摘要：Four experiments are described that employed a form of go/no go procedure that we refer to as a precursor to the Relational Evaluation Procedure (pREP). The pREP was used to train and test for equivalence responding, and the results of this procedure were compared with the results achieved using a standard matching-to-sample procedure. Each trial in the pREP involved a sample stimulus and either a positive or negative comparison stimulus being presented successively on a computer screen, followed by a 5-sec response interval. Subjects were required to press the space bar of a computer keyboard on sample-positive comparison trials and to not press the space bar on sample-negative comparison trials. Using either the pREP or a matching-to-sample procedure 20 subjects were trained in the following four tasks, A1 --> B1, A2 --> B2, B1 --> C1, B2 --> C2. They were then tested for four symmetrical relations (B1 --> A1, B2 --> A2, C1 --> B1, C2 --> B2), and two equivalence relations (C1 --> A 1 and C2 --> A2), using both the pREP and matching-to-sample procedures, with the order of presentation of the two types of test varied across experiments. The results of the four experiments reported here demonstrate that the pREP is less effective than the matching-to-sample procedure in generating equivalence responding. However, performance on the pREP tests improved when the subjects had prior exposure to matching-to-sample training and/or testing. This finding suggests that further study of the interactions between these two procedures, combined with suggested refinements to the pREP itself, may contribute to a fuller understanding of those variables most relevant to producing equivalence responding.
