Empirical analysis of biasing procedures to establish reject-control relations

ABSTRACT The establishment of reject control during conditional discrimination training disrupts the formation of experimenter-designed equivalence classes. In the matching-to-sample context, reject control describes a situation in which the choice is under control of a sample and S− relation. Specifically, participants learn to choose the correct comparison by rejecting or responding “away from” the S−. The present study investigated procedures that increase the likelihood of reject control during conditional discrimination training. Experiment 1 examined the effects of programmed differences in the distribution of sample/S+ and sample/S− relations. In Experiment 2, we used the same training protocol as Experiment 1, while adding constrained access to S+ observing by using a “mask” to cover S+ stimuli. All experiments presented conditional discrimination training followed by equivalence, symmetry, transitivity, and reflexivity tests. In Experiment 1, participants failed to meet criteria during baseline training or, if participants passed baseline relations, equivalence testing. In Experiment 2, reject-control patterns were observed in equivalence tests. Results suggest that reject control is not as easily established as supposed.