Geometry and Dynamics of Neural Computation

The computation-through-dynamics framework postulates that neural circuit computations can be understood through the lens of dynamical systems theory. In recent years, dynamical systems trained to perform behavioural tasks or fit directly to data have reshaped our understanding of the neural processes underlying various sensory, motor, and cognitive functions. In parallel, there has been increasing interest in the geometry of neural population activity, which has influenced our understanding of the neural representations of task variables. However, the relationship between the dynamic and geometric views of neural circuit computation remains under-explored. In particular, both geometry and dynamics emerge from the underlying neural circuitry and are shaped by the behavioural task under consideration, suggesting the possibility of an integrated perspective on geometry and dynamics in neural systems.
The present workshop aims to foster discussion between researchers working on dynamical systems and geometric methods to develop a unified view of geometry and dynamics in neural computation.