Animals often interact in complex and dynamic sensory environments in which behavior must be continuously adapted. Understanding how the brain processes such environments is a central challenge in neuroscience because it requires explaining how multiple, often conflicting sensory cues are integrated over time, how internal state and behavioral feedback shape perception, and how stable behavioral strategies emerge from these interactions. Despite its importance, this problem remains poorly understood because many of the relevant sensory signals are generated by other individuals and are therefore difficult to quantify, control, and manipulate experimentally.
This question is important for society because social behavior is fundamental to animal and human life, and its dysfunction contributes to a wide range of neuropsychiatric conditions. More broadly, understanding how brains operate in complex and dynamic environments can reveal general principles of adaptive decision-making with implications extending from biology and medicine to artificial intelligence and robotics.
The overall objective of this project was to uncover the computational principles and neural circuit mechanisms that govern social behavior in complex environments. Using Drosophila as a model system with powerful genetic access to neural circuits, the project established quantitative frameworks linking multimodal sensory input to behavior at high temporal resolution, determined how acoustic, chemical, and visual cues are integrated in a context- and state-dependent manner, identified how conflicting sensory information is resolved to guide behavioral choice, and revealed how these computations are implemented in neural circuits and extend to interactions in small groups. To achieve these goals, the project combined machine learning-based behavioral quantification, computational modeling of sensorimotor transformations, and experimental interrogation of neural circuits using optogenetics and calcium imaging.
The project demonstrates that social behavior is governed by dynamic, state-dependent integration of multimodal sensory cues rather than fixed stimulus–response mappings. Behavioral variability reflects structured and adaptive sensorimotor transformations shaped by internal state and closed-loop interactions between individuals. Neural circuits implement these computations through recurrent dynamics and interaction motifs such as excitation and inhibition, enabling flexible behavioral choice. Together, these results provide a mechanistic and generalizable framework for understanding how brains generate adaptive behavior in complex social environments.
