A synaptic and temporal ensemble interpretation of spike-timing-dependent plasticity

In this thesis we are concerned with activity-dependent neuronal plasticity in the nervous system, in particular the phenomenon of spike-timing-dependent-plasticity or STDP. We find that the experimental evidence for STDP may be interpreted in a variety of ways. Inspired by this observation, we propose a new model of spike-timing plasticity in the form of a synaptic switch rule. The switch rule governs changes at individual synapses, and only when the rule is averaged over multiple synapses and multiple spike-pairs does an STDP-like rule emerge. The STDP-like rule is therefore an ensemble property of our model, one that is nowhere instantiated at any individual synapse. We find that our switch rule explains a variety of spike- and rate-based plasticity results as a result of its intrinsic structure. We also find that stable, competitive dynamics emerge naturally due to multi-spike interactions. At no stage are we required to introduce additional modifications to accommodate particular experimental results or avoid otherwise undesirable learning behaviours. Indeed, ensuring consistency with various experimental results serves to neatly constrain the parameters of our model in a concise manner. This is in contrast to many other models of STDP, which are often required to introduce additional modifications and non-linearities to explain experimental results on a case-by-case basis. Furthermore, out synaptic switch rule is considerably simpler than many competing models of STDP and places a much lower computational burden on individual synapses. We are therefore freed from the need to postulate precise coincidence detections mechanisms and, as a result, our synaptic switch rule is broadly consistent with a range of possible biological implementations.