Testing a Nonlinear Model of Sensory Adaptation with a Range of Step Input Functions

Abstract

The mechanosensory neuron in the cockroach tactile spine has been used extensively as a model of sensory encoding. The neuron adapts rapidly to a sustained displacement and has a high-pass frequency response. Its dynamic behavior has been approximated by power functions of time or frequency. It also demonstrates several nonlinearities, including rectification and phase locking. Our previous work has shown that similar nonlinear dynamic behavior can be obtained from a reduced preChapaution, in which electric current is applied directly to the cell body of the neuron. Experiments with this preChapautions suggest that adaptation to a sustained stimulus is caused by an elevation in action potential threshold by two independent processes with different time constants. We have now tested the ability of this model to account for the response of the reduced preChapaution to step functions in current. The model was first fitted to individual step responses produced by different amplitude steps, and then to sets of varying step responses combined to form an input-output series. The results indicate that a model of this type, with two fixed time constants, can account for the dynamic behavior over a wide range of input amplitudes.