Dissociation between observational learning and social comparison neural signals

Aims. When making decisions in the presence of our peers, several signals can shape learning. A first signal, elicited by social comparison, concerns the relative skill or welfare of an individual compared to our own and is important for learning dominance relationships. By contrast, observational learning signals enable us to learn from the successes and mistakes of others. In most studies, social comparison and observational learning signals are confounded. Our aim was to disentangle their patterns of neural activity. Methods. Thirty participants performed a probabilistic instrumental learning task. The learning task was designed to manipulate both the learning context - they learned either from the same or different sets of cues as their counterpart – and the outcome valence. We used a reinforcement learning model and multi-voxel pattern analysis on fMRI data to investigate latent behavioural variables estimated and their supporting neural activity. Results. Participants benefited more from observation in the same than in the different cues condition, suggesting an effect of observational learning. Interestingly, they also learned better from rewards than from punishments, especially in the different cues conditions and when observing a competitive, high performing counterpart. Conclusions. Our results show that both observational learning and social comparison affect learning performance in our task. Further analysis will be performed to dissociate the neural basis of both processes. We expect the vmPFC, striatum and ACC to integrate information from different sources of learning (direct experience vs. observation) and the dmPFC to support social comparison. Data analysis is currently ongoing.