Choice and the dependability of stimuli that predict shock and safety.

This study assessed whether choosing a signalled shock condition over an unsignalled one is controlled by a stimulus that predicts the presence of shock (Experiment I), or by a stimulus that predicts the absence of shock (Experiment II). The dependability of these stimuli as predictors of either the presence or the absence of shock was parametrically varied over a wide range, and subjects (rats) were given an option to change from an unsignalled to a signalled condition. In the first experiment, all shocks were preceded by signals; however, the probability of a signal being followed by shock varied from 1.0 to 0.02. The data obtained indicate that the dependability of the signal as a predictor of shock is unimportant. Rats changed to the signalled condition when the signal was completely dependable (all signals followed by shock) and when the dependability of the signal was systematically degraded. In the second experiment, all signals were followed by shock; however, some shocks were not preceded by a signal. The data show that the dependability of a stimulus predicting the absence of shock is important in that, as dependability decreases, changing to the signalled condition also decreases.