A Distributed Representation of Remembered Time
Abstract
The goal of the award was to extend a mathematical framework for representing time to also construct scale-invariant representations of space. The extensive neurophysiological work done on the hippocampal place code(which resulted in a Nobel Prize in 2014) places physical constraints on this mechanism. We accomplished this goal by developing a computational framework that describes a wide range of functional cellular correlates in the hippocampus and related brain regions. Critically, this cellular-level model is connected to behavioral memory performance via cognitive models that take in the mathematical form of the representation. We have exceeded the initial goals of the project by developing a representation of numerosity as well as space using the same framework and a detailed model for function translation that is constrained by neurophysiological data from the hippocampal theta oscillation literature. This is an important development, as it opens up new frontiers in cognitive computation. Taken together, the work performed in this three year period was crucial in developing a mathematical model for cognitive operations that is constrained by neurophysiology. This model connects data from the subcellular level, to the systems neuroscience level to cognitive modeling. The mathematical framework is capable of forming the basis of a general brain-inspired cognitive computer.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 19, 2015
- Accession Number
- AD1000737
Entities
People
- Marc W Howard