Instructional control of vasomotor responding : a test of conditioning, two process, and expectancy theories

For many years there has been a debate between proponents of competing learning theories over the role of expectancy in conditioning and extinction. Although this debate has been the subject of an intensive experimental literature, it has not been satisfactorily resolved. Many research designs are intrinsically incapable of dissociating the influence of expectancy from other factors held to be important by competing theories. However, the informed unpairing design (Brewer, 1974), in which subjects are informed of changed contingencies at the onset of extinction, is argued to provide a powerful test of expectancy, two factor, and conditioning theories. It is also argued that the bidirectional vasomotor response provides a solution to the related problems of expectancy manipulation and artifact control that confound previous research using this design . A programme of research based on the informed unpairing design, and using the vasomotor response, was undertaken. Responding in extinction following several expectancy manipulation procedures was compared in subjects given 25 continuously reinforced acquisition trials (CRF25); 100 continuously reinforced trials (CRF100); and 100 partially reinforced trials (PRF). It was found that responding in CRF25 and PRF groups in extinction was abolished by unpairing instructions coupled with removal of the thermal stimulator used for UCS presentation. A significant reduction in responding in CRF25 and PRF groups was also obtained following unpairing instruction alone as compared with groups given no instruction. The CRF25 group instructed at the onset of extinction that they would be reinforced on a PRF schedule showed a non significant trend for greater resistance to extinction than those given no instruction. These results provide strong support for an expectancy based, rather than two factor or conditioning based, explanation of responding in these groups. However, no effects of expectancy manipulation on responding in extinction were obtained in CRF100 groups, and attempts to generate responding in two groups by instruction alone proved unsuccessful. These results are consistent with the hypothesis that there may be two conditioning processes; one expectancy dependent and the other expectancy independent. Existing learning theories based on two conditioning processes are unable to account for all of the results of the present research. However, it is argued that the results of the present study are consistent with a number of studies in the skill learning literature. Models proposed to account for skill learning which distinguish between processes involved in acquisition and early performance, and those involved in much practised responding, provide a possible explanation for the obtained results. Consequences of the research for the behaviour therapies are discussed.