Cosyne Workshop: The Dynamic Brain

The brain is tasked with navigating the complex and ever-changing world, where it must adapt to time-varying environmental conditions, evolving internal states, and the needs of the organism. To achieve this, the brain must balance stable, hard-wired circuits with flexible, dynamic processes. Recent advances have challenged the traditional view of innate behaviors as being governed solely by rigid, pre-programmed circuits. Instead, the emerging perspective emphasizes the role of dynamic neural computations—mediated by the coordinated activity of large neural populations—as key to understanding both adaptive and innate behaviors.

This workshop will bring together experts in computational neuroscience, dynamical systems, memory and behavior to explore how population-level dynamics, such as attractors, manifolds, and neuromodulation, shape both innate behaviors and adaptive processes. Central to this discussion is the question of how dynamics that govern complex behaviors—spanning motor, cognitive, and affective processes—are instantiated across model systems.

A major focus of the workshop will be computational methods for analyzing and modeling neural dynamics. What latent states underlie neural computations, and how can they be inferred? How can dynamical systems approaches help uncover the principles governing neural population activity? What are the best strategies for identifying latent neural states in both stable and non-stationary environments? How do neural circuits organize and regulate innate behaviors? How do factors such as biological variability and network structure shape these dynamics?

We will explore how computational methods—from dimensionality reduction and latent variable modeling to recurrent neural networks—in conjunction with experimental findings can uncover meaningful patterns in neural data, reveal underlying neural mechanisms, and advance our understanding of neural coordination across time scales and behaviors.

Questions & Discussion ideas can be added here: (Drive Link)