Teaching compliance with experimentally managed self-instructions can accomplish reversal shifts

An adult with mild mental retardation was asked to perform a complex visual sorting task according to instructed features. First, she was taught to na me the six objects to be shown in a set of pictures (bird, strawberry, dog, turtle, glass, and cake). Then she was taught to comply with a series of i nstructions (e.g., "Put all the birds here, and the others there") as she s orted through a deck of 15 pictures, 5 of which had the feature just specif ied. This was repented for three of the six features (the other three were saved for subsequent generalization tests). Each time a new object was spec ified constituted a reversal shift of the previous problem. At first, the p articipant failed these shifts. Then a relevant self-instruction was invoke d by asking, "What are you looking for?" She always answered correctly (e.g ., "Birds go here"), but still failed to sort correctly. Then she was given feedback pointing out that she had not acted on what she told herself; she continued to fail. Then she was taught that telling herself what she was l ooking for led immediately to only the five pictures she cited, so that her verbal response would be linked to the target feature. Thereafter, she not only self-instructed correctly, but also sorted accordingly even with new untaught problems. And thereafter, when she was told not to self-instruct, she sorted incorrectly; and when she was again told to self-instruct, she r esumed sorting accurately.