Beyond Spike Timing Theory – Thermodynamics of Neuronal Computation

Abstract

This paper highlights ionic fluxes as information carriers in neurons. The theoretical framework regarding information transfer is presented as changes in the thermodynamic entropy that underlie specific computations determined by ionic flow. The removal or accumulation of information is analyzed in terms of ionic mass transfer related with changes in Shannon information entropy. Specifically, information transfer occurs during an action potential (AP) via the voltage gated ion channels in membranes and the same physical mechanism can be extended to various types of synapses. Since sequential APs from a selected neuron are not alike, then every spike may transfer slightly different amounts of information during their occurrence. The average efficiency in information transfer during APs is estimated using mutual information measures and Hodgkin-Huxley model. This general scheme of ions as carriers of information represents the required physical machinery for a dynamic information transfer that is missing in the current spike-timing description.