A nonlinear model of step responses in the cockroach tactile spine neuron.

Rapid sensory adaptation in the cockroach tactile spine neuron has previously been associated with a labile threshold for action potentials, which changes with the membrane potential by a process involving two time constants. A feed-forward, variable-threshold model has previously been used to account for the frequency response function of the neuron when stimulated with small-signal, white-noise currents. Here, we used a range of accurately controlled steps of extracellular current to stimulate the neuron. The same model was able to predict the individual step responses and could also fit the entire set of step responses from a single neuron if an initial, saturating, static nonlinearity was included. These results indicate that the two-time-constant, variable-threshold model can account for most of the rapidly adapting behavior of the tactile spine neuron.