A biophysical calcium-dependent model of synaptic plasticity: homeostasis, selectivity and beyond.
Luk C. Yeung, Leon N Cooper and Harel Z. Shouval
We have recently proposed a Unified Calcium Model of synaptic plasticity,
where the intracellular calcium concentration is the key associative signal
for the Hebbian synaptic plasticity. This model has been able to account for
various plasticity induction protocols, such as rate-based and spike
time-dependent plasticity. Furthermore, it implies novel, experimentally
testable consequences. However, like any traditional Hebbian learning
rule, it is inherently unstable.
Homeostatic metaplasticity had previously been suggested as a means of
stabilizing Hebbian plasticity, such as in the BCM sliding threshold
mechanism. We formulate a biophysical mechanism for the metaplasticity,
based on a cell-wide activity-dependent regulation of NMDAR conductance.
This form of metaplasticity stabilizes the Unified Calcium Model, while
maintaining the results from the standard plasticity-induction protocols.
In fact, we show that metaplasticity can account for the experimental
results of synaptic scaling, where the synaptic weights scale down as the
neuronal activity increases.
In addition to robustness, interesting collective features arise from the
combined plasticity-metaplasticity system. We show that the simulated neuron
is sensitive to differences in the spike-train statistics, being able
develop stable and selective receptive fields in different input environments.