Continuous functions for the analysis of sensory transduction.

Sensory transduction at a primary receptor neuron yields a current that drives the generation of action potentials. Due to the inaccessibility of that current for direct measurements the analysis of sensory transduction requires the use of neuronal output functions that give an indirect measure for the "input" current, i.e. the current at the impulse initiating site. Three continuous neuronal output functions are investigated with respect to their ability to reconstruct the input current (i) the membrane potential recorded under sodium channel block referred to as "receptor potential", (ii) the interspike-interval function (Awiszus 1988a) and (iii) the phase lag function which is introduced in this paper. The behaviour of these three functions for constant and dynamically varying input is studied at the Hodgkin-Huxley model (Hodgkin and Huxley 1952) because for this model neuron it is possible to compare the input current estimates obtained from the output functions with the true input current. It was found that for constant and for sufficiently slow varying input all three functions allow a valid reconstruction of the input current time course. On the other hand, if the input current changes rapidly all three estimated input current time courses show considerable deviations from the true time course. The largest maximal deviation is shown by the current estimate obtained from the receptor potential whereas the phase lag function yields the smallest input current misjudgement. An experimental example to illustrate the procedure to obtain the phase lag function for a muscle spindle primary afferent is given.