In competitive social environments, such as those often encountered by male mice, these feedback loops amplify disparities and promote the emergence of different roles. By contrast, female mice may experience reduced competitive pressure, allowing more-equitable outcomes within their groups. In our model, the distributions of β values required to reproduce experimental behaviours were compatible with the hypothesis that the presence of other males or other females in the unisex triads increases or decreases exploitation, respectively, possibly owing to social effects on the internal state of the individual^4,37,51,56. This contingency-based framework emphasizes the importance of considering both contingent aspects (such as random group composition of initial parameters and/or initial random reward discovery) and structural factors (for example, one animal becoming a Scrounger favours others to become Workers) in shaping sex-dependent behaviours and social structures.

In conclusion, this work shows how individual and social factors interact to shape behavioural specialization in animal microsocieties. By bridging neurophysiology, behavioural science and computational modelling, our findings offer a nuanced understanding of the dynamical interplay between individual cognition and collective behaviour, with broad implications for the study of social systems across taxa.

Methods

Animals

Male and female C57BL/6J mice (8 to 12 weeks old; Janvier Labs, France) or DAT-iCre mice on a C57BL/6J background were group-housed under standard conditions (12 h:12 h light:dark, ~22 °C, ~50% humidity) and maintained in triads prior to testing (see Supplementary Methods for details). All experiments and procedures were performed in accordance with European Commission directives 219/1990, 220/1990 and 2010/63, and approved by the ESPCI and the ethical committee no. 059 under APAFIS #34335-2021121318085835.

Lone condition and microsociety experiment

Mice were placed either individually or in triads in a 50 × 50 cm large environment and continuously tracked using the Live Mouse Tracker system³⁶, allowing monitoring of identity and behaviour of individual mice over extended periods. Before the experiment, all mice were implanted with a radio frequency identification (RFID) microchip (Biomark APT-12 PIT Tag, Biomark) under the shoulder skin.

Triads (3 males, 3 females, or 2 females and 1 male) stayed for 8 days and 7 nights in the environment, whereas lone mice were housed for 5 days and 4 nights.

The cage was composed of different zones that were freely accessible by all the mice at any moment. The arena contained a lever on one side and a food dispenser/magazine on the opposite side. Each lever press delivered a 20 mg pellet (TestDiet 5TUL/1811142 purified, Bio-Concept) and the lever becomes inactive for 5 s. A nose poke in the magazine leads to a beam break that is a proxy of the consumption of the pellet by the mouse. Outside the 5 s delay after each lever press, all the mice can press at any moment and consume the pellet at any time following its release. Mice were weighed and health-checked daily. No mouse was removed from the task either for weight loss or for aggression.

Social replacement experiment

To test role stability, we performed a reconfiguration experiment in which a male previously identified as a Scrounger during the first week of group housing was rehoused with two task-naive males (Scrounger–naive–naive condition). Mice were kept in the same semi-natural environment as described above, and behaviour was recorded continuously for an additional 7 days. We retained only triads in which a Scrounger was reliably identified at the end of the first week (13 out of 15 triads). To track individual trajectories across time, the retained Scrounger from week 1 was linked to its corresponding behavioural position in week 2, allowing comparison of lever press counts, distances to archetypes, and profile classification before and after reconfiguration.

Live Mouse Tracker system

Behaviour was monitored continuously (24 h daily, 7 days a week) using a dual acquisition pipeline combining time-stamped, animal-identified operant events (lever, magazine/dispenser TTLs (Transistor-Transistor Logic); MedAssociates) with continuous video tracking to quantify locomotion and social/spatial organization (Live Mouse Tracker)³⁶. Individual lever presses, nose pokes and complete sequences were extracted from the database by matching TTLs to mouse identity within predefined lever/magazine zones, and a complete sequence was defined as a lever press followed by the same mouse reaching the magazine within 6 s, otherwise the sequence was not counted (typically if another mouse was already at the magazine waiting for the food) (see Supplementary Methods for details). Gain was defined as retrieving a pellet within 6 s after a conspecific pressed the lever, whereas loss was defined as a pellet retrieved by a conspecific within 6 s after the subject’s own press.

Tube test

We assessed social hierarchy in a subset of mice using a tube test conducted before and after the microsociety task. Three days before testing, mice were habituated to traverse a 30-cm tube (2.5-cm diameter) connecting two cages; habituation was reached after ≥10 successful crossings. On the following day, tube-test trials were performed by placing two mice at opposite ends of the tube; in triads, each mouse faced the two others once per trial. A win/loss was scored when one mouse retreated such that its hind paws touched the cage floor. Five trials were run before the microsociety task, and five additional trials after 7 days of group housing; the apparatus was cleaned between sessions. For Extended Data Fig. 4a, we quantified each mouse’s rank across the five pre-task trials and compared the final pre-task rank (trial 5) to the post-task rank to visualize rank stability and shifts.

Behavioural analyses

Behavioural data were extracted from the database (PyCharm/MySQL) and analysed in MATLAB and R. Lever press counts (#LP) were computed per day for each mouse. To account for inter-cage variability in overall activity, #LP was normalized within each cage by dividing the #LP per mouse by the total #LP in the cage on that day (%LP). This normalization was applied prior to all archetypal analyses involving social triads. The percentage of complete sequences (%CS) was calculated individually for each mouse, as the number of complete sequences divided by the total #LP on the same day. For Extended Data Fig. 3a,b, polar plots were generated from xy coordinates of representative male (n = 42) and female (n = 42) mice, using a coordinate system centred and aligned so that 0° corresponded to the food dispenser. This standardization allowed us to quantify the time each mouse spent oriented towards the food dispenser during particular events—specifically, in response to a conspecific mouse lever press. The arena was partitioned into four zones (water, food, lever-left area and lever-right area), and mouse orientation was quantified at and 1 s after conspecific lever presses.

Archetypal analysis

Archetypal analyses and visualizations were performed in R using the archetypes package (version 2.2-0.1). Archetypal analysis identifies k idealized behavioural profiles (archetypes) spanning the boundaries of variability in a multivariate dataset and yielding individual α-coefficients and distances to each archetype. The number of archetypes (k) was selected using residual sum of squares as a function of k (elbow criterion) and inspecting solution interpretability in lone and social datasets (see Extended Data Figs. 1d and 2c). For Fig. 1, archetypes were computed in a 2D space (#LP, %CS), yielding two archetypes (Achievers and Storers). In Fig. 2, the three-archetype reference space (Workers, Scroungers and Storers) was built from the full dataset (n = 195) using eight features: %LP and %CS (days 5–7) and food pellets gained from/lost to conspecifics. New cohorts (for example, mixed-sex or dopamine-manipulated triads; Fig. 5) were projected into this fixed space to derive α-coefficients, assign behavioural roles, compute distances to archetypes and compare group compositions. The same archetypal framework was applied to simulated triads (e-triads): archetypes were defined from a reference simulation with separated β values (high-β ‘males’ and low-β ‘females’), enabling robust identification of emergent strategies (Fig. 4g). All other simulated conditions (including Extended Data Fig. 7) were projected into this space. Distances to archetypes derived from α-coefficients were related to dopaminergic signals and firing rates using linear models; expected values at each archetype were estimated from the intercept at zero distance with 95% confidence intervals from prediction errors (Fig. 3). Cage-level compositions were compared to sex-specific null distributions generated by random sampling (10,000 iterations) preserving empirical archetype proportions (Fig. 2i). See also Supplementary Methods for details.

In vivo electrophysiology

Mice were deeply anaesthetized with isoflurane (3% induction, 1–2% maintenance) and extracellular single-unit recordings were performed in the VTA using glass micropipette electrodes (6–9 MΩ, 0.5% NaCl). Signals were amplified and digitized at 25 kHz (spike 2) while sampling the central VTA (anterior–posterior (AP) −3.1 to −4.0 mm, medial–lateral (ML) 0.3–0.7 mm, dorsal–ventral (DV) 4.0–4.8 mm), with electrode tracks spaced by ≥0.1 mm. Spontaneously active dopamine neurons were identified using established electrophysiological criteria (Supplementary Methods). Activity and bursting (% spikes within bursts, %SWB) were quantified in 60-s windows shifted every 15 s (Supplementary Methods).

Stereotaxic surgeries

Stereotaxic surgeries were performed in 6- to 8-week-old DAT-Cre mice under isoflurane anaesthesia. For fibre photometry, AAV1-Syn-FLEX-GCaMP7c was injected unilaterally into the VTA (300 nl, 100 nl min⁻¹; AP −3.20 mm, ML ±0.5 mm, DV −4.20 mm), followed 2 to 3 weeks later by unilateral optic fibre implantation above VTA and fixation with dental acrylic; buprenorphine was given post-operatively. For optogenetics (DAT-Cre males), AAV5-DIO-ChR2-EYFP (or EYFP control) was injected bilaterally in VTA (300 nl per side) and an optic fibre was implanted unilaterally above VTA at a 10° angle (AP −3.20 mm, ML ±0.9 mm, DV −3.95 mm). For chemogenetics (DAT-Cre females), AAV5-DIO-hM4Di-mCherry (or mCherry control) was injected bilaterally in VTA (300 nl per side). Mice recovered in a heated cage and were monitored daily; behavioural testing began at least one week after surgery, and injection or implant sites were systematically verified post hoc by immunohistochemistry (Supplementary Methods).

Immunohistochemistry

After euthanasia, brains were extracted and fixed in 4% paraformaldehyde for at least 3 days at 4 °C, and 60-µm-thick sections were taken through the midbrain on a vibratome. Free-floating sections were blocked (PBS, 3% BSA, 0.2% Triton X-100) and incubated overnight at 4 °C with a mouse anti-tyrosine hydroxylase primary antibody (Sigma T1299, 1:500). Sections were then rinsed and incubated for 3 h at room temperature with a Cy3-conjugated goat anti-mouse secondary antibody (Jackson, 1:500), mounted with ProLong Gold plus DAPI, and imaged on a Zeiss epifluorescence microscope (ZEN); grayscale images were acquired and false-coloured in ImageJ for visualization (Supplementary Methods).

Fibre photometry

DAT-Cre mice injected with AAV1-Syn-FLEX-GCaMP7c and implanted in the VTA underwent fibre photometry experiments in the lone or social context. A Doric Lenses fibre photometry system was employed to record fluorescence signals reflecting dopaminergic neuron activity in the VTA. Fluorescence was excited with a 465-nm LED driven in lock-in mode (220.537 Hz) and routed through a Mini Cube (FMC4_AE(405)_E(460–490)_F(500–550)_S) to the implanted fibre via a patch cord and zirconia sleeve; emitted light was detected with a photoreceiver (AC low setting). The received light signal was converted to electrical signals by a photoreceiver using the AC low setting, before being transmitted through another optic patch cord to the Mini Cube via a dedicated fibre optic adaptor. Signals were acquired in Doric Neuroscience Studio at 12 kHz and low-pass filtered at 12 Hz. For the lone condition, the mice were recorded between one and two hours at the beginning of the dark cycle, when they were active, during the first day and the last two days of the experiment. For the social condition, the mice were also recorded at the beginning of the dark cycle, one after the other, between one and two hours each.

Analysis of fibre photometry recordings

Fluorescence signals were first detrended (biexponential fit) to correct for photobleaching, then re-centred by adding back the pre-detrend mean; ΔF/F was computed relative to a baseline fluorescence signal. Dopamine-related activity was quantified using peri-event time histograms time-locked to behaviourally defined TTL events (lever press, nose poke) with 100-ms bins, converted to event-wise z-scored ΔF/F using the 5-s pre-event baseline, and then smoothed by Gaussian convolution (MATLAB, gausswin; 100-bin window). For each event type, peristimulus time histograms were computed over a −10 s to +10 s window, and response magnitude was quantified as the mean z-scored ΔF/F in a post-event window (0 to +1.5 s) relative to a pre-event window (−10 to −5 s). Statistical significance of event-evoked responses was assessed by paired comparisons of pre-event versus post-event window values across mice, using paired t-tests or Wilcoxon signed-rank tests as appropriate (described in figure legends, Supplementary Table 1 and Supplementary Methods). Focused versus unfocused conspecific lever press events in Scroungers (Fig. 3) were scored visually on the basis of orientation to the lever/dispenser and proximity to the dispenser (within 5 cm; Supplementary Methods).

Optogenetic experiments

In male DAT-Cre mice injected with AAV5-Ef1α-DIO-ChR2(H134R)-EYFP or AAV5-Ef1α-DIO-EYFP, optical stimulations were performed with an ultra-high-power LED (470 nm, Prizmatix) coupled to a patch cord (500 μm core, NA = 0.5, Prizmatix) with an output intensity of 5–10 mW. We applied a 20 Hz optogenetic stimulation protocol (5 ms light pulse) for 15 min, delivered twice: 24 h and 1 h before the start of the microsociety task (in the social environment). No significant changes were observed in the behaviours of the mice after the stimulation.

Chemogenetic experiments

In female DAT-Cre mice injected with AAV5-hSyn-DIO-hM4Di-mCherry or AAV5-hSyn-DIO-mCherry, CNO (water soluble, Hellobio) was administered through a water bottle. The CNO solution was introduced 24 h before the microsociety experiment and remained available during the first day (day 1), and was then replaced by normal water. The concentration of CNO was determined on the basis of a dosage of 5 mg kg⁻¹ assuming a daily water consumption of 5 ml per mouse. A 200 µl solution of CNO at a concentration of 10 mg ml⁻¹ was prepared and administered in 100 ml of water in each bottle.

Modelling

Building a behavioural model of e-mouse behaviours in lone and social conditions

The environment of experiments was modelled as six states (Fig. 4a,e, rooms 1–4, and lever and dispenser positions). The number and sex of agents (e-mice) present in the environment was varied, with; (1) one (male or female) e-mouse in the lone experiment; (2) three male or female e-mice in social experiments; or (3) 1 male and 2 females in the mixed-box experiment. State transition occurred at each time step, with probabilities of transitions determined by a softmax based on Q-values of all accessible states from the current state. In the softmax, the inverse temperature parameter β controlled the exploitation–exploration trade-off, with lower β values producing more stochastic exploration and higher β values promoting exploitation of higher-valued transitions. Q learning occurred after each transition from a departure state to an accessible arriving state. After each transition, the value of the selected move was updated from a prediction error combining the obtained reward and the best expected future value from the arrival state. Updates were scaled by a learning rate (α) and a discount factor (γ) controlled the weight of future outcomes. Furthermore, e-mice encountered satiety, which scaled action probabilities, the learning rate and fatigue, which affected both pressing and eating. Satiety and fatigue were used to scale an action pace in the simulation that was consistent with the experimental measures. The lever could not be pressed for 3 time steps after each press (that is, to mimic the 5 s lever unavailability in experiments). Complete modelling information regarding the full model and observables are given in Supplementary Methods.

Reduced model of social interactions

We built a reduced theoretical model (the ‘reduced model’) to assess, within a mathematically tractable framework, the causal mechanisms whereby specialized behaviours emerge under social interactions. To do so, we derived the reduced model from the reinforcement learning one, based on a continuous time version of Q dynamics—that is, ordinary differential equations (ODEs). In the ODE system, learning and behavioural dynamics operated at a slower time scale than that of individual choices in the full model, so that actions (that is, state transitions, lever pressings and eating) were described probabilistically. In this framework, we performed qualitative analysis to determine the number and stability of fixed points of learning and behavioural (state) variables, as a function of parameters (with a focus on β, essential in setting social interactions in the full model). Moreover, to reduce dimensionality for better tractability, we considered a simpler setup in which the environment contains only two positions for e-mice (the lever L and the food dispenser D) and only two e-mice (which allowed us to assess social interactions), and we did not consider fatigue or satiety. The reduced model ODEs could be expressed under a tractable form (Supplementary Methods). The main results of the qualitative analysis of the system are recapitulated in Supplementary Table 2 (Supplementary Information). Full information on the reduced model derivation and analysis are given in Supplementary Methods.

Statistics

A priori power analyses were not used to predetermine sample sizes. Animals were randomized to groups at the time of viral infection or behavioural testing. Statistical analyses were performed using MATLAB and R. Normality was assessed with the Shapiro–Wilk test; normally distributed data were analysed with independent or paired t-tests, and non-normal data with Mann–Whitney or Wilcoxon signed-rank tests. Repeated-measures ANOVA (one-way or two-way, as appropriate) was used for designs with multiple factors, with Bonferroni–Holm post hoc correction. Chi-square (χ²) tests were used to compare proportions or categorical distributions between groups.

For the archetypal analyses, linear regression models (lm function in R) were used to assess the relationship between behavioural or dopaminergic responses and distance to each archetype. Models included both main effects and archetype × distance interactions, allowing estimation of archetype-specific slopes and intercepts. To compare predicted responses at zero distance (intercepts), marginal means were extracted with the emmeans package and pairwise contrasts performed with Tukey correction. Confidence intervals for predicted values (95% confidence interval, shown in figures) were obtained via standard error propagation (predict(…, se.fit = TRUE)), and model significance was assessed using type II ANOVA (car::Anova) and adjusted R². These procedures were applied identically to behavioural, photometry and neurophysiological datasets.

Unless otherwise specified, all statistical tests were two-tailed. In cases in which specific hypotheses were tested regarding the directionality of the effect—such as expected increases or decreases in dopaminergic firing following ChR2 stimulation or hM4Di inhibition—one-tailed tests were used and explicitly reported.

Data presentation (mean ± s.e.m. or mean ± 95% confidence interval), significance thresholds and the statistical tests used are specified in each figure legend; full statistical details (details test statistics, degrees of freedom, exact P values and multiple-comparison procedures) are provided in Supplementary Table 1.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Source data are provided with this paper and a comprehensive table reporting all statistical tests are provided in the Extended Data. In addition, the data underlying the figures, together with analysis scripts, are deposited on Zenodo (https://doi.org/10.5281/zenodo.17121126)⁶¹.

Code availability

Code for the reinforcement learning model and all data analyses is available on Zenodo (https://doi.org/10.5281/zenodo.17121126)⁶¹.

References

Oldfield-Box, H. Social organization of rats in a “social problem” situation. Nature 213, 533–534 (1967).
Article CAS PubMed ADS Google Scholar
Barnard, C. J. & Sibly, R. Producers and scroungers: a general model and its application to captive flocks of house sparrows. Anim. Behav. 29, 543–550 (1981).
Article Google Scholar
Beshers, S. N. & Fewell, J. H. Models of division of labor in social insects. Entomology 46, 413–440 (2001).
Article CAS Google Scholar
Bergmüller, R. & Taborsky, M. Animal personality due to social niche specialisation. Trends Ecol. Evol. 25, 504–511 (2010).
Article PubMed Google Scholar
Gartland, L. A., Firth, J. A., Laskowski, K. L., Jeanson, R. & Ioannou, C. C. Sociability as a personality trait in animals: methods, causes and consequences. Biol. Rev. 97, 802–816 (2022).
Article PubMed Google Scholar
Sumpter, D. J. T. The principles of collective animal behaviour. Philos. Trans. R. Soc. B 361, 5–22 (2006).
Article CAS Google Scholar
Couzin, I. D. & Krause, J. Self-organization and collective behavior in vertebrates. Adv. Study Behav. 32, 1–75 (2003).
Article Google Scholar
Ioannou, C. C. & Laskowski, K. L. Conformity and differentiation are two sides of the same coin. Trends Ecol. Evol. 38, 545–553 (2023).
Article PubMed Google Scholar
Cooper, G. A. & West, S. A. Division of labour and the evolution of extreme specialization. Nat. Ecol. Evol. 2, 1161–1167 (2018).
Article PubMed Google Scholar
Webster, M. M. & Ward, A. J. W. Personality and social context. Biol. Rev. 86, 759–773 (2011).
Article PubMed Google Scholar
Loftus, J. C., Perez, A. A. & Sih, A. Task syndromes: linking personality and task allocation in social animal groups. Behav. Ecol. 32, araa083 (2021).
Google Scholar
Faure, P., Fayad, S. L., Solié, C. & Reynolds, L. M. Social determinants of inter-individual variability and vulnerability: the role of dopamine. Front. Behav. Neurosci. 16, 836343 (2022).
Article CAS PubMed PubMed Central Google Scholar
Giraldeau, L.-A. & Dubois, F. Social foraging and the study of exploitative behavior. Adv. Stud. Behav. 38, 59–104 (2008).
Article Google Scholar
Ahn, S. et al. A role of anterior cingulate cortex in the emergence of worker–parasite relationship. Proc. Natl Acad. Sci. USA 118, e2111145118 (2021).
Article CAS PubMed PubMed Central Google Scholar
Colin, C. & Desor, D. Differenciations comportementales dans des groupes de rats soumis a une difficulte d’acces a la nourriture. Behav. Process. 13, 85–100 (1986).
Article Google Scholar
Smith, J. M. Evolution and the Theory of Games (Cambridge Univ. Press, 1982).
Fawcett, T. W., Hamblin, S. & Giraldeau, L.-A. Exposing the behavioral gambit: the evolution of learning and decision rules. Behav. Ecol. 24, 2–11 (2013).
Article Google Scholar
Laland, K. N., Sterelny, K., Odling-Smee, J., Hoppitt, W. & Uller, T. Cause and effect in biology revisited: is Mayr’s proximate–ultimate dichotomy still useful? Science 334, 1512–1516 (2011).
Article CAS PubMed ADS Google Scholar
Dridi, S. & Lehmann, L. On learning dynamics underlying the evolution of learning rules. Theor. Popul. Biol. 91, 20–36 (2014).
Article PubMed Google Scholar
Schultz, W., Dayan, P. & Montague, P. R. A neural substrate of prediction and reward. Science 275, 1593–1599 (1997).
Article CAS PubMed Google Scholar
Bayer, H. M. & Glimcher, P. W. Midbrain dopamine neurons encode a quantitative reward prediction error signal. Neuron 47, 129–141 (2005).
Article CAS PubMed PubMed Central Google Scholar
Eshel, N., Tian, J., Bukwich, M. & Uchida, N. Dopamine neurons share common response function for reward prediction error. Nat. Neurosci. 19, 479–486 (2016).
Article CAS PubMed PubMed Central Google Scholar
Hamid, A. A. et al. Mesolimbic dopamine signals the value of work. Nat. Neurosci. 19, 117–126 (2016).
Article CAS PubMed Google Scholar
Solié, C., Girard, B., Righetti, B., Tapparel, M. & Bellone, C. VTA dopamine neuron activity encodes social interaction and promotes reinforcement learning through social prediction error. Nat. Neurosci. 25, 86–97 (2022).
Article PubMed Google Scholar
Gunaydin, L. A. et al. Natural neural projection dynamics underlying social behavior. Cell 157, 1535–1551 (2014).
Article CAS PubMed PubMed Central Google Scholar
Wilson, R. C., Bonawitz, E., Costa, V. D. & Ebitz, R. B. Balancing exploration and exploitation with information and randomization. Curr. Opin. Behav. Sci. 38, 49–56 (2021).
Article PubMed Google Scholar
Cohen, J. D., McClure, S. M. & Yu, A. J. Should I stay or should I go? How the human brain manages the trade-off between exploitation and exploration. Philos. Trans. R. Soc. B 362, 933–942 (2007).
Article Google Scholar
Daw, N. D., O’Doherty, J. P., Dayan, P., Seymour, B. & Dolan, R. J. Cortical substrates for exploratory decisions in humans. Nature 441, 876–879 (2006).
Article CAS PubMed PubMed Central ADS Google Scholar
Kakade, S. & Dayan, P. Dopamine: generalization and bonuses. Neural Netw. 15, 549–559 (2002).
Article PubMed Google Scholar
Beeler, J. A., Daw, N., Frazier, C. R. M. & Zhuang, X. Tonic dopamine modulates exploitation of reward learning. Front. Behav. Neurosci. 4, 170 (2010).
Article PubMed PubMed Central Google Scholar
Cinotti, F. et al. Dopamine blockade impairs the exploration-exploitation trade-off in rats. Sci. Rep. 9, 6770 (2019).
Article PubMed PubMed Central ADS Google Scholar
Dongelmans, M. et al. Chronic nicotine increases midbrain dopamine neuron activity and biases individual strategies towards reduced exploration in mice. Nat. Commun. 12, 6945 (2021).
Article CAS PubMed PubMed Central ADS Google Scholar
Faure, P. in Handbook of Behavioural Neuroscience Vol. 32 (eds Cragg, S. J. & Walton, M. E.) 357–365 (2025).
Chen, C. S., Mueller, D., Knep, E., Ebitz, R. B. & Grissom, N. M. Dopamine and norepinephrine differentially mediate the exploration–exploitation tradeoff. J. Neurosci. 44, e1194232024 (2024).
Article CAS PubMed PubMed Central Google Scholar
Humphries, M. D., Khamassi, M. & Gurney, K. Dopaminergic control of the exploration–exploitation trade-off via the basal ganglia. Front. Neurosci. 6, 9 (2012).
Article PubMed PubMed Central Google Scholar
Chaumont, F., de et al. Real-time analysis of the behaviour of groups of mice via a depth-sensing camera and machine learning. Nat. Biomed. Eng. 7, 462–942 (2019).
Google Scholar
Torquet, N. et al. Social interactions impact on the dopaminergic system and drive individuality. Nat. Commun. 9, 3081 (2018).
Article CAS PubMed PubMed Central ADS Google Scholar
Fayad, S. L. et al. Individualistic reward-seeking strategies that predict response to nicotine emerge among isogenic male mice living in a micro-society. PLoS Biol. 22, e3002850 (2024).
Article CAS PubMed PubMed Central Google Scholar
Weissbrod, A. et al. Automated long-term tracking and social behavioural phenotyping of animal colonies within a semi-natural environment. Nat. Commun. 4, 2018 (2013).
Article PubMed ADS Google Scholar
Forkosh, O. et al. Identity domains capture individual differences from across the behavioral repertoire. Nat. Neurosci. 2, e00759–6 (2019).
Google Scholar
Hart, Y. et al. Inferring biological tasks using Pareto analysis of high-dimensional data. Nat. Methods 12, 233–235 (2015).
Article CAS PubMed Google Scholar
Strogatz, S. H. Nonlinear Dynamics and Chaos: With Applications to Physics, Biology, Chemistry, and Engineering (Addison-Wesley, 1994).
Bariselli, S. et al. Role of VTA dopamine neurons and neuroligin 3 in sociability traits related to nonfamiliar conspecific interaction. Nat. Commun. 9, 3173 (2018).
Article PubMed PubMed Central ADS Google Scholar
Chaudhury, D. et al. Rapid regulation of depression-related behaviours by control of midbrain dopamine neurons. Nature 493, 532–536 (2013).
Article CAS PubMed ADS Google Scholar
Laland, K. N. & O’Brien, M. J. Cultural niche construction: an introduction. Biol. Theory 6, 191–202 (2011).
Article Google Scholar
Wanat, M. J., Willuhn, I., Clark, J. J. & Phillips, P. E. M. Phasic dopamine release in appetitive behaviors and drug addiction. Curr. Drug Abuse Rev. 2, 195–213 (2009).
Article CAS PubMed PubMed Central Google Scholar
Zilkha, N. et al. Sex-dependent control of pheromones on social organization within groups of wild house mice. Curr. Biol. 33, 1407–1420.e4 (2023).
Article CAS PubMed PubMed Central Google Scholar
Sofer, Y. et al. Sexually dimorphic oxytocin circuits drive intragroup social conflict and aggression in wild house mice. Nat. Neurosci. 27, 1565–1573 (2024).
Article CAS PubMed Google Scholar
Zipple, M. N. et al. Competitive social feedback amplifies the role of early life contingency in male mice. Science 387, 81–85 (2025).
Article CAS PubMed PubMed Central ADS Google Scholar
Vogt, C. C. et al. Female behavior drives the formation of distinct social structures in C57BL/6J versus wild-derived outbred mice in field enclosures. BMC Biol. 22, 35 (2024).
Article PubMed PubMed Central Google Scholar
Wei, A. et al. Sexually dimorphic dopaminergic circuits determine sex preference. Science 387, eadq7001 (2025).
Article CAS PubMed Google Scholar
Chen, C. S., Knep, E., Han, A., Becker, R. & Grissom, N. M. Sex differences in learning from exploration. eLife 10, e69748 (2021).
Article PubMed PubMed Central Google Scholar
Davis, J. F., Krause, E. G., Melhorn, S. J., Sakai, R. R. & Benoit, S. C. Dominant rats are natural risk takers and display increased motivation for food reward. Neuroscience 162, 23–30 (2009).
Article CAS PubMed Google Scholar
Morgan, D. et al. Social dominance in monkeys: dopamine D₂ receptors and cocaine self-administration. Nat. Neurosci. 5, 169–174 (2002).
Article CAS PubMed Google Scholar
Kooij, van der, M. A. et al. Diazepam actions in the VTA enhance social dominance and mitochondrial function in the nucleus accumbens by activation of dopamine D₁ receptors. Mol. Psychiatry 23, 569–578 (2018).
Article PubMed Google Scholar
Battivelli, D. et al. Dopamine neuron activity and stress signaling as links between social hierarchy and psychopathology vulnerability. Biol. Psych. 95, 774–784 (2024).
Article CAS Google Scholar
Lozano-Montes, L. et al. Latency to reward predicts social dominance in rats: a causal role for the dopaminergic mesolimbic system. Front. Behav. Neurosci. 13, 69 (2019).
Article CAS PubMed PubMed Central Google Scholar
Freund, J. et al. Emergence of individuality in genetically identical mice. Science 340, 756–759 (2013).
Article CAS PubMed ADS Google Scholar
Honegger, K. & de Bivort, B. Stochasticity, individuality and behavior. Curr. Biol. 28, R8–R12 (2018).
Article CAS PubMed Google Scholar
Wolf, M., Bierbach, D. & Laskowski, K. L. Behavioural individuality in clonal fish arises despite near-identical rearing conditions. Nat. Commun. 8, 15361 (2017).
Article PubMed PubMed Central ADS Google Scholar
Solié, C. et al. Data and code for 'Dopaminergic mechanisms of dynamical social specialization'. Zenodo https://doi.org/10.5281/zenodo.17121126 (2026).

Download references

Acknowledgements

We thank O. de Oliveira, N. Karakaplan-Dherbe and E. Tubeuf at ESPCI animal facilities for support, and G. Ellis-Davies for feedback on a draft of the manuscript. C.S. thanks La Fondation des Treilles for supporting this work. We thank the Bettencourt Schueller Foundation for early support of this microsociety programme. This work has received support under the Major Research Program of PSL Research University ‘PSL-Neuro’, implemented by ANR (ANR-10-IDEX-0001). Our team is affiliated to DIM C-BRAINS, funded by the Conseil Régional d’Ile-de-France. Funding: Centre National de la Recherche Scientifique (CNRS UMR 8249), ESPCI Paris - PSL, Fondation pour la Recherche Médicale (FRM EQU201903007961 and EQU202503019989 to P.F., FRM SPF202005011922 to C.S.), French National Cancer Institute Grant (SPA-CPA-DOC23 to A.N., SPAV1-21-002 and SPAV1-23-005 to P.F.), French state funds managed by Agence Nationale de la Recherche (ANR-23-CE37-0017 VarSeek to P.F. and B.D. and ANR-25-CE37-6088 Social-VIP to P.F.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Author information

Authors and Affiliations

Brain Plasticity Laboratory, CNRS UMR 8249, ESPCI Paris, PSL Research University, Paris, France
C. Solié, A. Nicolson, R. Justo, Y. Layadi, B. Morin, C. Batifol, L. M. Reynolds, T. Le Borgne, S. L. Fayad, A. Gulmez, Y. Rodriguez Quevedo, J. Allegret-Vautrot, G. Centene Guglielmi, S. Didienne, N. Debray, J.-P. Hardelin, A. Mourot, C. Viollet, F. Marti & Ph. Faure
Human Genetics and Cognitive Functions, CNRS UMR UMR3571, Pasteur Institute, Paris, France
F. de Chaumont
Institut des Systèmes Intelligents et de Robotique (ISIR), Sorbonne Université, CNRS, Paris, France
B. Girard & B. Delord
Institut de Génomique Fonctionnelle, University of Montpellier, UMR 5203 CNRS, U 1191 Inserm, Montpellier, France
J. Naudé

Authors

C. Solié
View author publications
Search author on:PubMed Google Scholar
A. Nicolson
View author publications
Search author on:PubMed Google Scholar
R. Justo
View author publications
Search author on:PubMed Google Scholar
Y. Layadi
View author publications
Search author on:PubMed Google Scholar
B. Morin
View author publications
Search author on:PubMed Google Scholar
C. Batifol
View author publications
Search author on:PubMed Google Scholar
L. M. Reynolds
View author publications
Search author on:PubMed Google Scholar
T. Le Borgne
View author publications
Search author on:PubMed Google Scholar
S. L. Fayad
View author publications
Search author on:PubMed Google Scholar
A. Gulmez
View author publications
Search author on:PubMed Google Scholar
Y. Rodriguez Quevedo
View author publications
Search author on:PubMed Google Scholar
J. Allegret-Vautrot
View author publications
Search author on:PubMed Google Scholar
G. Centene Guglielmi
View author publications
Search author on:PubMed Google Scholar
F. de Chaumont
View author publications
Search author on:PubMed Google Scholar
S. Didienne
View author publications
Search author on:PubMed Google Scholar
N. Debray
View author publications
Search author on:PubMed Google Scholar
J.-P. Hardelin
View author publications
Search author on:PubMed Google Scholar
B. Girard
View author publications
Search author on:PubMed Google Scholar
A. Mourot
View author publications
Search author on:PubMed Google Scholar
J. Naudé
View author publications
Search author on:PubMed Google Scholar
C. Viollet
View author publications
Search author on:PubMed Google Scholar
F. Marti
View author publications
Search author on:PubMed Google Scholar
B. Delord
View author publications
Search author on:PubMed Google Scholar
Ph. Faure
View author publications
Search author on:PubMed Google Scholar

Contributions

Conceptualization: C.S. and P.F. Experimental methodology: C.S., A.N., R.J., F.d.C., C.V., A.M., S.D., N.D. and P.F. Investigation: C.S., A.N., B.M., Y.R.Q., C.B., L.M.R., T.L.B., S.L.F., A.G. and J.A.-V. Experimental data analysis: C.S., B.M., G.C.G., F.M., R.J. and P.F. Numerical modelling: B.D., J.N., B.G., Y.L. and P.F. Reduced model: B.D. Funding acquisition: P.F., B.D. and C.S. Writing—original draft: C.S., B.D. and P.F. Writing—review and editing: C.S., A.N., Y.L., C.V., J.-P.H., A.M., J.N. and P.F.

Corresponding authors

Correspondence to B. Delord or Ph. Faure.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature thanks Andrew Tapper, who co-reviewed with Susanna Molas, and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer review reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Lone condition.

(a) (Left) Number of lever-presses (#LP) across days in the lone condition for each animal. (Middle) Time course of the number of beam breaks (BB) for each animal. (Right) Time course of the percentage of complete sequences (%CS) for each animal. (b) Mean Z-score (mean ± s.e.m) VTA DA activity aligned to LP for incomplete sequences when the animal is moving after the LP (Top) or staying immobile (Bottom) (Wilcoxon tests). (c) (Top) Mean Z-score VTA DA activity at LP between complete sequences and moving incomplete sequences (Mann–Whitney/Wilcoxon rank-sum test). (Bottom) Correlation between the mean Z-score VTA DA activity at LP and the time to go to the food magazine for each animal (Pearson correlation). (d) Archetypal analysis of mouse behavior in the lone condition (n = 62 mice) based on an 8-dimensional feature space: number of lever presses (#LP) and percentage of complete sequences (%CS) from day 1 to day 4 of the microsociety experiment. Residual sum of squares is shown for a number of archetypes (N_a) ranging from 1 to 6. (e) Archetype solutions for N_a = 2, 3 and 4; the color code indicates Storer, Achiever and undefined profiles. Solutions with N_a > 2 do not reveal clearly distinct new profiles. (f) Cumulative distribution of the number of LPs (#LP, Left) and of the percentage of complete sequences (%CS, Right) during the last two days of the experiment, comparing Achievers and Storers (two-sample Kolmogorov–Smirnov tests). (g) In vivo electrophysiology in anesthetized mice. (Left) Spontaneous firing rate in hertz (Hz) as a function of the percentage of spikes within burst (%SWB) for each recorded neuron, grouped by behavioral profile. (Right) Mean %SWB comparing Achiever and Storer male mice (Wilcoxon rank-sum test). Data are presented as mean ± s.e.m. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Extended Data Fig. 2 Sex differences and archetypal structure in the social task.

(a) Cumulative probability distributions of the number of lever-presses (#LP) (Left) and the percentage of complete sequences (%CS) (Right) during the last two days of the microsociety experiment, comparing male and female mice (two-sample Kolmogorov–Smirnov tests). (b) Distribution of latencies between lever press (LP) and beam break (BB) (Left) and distribution of the number of consecutive LP before a BB (Right) in male versus female mice (two-sample Kolmogorov–Smirnov tests); N_LP indicates the total number of lever-press events contributing to the distributions (males and females shown separately), and n indicates the number of mice. (c) Archetypal analysis of mouse behavior in the social task (n = 195 mice) using an 8-dimensional feature space: percentage of LPs (%LP) and %CS from day 5 to day 7, and food pellet gain/loss; residual sum of squares is shown for the number of archetypes (N_a) ranging from 1 to 8. (d) Archetypal solutions for N_a = 2, 3, 4 and 5; the color code indicates Worker, Storer, Scrounger, and undefined profiles. Solutions with N_a > 3 do not yield clearly distinct additional profiles. (e) Time course of #LP across days by archetype (mean ± s.e.m; two-way repeated-measures ANOVA). Data are presented as mean ± s.e.m. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Extended Data Fig. 3 Behavioral characterization.

(a-b) Animal position at the time of a conspecific LP during the last day of the microsociety experiment. The cage was divided into four zones: water (1), feeder (2) and left/right lever zones (3/4). (a) Circular diagrams show the angular distribution of mice relative to the feeder zone at the time of a conspecific’s LP (Left) and 1 second after (Right), by behavioral profile (Worker, Scrounger, Storer). (b) Percentage of time spent in each zone (zones 1 to 4) 5 s before the conspecific’s LP, 5 s after, and during random time points, shown by behavioral profile (Fisher’s exact tests across zones/time windows, with Holm correction for multiple comparisons). (c) (Top-left) Time course of the number of LPs (#LP) during the first day (Day 1) by behavioral profiles (two-way repeated-measures ANOVA). (Top right) Mean #LPs per hour during the first day, across profiles (one-way ANOVA with Bonferroni–Holm correction for pairwise comparisons). (Bottom-left) Time course of the number of complete sequences (#CS) on day 1 by profile (two-way repeated-measures ANOVA). (Bottom-right) Mean #CS per hour on day 1 across profiles (one-way ANOVA with Bonferroni–Holm correction) (d) Correlations of individual #LP between the last day and day 2 (Left) or day 6 (Right) (linear regression, lm, shaded areas indicate 95% confidence intervals of the regression fit). (e) Number of food pellets stolen from a mouse (Loss) or by a mouse (Gain) after a LP on day 1 by profile (two-way repeated-measures ANOVA). Data are presented as mean ± s.e.m. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided unless otherwise stated. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Extended Data Fig. 4 Hierarchy analysis.

(a) (Left) Individual social dominance ranks (R1-3) across the tube-test trials performed daily for five days (D1-5), before (Pre) and after (Post) the microsociety experiment. (Right) Comparison of ranks from the last pre-experiment trial (D5 pre) to those obtained after the microsociety experiment (D5 post) (see Methods). (b) Distance to archetypes as a function of social dominance rank in males and females, analyzed separately in the relevant sex/archetype groups (one-way ANOVA, Pre and Post data analyzed separately; see Supplementary Table 1). Data are presented as mean ± s.e.m. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Extended Data Fig. 5 Additional analyses of VTA DA signals and electrophysiology.

(a-b) Mean Z-scored (mean ± s.e.m) VTA DA activity on day 1 aligned to food consumption (a) and own lever presses (b) by sex. (c) Per-mouse trial averages of Z-scored VTA DA activity aligned to food consumption, own LPs, and conspecific LPs, shown separately for Workers (Top, black) and Scroungers (Bottom, red). Responses were quantified as post- vs pre-event windows and compared using paired t-tests or Wilcoxon signed-rank test (see Methods) (d-g) Peak activity at own LPs (d, f) and conspecific LPs (e, g) as a function of distance to each archetype with linear regression (lm) fits. (h) In vivo electrophysiological recording in anesthetized mice post-microsociety experiment; representative traces of DA neuron firing in male and female mice. (i) Percentage of spikes within bursts (%SWB) for each recorded VTA DA neuron in male and female mice, either naive or after 7 days of microsociety experiment (Post), shown per neuron (Left) and per mouse (Right) (Wilcoxon rank-sum tests; p values Holm-adjusted for multiple comparisons). (j) Estimated %SWB at zero distance from each archetype (linear model with Archetype × Distance interaction, with post hoc contrasts using emmeans and Tukey adjustment; mean ± 95% CI). Experimental data are presented as mean ± s.e.m.; model-based estimates are reported as mean ± 95% CI. Shaded areas indicate 95% CI of the regression fit. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided unless otherwise stated. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Extended Data Fig. 6 Parameter sensitivity of the Q-learning model.

(a) Q-learning simulation of one e-mouse with « low » β (β = 0.5; compare to Fig. 4a, where β is high). (b-c) Dependence of the LP count (#LP) (b) and of the percentage of complete sequences (%CS) (c) on the learning rate (α), inverse temperature (β) and temporal discount factor (γ) in the full model and lone condition. In each panel, the considered parameter is varied, with other parameters set at their standard value (see Methods). (d) Same as (a), with three e-mice with low β (β = 0.5; compare to Fig. 4e, where β is high). (e-f) Dependence of #LP (e) and of %CS (f) on parameters, as in (b-c), but in the social condition, with three e-mice having the same parameter values.

Extended Data Fig. 7 Behavioral archetypes and β-value distributions in simulated social triads.

(a) Distribution (mean and individual data) of β-values across the three behavioral archetypes—Worker, Scrounger, and Storer—in simulated triads composed of either three high-β males (N = 1000 triads) or three low-β females (N = 1000 triads). (b) Distribution of the Worker-to-Scrounger β-ratio across male-only triads. The distribution is centered around 1 (mean = 1.007), indicating only a minimal but statistically significant bias (p < 0.001). (c) Q-learning simulation of three e-mice with two low-β individuals (females, β ∈ [0, 0.25]) and one high-β individual (male, β ∈ [2.25, 2.5]). (d) Proportions of Worker, Scrounger, and Storer archetypes as a function of sex across 1000 triads, revealing sex-dependent specialization patterns. (e) Distribution (mean and individual data) of β values per archetype (Worker, Scrounger, Storer) in mixed triads, highlighting β differences in value-based behavior. (f) Distribution of archetypal compositions per triads, grouped by β value (low vs. high).

Extended Data Fig. 8 Dynamical modeling.

(a-c) Bifurcation diagrams corresponding to Fig. 4h, shown as a function of the residence probability at the food dispenser (Dispenser) (a), lever-press (LP) (b) and complete sequence (CS) probabilities (c) in the reduced model. (d-f) Model dynamics corresponding to Fig. 4i-k in the plane defined by the residence probability at Dispenser (Left), LP (center) and CS (Right), for females (d), males (e) and in the mixed-sex box condition (f). (g-j) Same as (a-c) for the reduced model when no linearization of the softmax function is done. The bifurcation diagrams are qualitatively and quantitatively similar to those obtained with linearization (compare to Fig. 4h and 4i-k).

Extended Data Fig. 9 Experimental manipulation supplementary information.

(a-g) Experimental testing of model predictions: Mixed-sex triads and behavioral flexibility. (a) Mixed-sex triads composed of 1 male and 2 females. (b) Mean number LPs (#LP) and %of complete sequence (%CS) for males (green, n = 12) and females (blue, n = 24) (unpaired t-tests for #LP and %CS). (c) Distance to the Storer (Left) and Scrounger (Right) archetypes by sex (Wilcoxon rank-sum tests). (d) Triads composition. (e) Reconfiguration experiment: a male previously identified as a Scrounger (week 1) was rehoused with two naive males for a second week (week 2). (f) Number of LPs for the identified Scrounger across week 1 and week 2 (paired t-test). (g) Archetypal analysis of all individuals from both weeks; retained animals from week 1 are linked across time by arrows. (h-o) Bidirectional manipulation of VTA DA activity modulates behavioral specialization. (h) Experimental male mouse triads: (i) three YFP mice (YFP,control), (ii) one ChR2-stimulated mouse with two YFP mice (1ChR2), and (iii) three ChR2-stimulated mice (3ChR2). (i) Example image of ChR2 expression and fiber placement. (j) Distance to the Storer archetype across groups (Kruskal–Wallis test followed by one-sided post hoc Wilcoxon rank-sum tests with Holm adjustment). (k) Triad composition by condition. (l) Experimental female mouse triads: (i) three mCherry mice (mCh, control), (ii) one hM4Di-inhibited mouse with two mCh mice (1hM4Di), and (iii) three hM4Di-inhibited mice (3hM4Di). (m) Example image of hM4Di expression. (n) Distance to the Scrounger archetype across groups (Kruskal–Wallis test followed by one-sided post hoc Wilcoxon rank-sum tests with Holm adjustment). (o) Triad composition by condition. Data are presented as mean ± s.e.m. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided unless otherwise stated. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Extended Data Fig. 10 Optogenetic stimulation and chemogenetic inhibition of VTA DA neurons alters dopaminergic activity.

(a) Experimental timeline: DAT-Cre mice were transfected with either YFP or ChR2 in VTA DA neurons and implanted with optic fibers. The stimulation protocol was applied 24 h and again 2 h before in vivo electrophysiological recordings under anesthesia. (b) In vivo electrophysiology in anesthetized mice. For each recorded VTA DA neuron, spontaneous frequency in hertz (Hz) is plotted against the percentage of spikes within burst (%SWB) in YFP- and ChR2- expressing mice after the stimulation protocol (Wilcoxon rank-sum test). (c) Mean firing rate (mean ± s.e.m., with individual data) in YFP and ChR2 mice after the stimulation protocol (Wilcoxon rank-sum test). (d) Mean %SWB (mean ± s.e.m., with individual data) in YFP and ChR2 mice after the stimulation protocol. (e) Experimental timeline: DAT-Cre female mice were transfected with either mCherry or hM4Di in VTA DA neurons. CNO was provided in the drinking water 24 h before recordings and removed 2 h before anesthesia. (f) In vivo electrophysiology in anesthetized mice: spontaneous firing rate in hertz (Hz) against %SWB for each recorded neuron in mCherry- and hM4Di- expressing mice after the CNO protocol. (g) Mean firing rate (mean ± s.e.m., with individual data) in mCherry vs hM4Di mice (Wilcoxon rank-sum test). (h) Mean %SWB (mean ± s.e.m., with individual data) in mCherry vs hM4Di mice after the CNO protocol (Wilcoxon rank-sum test). Data are presented as mean ± s.e.m. Details (test statistics, df, exact p values) are reported in Supplementary Table 1. All tests were two-sided. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Supplementary information

Supplementary Information (download PDF )

Supplementary Methods and Modelling, Supplementary Tables 1 and 2 and Supplementary References.

Reporting Summary (download PDF )

Peer Review File (download PDF )

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Cite this article

Solié, C., Nicolson, A., Justo, R. et al. Dopaminergic mechanisms of dynamical social specialization. Nature (2026). https://doi.org/10.1038/s41586-026-10301-4

Download citation

Received: 19 February 2025
Accepted: 19 February 2026
Published: 01 April 2026
Version of record: 01 April 2026
DOI: https://doi.org/10.1038/s41586-026-10301-4