Using neural networks to determine the strength of association between mechanical variables and spike responses in rapidly adapting cutaneous mechanoreceptor neurons in mouse skin

Rapidly adapting mechanoreceptor neurons were recorded in a preparation of isolated skin and nerve from hairy skin of a rat. The skin was stretched dynamically, and tensile stress, strain, and the spike responses of single neurons were recorded. We determined the strength of association between spikes and stress, strain, and their rates of change using multiple logistic regression and a neural network. Both methods revealed that spikes were more strongly associated with the rate of change of stress than with other variables.