Neural Modeling: Synaptic Plasticity

Reviewing the code of the SPNet++ I realized that the implemented synaptic plasticity mechanism in the simulation is much simpler than I expected. It turned our that the code implements the nearest-neighbor spike model described in Relating STDP to BCM article by Eugene M. Izhikevich and Niraj S. Desai. According to the authors of the article it may be sufficient to only consider two postsynaptic spikes -- the one that occurs before and the one that occurs after the presynaptic firing -- while determining whether a synapse should be potentiated or depressed based on recent spike activity. In addition to attempting to unify different forms of plasticity -- spike-timing-dependent plasticity (STDP) and a standard long-term potentiation and depression (LTP/LTD) -- into a single framework, this approach appears to be very computationally efficient.

update 2006/01/14: A model of STDP based on spatially and temporally local information: Derivation and combination with gated decay paper presents an alternative rule, which is also described as computationally efficient, yet simple and reliable. This paper also provides five types of gating functions: no gating, presynaptic, postsynaptic, dual OR and dual AND gating.

There is also more information on different types of synaptic plasticity in the post on synaptic connectivity.

Leave a comment

what will you say?