Biological evaluation of a Hebbian-Bayesian learning rule

• A correlation based Hebbian-Bayesian learning rule formulated on theoretical, probabilistic grounds has been extended to an incremental version running in continuous time and with spiking units. This learning rule has, however, not previously been evaluated in any detail with regard to biological plausibility and ability to mimic synaptic long-term potentiation and depression. It is demonstrated here that this learning rule indeed captures several fundamental aspects of Hebbian spike-timing dependent synaptic plasticity. A slightly modified version of the model gives a quantitative fit to data.

Nyckelord

synaptic plasticity model

BCPNN

spike-timing dependency

long-term potentiation

hippocampal slices

neurons

memory

model

mechanism

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)