Periodicity of response in operant extinction

The distribution of responses during experimental extinction trials for a group of 11 rats, trained in the Skinner-box, was examined. Our analysis supports an earlier suggestion that the extinction trial may be separated into respondent and silent periods. Response rates during respondent periods show relatively little decline throughout the extinction period investigated, whereas silent periods increase with positive acceleration. The traditional exponential function used to describe response rates during experimental extinction trials should therefore be regarded as a very rough approximation to the mathematical characteristics of the animal's behaviour. Its use as a basis for quantifying critical intervening variables, such as habit strength or behaviour potentiality, is consequently seriously challenged.