Neuronal Correlates of Motor Control and Economic Decision-Making During Social Interactions in Non-Human Primates

A cornerstone of human evolution lies in social behavior and cooperation, proposed to be the third pillar of evolution alongside mutation and natural selection (Nowak, 2006). Among the various forms of social behavior, motor interaction between individuals — referred to as Joint Action — represents the dynamic core of cooperation. Traditionally, motor control studies have focused primarily on solipsistic movements, revealing the remarkable complexity of our motor systems. Beyond the standard sensorimotor processes necessarily involved during planning and execution of our own actions, it has been hypothesized that the brain employs internal models — neural representations that simulate aspects of the body and the environment — which allow for the prediction of action consequences, as well as for learning, planning, and controlling movements. These internal models are essential for managing the inherent complexity of motor behavior (D. M. Wolpert et al., 1995; Miall & Wolpert, 1996). How these models apply to motor interactions between individuals remains a topic of debate. In cooperative behavior — particularly during decision-making in social contexts — it is essential to predict a partner’s intentions and movements, which are inherently unpredictable. In understanding others’ actions, the discovery of the mirror neuron system represented a ground-breaking insight into how the brain maps observed behavior onto internal motor processes (Gallese et al., 1996, 2004). Building on this, it has been hypothesized that motor control theories could be extended from solipsistic to social context (Wolpert et al., 2003). However, while these works have served as a foundational pillar in the study of action observation, they do not fully account for motor interactions between individuals, where simultaneous movements of two or more agents occur in interactive settings. Indeed, recent hypothesis suggest that sharing actions require a more sophisticated model of motor control — one that involves a shared representation integrating both one’s own and the partner’s actions (della Gatta et al., 2017; Sacheli et al., 2018; Kourtis et al., 2019; Lacal et al., 2022; Marschner et al., 2024), and relies on a predictive motor representations of collective goals (Pezzulo et al., 2025). It was previously showed that non-human primates are able to coordinate each other in a joint fashion, by dynamically adjusting their motor behavior to foster synchronization (Visco-Comandini et al., 2015). A specialized class of neurons encoding joint actions in dorsal premotor cortex have also been discovered (Ferrari-Toniolo et al., 2019). Furthermore, empirical evidence has been provided that, like humans, also macaques exploit on a shared representation (Lacal et al., 2022), suggesting the formation of dyadic motor plan, thus suggesting proactive adjustments are required to successfully perform Joint Actions. Within this framework, the first study of this thesis addressed whether and how cooperative actions are represented at the neural population level and whether such proactive mechanisms are reflected in neural activity. The aim of Study 1 was to investigate how Local Field Potentials (LFPs) recorded from dorsal Premotor Cortex (area F2) encode the spatiotemporal components of Joint Action, potentially providing the first neuronal population-level evidence for a dyadic motor representation. To this purpose, we simultaneously recorded LFPs from the PMd of two monkeys engaged in a pre-cued “center-out” joint action task, that was first adopted in the lab of Prof. Battaglia Mayer to study the neural underpinnings of joint performance in non-human primate models (Visco-Comandini et al., 2015; Ferrari-Toniolo et al., 2019). Our findings provided new insights into the population-level neural mechanisms associated with Joint Action coding in the dorsal Premotor Cortex, as reflected in low-frequency LFPs. Surprisingly, these low-frequency LFPs were found to predict the quality of dyadic interaction, and neuronal synchronization in the LFP components mirrored their timing coordination during motor planning. Decoding of LFPs further revealed modulations in premotor activity associated with interpersonal coordination and its effectiveness. Moreover, a dynamic sequential coding pattern emerged, whereby the action context was represented early during planning, and coordination features later, near movement onset, suggesting that premotor activity enables a predictive–corrective process, transforming preparatory plans into execution-phase control for effective coordination between agents. These findings suggest monkeys leveraged on shared motor representations, and provide a neuronal substrate for the predictions necessary for joint temporal and spatial coordination. In a second study we investigated how the cost of acting together influences decision-making processes during cooperative actions. While joint actions enable individuals to achieve goals that would be unattainable alone, they involve a tangible cost stemming from the effort required to coordinate with others. These coordination costs might be also associated with a decreased probability of success due to the increased complexity and motor demands involved. Indeed, when deciding to act together a cognitive effort is required as it is necessary to anticipate the partner’s intentions to successfully perform joint decisions. Added to this is the difficulty of coordinating together within temporal and spatial constraints. Indeed, despite the implementation of the strategy to reduce individual motor variability to face with the partner’s motor unpredictability, monkeys showed a lower probability to successfully perform joint actions compared to individual actions (Visco-Comandini et al., 2015; Lacal et al., 2022). Therefore, when monkeys are faced with the choice of whether to act alone or with others, decisions might impinge upon a careful cost-benefit evaluation. Thus, in Study 2 we asked whether monkeys consider the cost-benefit of joint action while free to choose whether acting alone or together, and whether they are able to update their decision processes to the partner’s intentions. The primary goal of this work was to investigate the neuronal basis of economic decision-making during such social context, by focusing on the dorsolateral prefrontal cortex (dlPFC; area 9), a region demonstrated to be implicated in social and economic decision-making (Padoa-Schioppa et al., 2014; Falcone et al., 2016; Franch et al., 2024). In Study 2 we first explore the behavioral strategies that monkeys employ to make joint decisions, using motor control as a window into the underlying decision-making processes (Lacal, 2020; Quarta et al., in prep.). Specifically, we asked whether monkeys take into account the cost of cooperation, and we aimed at estimate this subjective cost by varying the level of reward payoff associated with individual versus joint actions. To this purpose, we adopted a novel joint action choice task in which two monkeys were free to choose between acting alone (SOLO) or jointly (JA), based on the varying quantity of reward associated to each offer. Two targets were presented, each indicating both the reward magnitude and the action type ("SOLO" or "JA") required to obtain the desired reward. Using an isometric joystick, each monkey expressed its choice by guiding a cursor towards the chosen target to gain the reward associated with. Behaviorally, monkeys demonstrated the ability to evaluate the trade-offs between coordination costs and reward benefits. We inferred their subjective cost of cooperation and found that the two monkeys shared a similar valuation. Over time, they showed an increased tendency to choose the joint action, which was accompanied by a reduced subjective cost of cooperation. These findings suggest that monkeys were capable of representing their partner’s choices, and learned to make better joint decisions to maximize overall reward. At neuronal level, preliminary analyses suggest that the dlPFC’s neurons may play a critical role in economic decision-making within social contexts, and may encode the complex cost-benefit computations required when deciding whether to act alone or together. Interestingly, we found neurons preferentially modulated by the decision to act together (Joint Action Selection cells) with respect to individual actions. We also observed cells encoding the future choice (and thus predicting it) of the partner (Other-Choice cells). We also found neurons whose activity signals the type of actions (SOLO or JA), often enhancing the firing rates by the joint action during motor planning and execution. These findings offer, for the first time, evidence of neural modulations during joint behavior both during the decision-making phases, and action execution, suggesting a critical role of dlPFC in “good-to-action” transformation when acting in social context.