Colloquium, David Heeger, New York University
Canonical Computation in Brains and Machines
Working memory is a cognitive process that is responsible for temporarily holding and manipulating information. Most of the empirical neuroscience research on working memory has focused on measuring sustained activity in prefrontal cortex (PFC) and/or parietal cortex during simple delayed-response tasks, and most of the models of working memory have been based on neural integrators. But working memory means much more than just holding a piece of information online. I will present a new theory of working memory, based on a recurrent neural circuit that I call ORGaNICs (Oscillatory Recurrent GAted Neural Integrator Circuits). ORGaNICs are a variety of LSTMs (imported from the machine learning/AI literature) that can be used to explain the complex dynamics of delay-period activity in real neural circuits during a working memory task, and that offer some computational advantages over conventional artificial recurrent neural nets. Time permitting, I’ll also say a few words about inference, exploration, prediction, and the role of feedback (top-down) in the brain.