The temporal dynamics of synapses and synaptic decoding

The strength of many synapses is modified by various use and time-dependent processes, including facilitation and depression. A general description of synaptic transfer characteristics must account for the history-dependence of synaptic efficacy and should be able to predict the postsynaptic response to any temporal pattern of presynaptic activity. Here I develop the synaptic decoding method which, after training on data from random spike sequences, predicts the postsynaptic response to arbitrary presynaptic spike trains. The method characterizes both the strength and the temporal dynamics of a synapse. Using this method, I fit the responses of several neuromuscular junctions in the stomatogastric system of the crab Cancer borealis which allows a comparison of their responses to sequences of action potentials and the time course and degree of facilitation that they exhibit. The method is extended to characterize modulation of these neuromuscular junctions. At these neuromuscular junctions modulators effect both synaptic strength and facilitation but preserve a characteristic relationship between them. The method is further extended to synapses in the rat primary visual cortex which are well characterized by two components of depression and one component of facilitation. With theoretical and modeling studies I show that the temporal dynamics of synapses have important functional consequences. Specifically, (1) short-term depression of intra-cortical synapses provides a dynamic gain-control mechanism and (2) temporal dynamics of cortical synapses can account for cortical contrast adaptation. I relate the macroscopic model of temporal dynamics to a microscopic synapse model based on the probabilistic release of transmitter which explains the experimentally observed relationship between facilitation and strength. Based on the microscopic synapse model I construct a model network and show that the temporal dynamics of synapses can play a significant role in emergent network phenomena. Finally, I show that Hebbian learning algorithms that modify synaptic reliability also affect the temporal dynamics of synapses. With correlated activity between presynaptic and postsynaptic neurons, unreliable synapses that display facilitation develop into reliable synapses that display depression.