Implications of dynamic synapses on synaptic plasticity and neural coding.

摘要

Temporal dynamics are a general feature of synaptic transmission, recently novel aspects of these dynamics have been described in neocortex. These studies of dynamics of synaptic connections between a pair of neurons have shown that successive stimulation gives rise to depression in the strength of the postsynaptic cell response. The central theme of this work is to investigate the implications of these dynamic synapses. One of the important debated issues in the field is the site of expression of synaptic strength. In synaptic transmission, strength of a synapse can be defined either as a presynaptic or a postsynaptic property. First, we examine the effect of these different assumptions on the spatio-temporal receptive fields of simple cells in visual cortex. We have shown that these different assumptions about the origin of receptive field structure lead to very different spatio-temporal dynamics in the case of flashed-bar stimulus. In addition, the results of the reverse correlation study suggests a possible test for differentiating between models. Second, we consider the effects of noise on spatio-temporal neural codes. The single cell firing rate interpretation of the neural code has long dominated the field of electro-physiology. Although spatio-temporal codes are much richer than firing rate codes, the neural responses are very variable. This variability is often referred to as noise. We therefore addressed three questions, (i) how detrimental is the stochastic nature of the neurons to spatio-temporal codes, (ii) can temporal coding enhance coding performed by a population of neurons in the high noise regime, (iii) what are physiologically plausible ways such codes can be generated.;We show that for the random spatio-temporal code, a number of patterns exponential in the dimensionality of the code can be stored for large levels of noise. Based on multiresolution orthogonal transformations (such as wavelet transformation), a code that multiplexes information in time as the neuron codes for more variables has been developed. It is also shown how synaptic dynamics is beneficial in utilizing a temporal code that outperforms a rate code (which is more noise prune) even in the high noise regimes.