Entropy and Information in Neural Spike Trains

摘要: The nervous system represents time dependent signals in sequences of discrete, identical action potentials or spikes; information is carried only the spike arrival times. We show how to quantify this information, bits, free from any assumptions about which features train input signal are most important, and we apply approach analysis experiments on a motion sensitive neuron fly visual system. This transmits stimulus at rates up 90 bits/s, within factor 2 physical limit set by entropy itself.