Assessment of safety generalization and demonstration as a function of various training stimulus parameters

Two experiments explored training of safe leg position behavior in an office setting. Both experiments measured (a) generalization of safety-related stimuli along a generalization gradient and (b) demonstration of the limits of “safe” behavior. Experiment 1 used a 2 x 2 between-subject factorial design plus a non-differential stimulus training group to assess the effects of (a) safe [S+] versus both safe and at-risk stimulus classes [S+ and S-] and (b) the similarity between the safe [S+] and at-risk [S-] stimuli along a continuum. The results show that training with stimuli from both classes, and greater similarity between S+ and S- stimuli, will increase correct classification of at-risk stimuli (i.e., decrease responding to S-) and decrease over-extension of the limits of safe behavior. Experiment 2 used a 2 x 2 between-subject factorial design to determine the effects of (a) the relative frequency (i.e., ratio) of safe to at-risk training trials and (b) the absolute frequency of safe and at-risk training trials. The results show that training should use a lesser S+ relative frequency to increase correct classification of at-risk stimuli and decrease overextension of the limits of safe behavior. The absolute frequency of trials did not show an effect on either of the generalization or demonstration dependent variables.