Detecting Changes: a Probabilistic Interpretation of Short-Term Plasticity
Neurons are highly sensitive to changes through various aspects of short term synaptic plasticity (STP). In particular, short- term synaptic depression (STD) which is believed to enhance detectability of changes. We use a probabilistic approach to derive the dynamics of STP that signals a change in the input. We found that, depending on the parameters, the underlying synaptic computation is an integration of dynamic EPSPs that ranges from depression to facilitation. The results suggest that deviations of the membrane potential of a sensory neuron from its resting level might reflect transition probabilities of hidden variables to which it is tuned. Furthermore, in comparison to existing models, predictions with respect to input statistics are discussed.