Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

BCCN/BFNT AG-Seminar

Tuesday, 20.05.2014 17:15 c.t.

Perceiving, learning and forgetting – a physicist’s take

by Prof. Dr. Anita Mehta
from S.N. Bose National Centre for Basic Sciences, Kolkata, India

Contact person: Annette Witt

Location

Ludwig Prandtl lecture hall

Abstract

I will present an outline of my recent research, which uses the tools of statistical physics to examine problems in cognition and neuroscience. First, I will discuss our analysis of eyetracking data from an experiment designed to explore the effect of literacy on audiovisual processing. Fixations were found to be most strongly affected by literacy levels while scale-invariance was observed in the corresponding saccadic distributions. Our analysis suggests that saccades are characterised by Levy-like dynamics, with exponents that are reminiscent of those obtained in studies of foraging. Next, I will describe an agent-based model of synaptic competition, where the prediction of memory is inherent in a timescale for forgetting a learned signal. Among its main results is the prediction that learning is optimal for extremal values of synaptic strength. Finally, I will describe a theoretical model of metaplastic synapses which manifests power-law forgetting. I will also touch on its recent use to explain a novel form of potentiation, which was found to be use-, but not time-dependent, in experiments done on Lymnaea neurons.

back to overview