Neural Expectation: Cerebellar and Retinal Analogs of Cells Fired by Learnable or Unlearned Pattern Classes

Abstract

This article contains a confluence of several streams of theoretical ideas. Earlier mathematical work on associative learning showed that the unit of long term memory is a spatial pattern, or a pattern of ‘reflectances’. After noticing this, I applied minimality, or Occam’s razor, when I asked: “If the system can learn only reflectance patterns, then does it not discriminate only reflectance patterns? Would not discriminatory capabilities atrophy due to disuse if they could not abet later recognition acts by being learned?” This ‘equal tuning principle’ for learning and perception led me to construct the minimal networks capable of discriminating reflectance patterns.