The recognition of social cues is a major determinant for the selection of social behaviors under specific contexts. To gather social information, animals are tuned to specific multi-sensory stimuli emitted by interaction partners signifying e.g., age, sex, and hierarchy status. The dynamic nature of social interactions also means that the social cues are presented in a transient and stochastic manner. How the brain gathers and integrates different streams of multi-sensory social information, which potentially arrive with different timing, remains unclear.

As a model system to address this question, we chose to study the medial amygdala (MeA), a critical hub in the innate social behavior circuit. Previous studies demonstrated that the MeA receives strong pheromonal inputs, and that activating specific cell populations within the MeA can drive specific innate social behaviors including mating, aggression, allogrooming, and parenting. Despite this, how the MeA integrates incremental social information such as monomolecular pheromonal inputs to form an abstract social representation critical for the expression of specific behaviors remains elusive.

To this end, we developed an explantable semi-chronic recording method using Neuropixels probes (termed Re-pix) to measure the responses of the MeA neurons to a variety of social cues in awake animals, including monomolecular pheromones, pheromonal blends, volatile scents, and other stimuli including social touch. The large-scale recording of MeA neurons allowed us to chart the selectivity of single MeA neurons to different types of social stimuli. Moreover, we found that MeA neurons respond by at least two different types of kinetics and that the dynamics of neural response depends on stimulus modality and is modulated by prior social experience. Taken together, these results suggest the convergence of multi-pheromonal and multi-sensory inputs within MeA neurons with experience-dependent modulation.