Social cognition is believed to contribute to the collective success of groups. To learn from others, individuals must form representations of their own and conspecifics’ actions and outcomes, integrating this information to enhance their understanding of the environment. Neurons in the frontal cortex of socially interacting mice represent various aspects of conspecific behavior, including fear, pain, competition, and social hierarchy. However, whether these neural circuits are involved in observational reinforcement learning during goal-directed tasks remains to be determined. To investigate this, we adapted a two-armed bandit task for a joint design where a head-fixed demonstrator and an observer mouse facing each other alternately forage for a water reward in a two-alternative forced-choice scenario. Preliminary data indicates that observer mice use visual cues of expert demonstrators’ actions and outcomes to better infer the action with the higher reward probability, thus enhancing their decision-making. Notably, mice switched their licking behavior more effectively towards the spout with a higher reward probability when they could visually observe their conspecific, compared to when a partition obstructed the view. Chronic recordings from the Anterior Cingulate Cortex using Neuropixels probes in observing mice revealed neurons that encoded multiple task features, including auditory cues, rewards, and specific actions. Key-point-based analysis identified units responsive to motor actions for both the self and conspecifics, in a mirror-neuron-like fashion. Further investigations will aim to elucidate the neural underpinnings of latent variables about the environment and conspecific in the mouse brain during joint foraging.