Predicting spike timings of current-injected neurons.

Abstract

Typical neurospiking models were examined on the ability to reproduce and predict the spike sequences of a biological neuron for a variety of fluctuating currents, using a fixed set of parameter values. The predicting accuracy was found to be particularly good for the Hodgkin-Huxley models augmented with the Ca(2+)-dependent potassium current generating slow afterhyper-polarization and/or the muscarine-sensitive potassium current. In the successful parameter determination method, the effective membrane time constant is estimated very short, typically about 5 ms. The biological neurons we examined were distinctly classified into two types according to the estimated percent contents of those ionic channels.