Big Question: From neural circuits to computation

Principal Investigators: Stephanie Palmer, Organismal Biology and Anatomy ; Jason MacLean, Neurobiology

Funding Type: Vision

Focus Areas: Cognition, Information

Big Idea: Understanding what and how neural circuits compute is the central question of modern neuroscience. Several current approaches view neural circuits as stationary sets of connection weights, whereas our work focuses on the time evolving activity in the brain.  We postulate that the computational structure in the brain is contained in these dynamics rather than the fine details of static neuron-neuron connections.  We further hypothesize that the structure of these dynamics implement both priors and approximations the brain has implemented over evolutionary time to best represent and react to stimuli from the environment.  That is, they perform approximate computations that are optimized for the particular structure of inputs they receive, shaped by the organism’s ecological niche.  By making large-scale recordings of circuit dynamics in primary visual cortex in the intact brain, we will analyze and detect computational structure from the circuit activity as the inputs change in real time. By showing both standard artificial stimuli as well as clips from our database of natural motion video, we will infer what features of neural dynamics encode the unique statistics of the natural visual world. Additional experiments will allow us to to show how a neural circuit adjusts what it computes via dynamic changes in its activity.