Internal Models of the External World

Dr. Arvind Murugan (host Shankar Mukherji), James Franck Institute and Department of Physics, University of Chicago, Illinois
September 18, 2017 at 4:00 pm
241 Compton
Event Description 

Biological organisms and pattern recognition algorithms interact with the external world in the context of an internal model. Such internal models help focus on the few relevant aspects of a high dimensional external signal while ignoring the rest.

We consider two biological examples of such internal models encoded using dynamical attractors -  place cell neural networks (encoding internal models of space) and bacterial circadian clocks (encoding internal models of time). In these examples, we connect the physics of attractors to biologically relevant questions -  How quickly can internal models be updated when they disagree with external input? What are the biophysical constraints on the reliability of internal models?  When do internal models help and when do they hurt?

 

Coffee:  3:45 pm, 241 Compton