Abstract

Please join us for our Computational Neuroscience journal club on Wednesday 1st May at 2pm UK time in the CBL seminar room

The title is “Feedback Controllability is a Normative Theory of Neural Population Dynamics”, presented by Rui Xia and Youjing Yu.

Summary:

In the production of complex behaviors such as locating, identifying, and grasping food, the brain employs feedback mechanisms, wherein the outputs of a system are rerouted as inputs. Despite substantial evidence supporting feedback control as a normative theory of behavior, if and how feedback control explains neural population dynamics has been largely unarticulated.

In this journal club, we will explore the concept of feedback controllability as a normative theory for understanding neural population dynamics [1]. We will first introduce foundational mathematical concepts of control theory, including controllability, feedback controllers, particularly the Linear Quadratic Regulator. Subsequently, we will discuss the novel dimensionality reduction methods designed to identify subspaces within neural population data that are most feed-forward controllable (FFC) vs. feedback controllable (FBC). Experimental results are then presented to show FBC subspaces as better decoders of behavior and how these two subspaces carry distinct computations with differing regimes of emergent dynamics.

[1] Bouchard, Kristofer, and Ankit Kumar. “Feedback Controllability is a Normative Theory of Neural Population Dynamics.” (2024)