Accuracy of a two-lever operant discrimination task : Three experiments

Little research is currently available concerning the most accurate method for assessing discriminative control during a two-lever choice operant task when extinction sessions are utilized. Past and present research has noted a decrease in correct bar presses during extinction when compared to correct bar presses during reinforced responding. Investigators have noted an apparent 'cue-search' exhibited by animals during extinction testing and have proposed that the elimination of the food reinforcement cue is responsible for searching behavior and subsequent low discrimination scores. A growing body of evidence obtained from operant discrimination experiments, psychopharmacology research, and motivational investigations is reviewed here and methods for elimination of animal searching behavior are proposed. Since extinction sessions, by definition, require the elimination of the reinforcement cue, methods for altering cue valence or cue importance are discussed. A series of investigations was performed to determine the effect of motivational states on performance of a two-lever operant discrimination task. In all experiments, animals were trained to discriminate between tactile cues and then tested during extinction periods. Experiment I investigated discrimination accuracy for groups trained under a food-motivated condition and tested either in a motivated or non-motivated state. Experiment II followed the general pattern of the first investigation but with an additional point on the motivation continuum. Experiment III investigated the effect of amount of preliminary training when in a motivated state and subsequently tested in a non-motivated condition. Results from these experiments indicated that the non-motivated animals performed the discrimination task significantly better than did animals in a motivated condition. In addition, results suggested that a minimal amount of training was necessary to produce accurate, non-motivated discrimination during extinction. There was no discernible relationship observed between number of responses occurring during extinction periods and deprivation level. Results are discussed in terms of the reinforcement cue, cue valence, and state-dependent learning theory.