Rodrigo Echeveste, Ph.D,  Argentina’s National Research Council (CONICET)

Abstract:

The dynamics of sensory cortices show a suite of basic, ubiquitous features that have so far escaped a common, principled theoretical account. These include strong inhibition-dominated transients at stimulus onset, gamma oscillations, and noise variability – all stimulus-dependent. We present a unifying model in which all these dynamical phenomena emerge as a consequence of the efficient implementation of the same computational function: fast probabilistic inference. For this, we use a novel approach and train a recurrent E/I neural circuit model of a V1 hypercolumn. The network is required to modulate not only the mean (as conventional) but also the variability of its stationary response distributions in order to match the corresponding input-dependent posterior distributions inferred by an ideal observer. The optimized neural circuit features a number of remarkable properties. Firstly, the network discovers non-equilibrium dynamics, a state-of-the-art machine learning strategy to speed up inferences. The circuit also exhibits realistic biological properties for which it was not trained directly. It achieves divisive normalization and displays marked transients at stimulus onset, as well as strong gamma oscillations, both scaling with stimulus contrast. Crucially, these dynamical phenomena do not emerge in other control networks trained to match either mean responses only, or mean responses and variances (without modulating response co-variances). This model allowed us to make predictions regarding the tuning of transient responses that we later confirmed in novel analyses of awake monkey V1 recordings.

Brief bio:

Rodrigo Echeveste obtained his Bachelor and Masters Degrees in Physics from Balseiro Institute in Argentina, and his PhD from the Goethe University in Frankfurt, Germany. After a three year stay in the UK as a postdoc at Cambridge University’s Computational and Biological Learning Lab (CBL), he returned to Argentina where he now holds a permanent research position from Argentina’s National Research Council (CONICET) at the Research Institute for Signals, Systems and Computational Intelligence, sinc (i).

Zoom Link: https://zoom.us/j/153319823

Or iPhone one-tap :
US: +13126266799,,153319823# or +16468769923,,153319823#
Or Telephone:
Dial(for higher quality, dial a number based on your current location):
US: +1 312 626 6799 or +1 646 876 9923 or +1 301 715 8592 or +1 346 248 7799 or +1 408 638 0968 or +1 669 900 6833 or +1 253 215 8782
Webinar ID: 153 319 823
International numbers available: https://zoom.us/u/adBWphwtx6

Speaker

Rodrigo Echeveste, Ph.D,  Argentina’s National Research Council (CONICET)