Characterizing Neural Code from a Minimum-Description-Length Perspective
Abstract
We have successfully completed the projects in our original proposal and also performed some additional, closely relatedstudies. (1) We have proposed a new theoretical framework for understanding sensory encoding based on the modernminimum-description-length principle. A key finding is our new hypothesis that firing rates of sensory projection neurons areproportional to optimal code length for stimulus features (i.e., negative log estimated probability of stimulus features). This is insharp contrast to the traditional view that sensory neurons' firing rates represent the probability of the neurons' preferredstimulus in the input. (2) We have proposed a new theoretical framework for understanding visual decoding. Experimentalstudies have firmly established that encoding is hierarchical, progressing from lower-level representations of simpler and lessinvariant features to higher-level representations of more complex and invariant features along visual pathways
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 17, 2020
- Accession Number
- AD1105888
Entities
People
- Jun Zhang
- Ning Qian
Organizations
- Trustees of Columbia University in the City of New York