The emergence of descriptions in a second-order conditional discrimination: its relationship with the type of training and the temporal allocation of transfer tests

Abstract

Two experiments evaluated (a) the emergence of verbal descriptions after four different types of training on second-order conditional discrimination tasks, and (b) the effects of these descriptions on performance during transfer and maintenance tests. Each experiment involved five groups of four subjectseach, the subjects being university students: each group was assigned to a different type of training or to a control condition without descriptions. We employed: a) training with correction and with 50% of trials with solution; b) training with correction and with 100% of trials with solution; c) training without correction and with 50% of trials with solution; and d) training without correction and with 100% of trials with solution. Each experiment involved tests of intramodal and extramodal transfer. ln the first  experiment subjects had to elaborate verbal descriptions after training and before the transfer tests, whereas in the second experiment descriptions were made after the transfer tests. In the first experiment, the subjects trained with correction and with 100% of trials with solution showed the highest level of correspondence and precision with respect to their descriptions of matching responses; in aIl groups, performance on transfer tests and post-experimental probes was related to the percentage of correct responses observed in training. In the second experiment, subjects trained with correction and with 50% of trials with solution showed the best results during training, but performance on transfer tests and post-experimental probes, as well as verbal descriptions, rarely corresponded to the training results. The influence of the temporal localization of verbal descriptions on transfer and maintenance is discussed.