Synaptic depression as an intrinsic driver of reinstatement dynamics in an attractor network

• Long-term memory consolidation is commonly assumed to occur through the reinstatement of previous activation states in the cortex by the action of the medial temporal lobe (MTL) memory system. In order to produce a sequence of reinstated patterns the MTL system either needs to be externally cued, or have an intrinsic dynamics enabling it to present earlier learned patterns. We show that the Hebbian depression of synaptic connections between the neurons in an attractor network can produce such a dynamics that retains information about relative attractor strength.

Nyckelord

synaptic depression

attractor networks

Bayesian confidence propagation neural networks

memory reinstatement

bayesian learning rule

memory

model

Publikations- och innehållstyp

ref (ämneskategori)

art (ämneskategori)