Abstract
Apparent evidence that awareness regularly mediates verbal conditioning actually may by attributable to a self-fulfilling experimental methodology. A more appropriate procedure was devised in which a subtle reversal of the roles of S and E distracted the actual S from acquiring the problem-solving set characteristic of previous methods. This "double-agent" approach was established in the context of an interview. The interviewer considered himself the E, and was told that his assignment was to reinforce a particular motor response in the interviewee. The interviewee was in fact the E; he used that motor response to reinforce a particular choice of verbal prompt in the interviewer. Thereby, the double-agent interviewee was able to accomplish seven reversals in the relative rates of 2 of the interviewer's prompts. Frequent probing of the interviewer indicated that he was not aware of these changes in his own behavior, and that he had other irrelevant hypotheses to explain his variable apparent success in reinforcing the interviewee's motor response. Variable success in obtaining the effect across Ss may eventually be accounted for by parameters of the setting and by the S's other characteristic uses of the verbal responses chosen for study. Thus, the results indicate that verbal operant conditioning can occur without awareness in some persons, and that a person may unknowingly let his verbal behavior be controlled by another person in order to himself gain control over certain aspects of the other's behavior