This project investigates the hypothesis that the amount of information processed by individual neurons per unit metabolic energy cost of doing so is maximized in nature. If the hypothesis is true, biological parameters which affect the information/energy ratio must be optimized accordingly. The parameters which will be considered here are the mean number of synapses per neuron and the mean size and duration of synaptic conductance events. Functional relationships are conjectured to exist between these parameters, the resulting information/energy ratio, and the mean amount of synaptic activity (equivalent to the mean firing frequency). The goal of the project is to obtain these relationships and determine whether the predicted optimum values of the parameters are near the ones observed in nature. To do this, computational models of dendrites and synapses will be constructed, using the most recent available experimental data for their properties and including sources of information loss such as quantal failure and channel noise. These models will then be studied in the context of both biology and information theory, and the information/energy ratio calculated as a function of the number of synapses and the synaptic parameters.
Crotty, Patrick; Sangrey, Thomas; Levy, William B (2006) Metabolic energy cost of action potential velocity. J Neurophysiol 96:1237-46 |