Thursday, October 16, 12:20 to 1:10 p.m., Room 109 Uhler
From Cortical Fields to Hidden Patterns: Neural Network Revisions to Köhler’s Isomorphism
Dr. Ray Pavloski
How is sensory input encoded both as the noisy, high-dimensional activity recorded from the brain and as low-dimensional and noise-free perceptual experience? In 1929, Köhler argued that there are physical fields distributed throughout the cortex that stand in a 1:1 relationship with the structures of perception. Current evidence suggests that perceptual experiences are self-organized biological patterns, but their relationship to the activities of neurons and neural networks remains unexplained. For the past several years, Dr. Pavlovski has used neural network modeling and analysis of computer simulations to study how patterns having the dimensions and structures of relations that characterize perceptual experience might be hidden within neural network activity. He will give an overview of this work, starting with his initial efforts to devise simple neural networks that produce hidden patterns and progressing to more recent efforts to find hidden patterns that are naturally produced by certain networks.
Presented by the Department of Psychology.