As methods for highly specific and precise manipulations of genetics and neuronal activity become the standard in neuroscience, there is growing demand for behavioral paradigms to evolve as well, beyond the simplified and reductive tests which are commonly used. This is especially evident in social behavior, where standard testing paradigms are typically short, involve only a pair of animals, and take place in stimulus-poor environments. Here, we present a series of studies using the Social Box, an experimental setup developed in our lab to automatically track groups of mice living in an enriched environment over days, and extract dozens of behavioral readouts at the individual, dyadic, and group level. We manipulated neuronal populations expressing the socially-relevant neuropeptides oxytocin (OXT) and urocortin3 (UCN3), and utilized genetic mouse models of human disorders affecting sociability – autism spectrum disorder (ASD) and Williams-Beuren Syndrome (WBS) – to demonstrate the importance of the social context in studying mouse behavior. Repeated optogenetic activation of Oxt+ cells recapitulated the known effect of reducing aggressive behavior in the classical resident-intruder paradigm, but in a group of conspecifics it led to an increase in such behaviors on the second day of activation. In parallel, chemogenetic activation of Oxt+ or Ucn3+ cells, separately or together, increased aggressive behavior in the context of a territorial conflict. Finally, behavior of ASD-like mice was mediated by the group composition, such that single-genotype groups showed greater genotype separation in multi-behavioral space than mixed-genotype groups. These findings emphasize the importance of considering contextual and environmental factors when designing and interpreting behavioral studies, which could affect the translatability of findings from mouse to human. Zoom link to join: https://weizmann.zoom.us/j/94822556146?pwd=VnY2eDVGeWdSNmFCVC9zZDVrWUtvUT09 Meeting ID: 948 2255 6146 Password: 884034