Sequence learning via biophysically realistic learning rules (Cone and Shouval 2021)


This work proposes a substrate for learned sequential representations, via a network model that can robustly learn and recall discrete sequences of variable order and duration. The model consists of a network of spiking leaky-integrate-and-fire model neurons placed in a modular architecture designed to resemble cortical microcolumns. Learning is performed via a biophysically realistic learning rule based on “eligibility traces”, which hold a history of synaptic activity before being converted into changes in synaptic strength upon neuromodulator activation. Before training, the network responds to incoming stimuli, and contains no memory of any particular sequence. After training, presentation of only the first element in that sequence is sufficient for the network to recall an entire learned representation of the sequence. An extended version of the model also demonstrates the ability to successfully learn and recall non-Markovian sequences.

Region(s) or Organism(s): Visual cortex

Cell Type(s): Abstract integrate-and-fire leaky neuron

Model Concept(s): Sequence learning; Eligibility traces

Simulation Environment: MATLAB

Implementer(s): Cone, Ian [iancone at rice dot edu]

References:

Cone I, Shouval HZ. (2021). Learning precise spatiotemporal sequences via biophysically realistic learning rules in a modular, spiking network. eLife. 10 [PubMed]


This website requires cookies and limited processing of your personal data in order to function. By continuing to browse or otherwise use this site, you are agreeing to this use. See our Privacy policy and how to cite and terms of use.