Recent evidence suggests that Wolpe's theoretical formulations are inadequate in accounting for the effectiveness of the therapeutic approach of systematic desensitization. Of the elements traditionally included in the desensitization procedure, neither training in muscle relaxation nor the construction of a graded hierarchy of fear-relevant scenes nor the concomitance of instructed imagination to muscle relaxation are necessary conditions for treatment success; instructed imagination by itself is the only necessary element of the desensitization procedure. Further, the effectiveness of the procedure does not appear to be due to the traditionally stated mutual antagonism between muscle relaxation and anxiety, but rather due to social variables involved in the patient-therapist relationship and cognitive variables involving expectancy of therapeutic gain, information feedback of success, training in the control of attention, and vicarious learning of the contingencies of behavior through instructed imagination. 311 1971 by the American Psychological Association, Inc.

The client's expectancy of improvement is currently regarded as an explanatory construct accounting, in part, for actual psychotherapeutic improvement. In spite of the widespread acceptance of this construct, the empirical data necessary to establish its validity have not been generated. Issues involving the conditions under which expectancy effects have been demonstrated, the circular definition of expectancy, the attribution of causality to expectancy, expectancy versus prediction, and the methodological confounding of expectancy and feedback are discussed.

Criticisms about the effects of therapy have changed markedly during the last 25 years. The challenge by Eysenck (1952) required the demonstration of the superiority of therapy procedures over events classed as nonspecifics. This challenge was met during the 1960s with a convergence of developments involving innovative therapy procedures, sensitive experimental designs, and discriminating conceptual systems. More recently, challenges have required the demonstration of the independence of the effects of therapy procedures from events classed as nonspecifics. When discriminations are made among the heterogeneous events classed as nonspecifics, it is apparent that the effects of therapy procedures have already been demonstrated to be independent of some nonspecifics. Logical tautology precludes the demonstration of independence from other nonspecifics. The role of client outcome expectancy as an interpretive artifact is questionable.Over a quarter of a century ago, a review of available literature ended with the conclusion that "the figures fail to support the hypothesis that psychotherapy facilitates recovery" (Eysenck, 1952, p. 323). A decade Editor's Note. Because the topic of nonspecific treatment effects in psychotherapy is an important and controversial one, I concurred with the suggestion of a consulting editor and invited comments on this article.-S. G.

Research designs in which psychotherapy treatments are compared to placebo conditions were derived from analogies with chemotherapy research. However, chemotherapy and psychotherapy research efforts are complementary rather than analogous. Chemotherapy placebos are delivered to eliminate the psychological variables that psychotherapy researchers attempt to elucidate. Psychotherapy treatments are actually analogous to chemotherapy placebos, rather than to chemotherapy treatments. It is gratuitous to continue to view chemotherapy artifacts, such as client outcome expectancies, as de facto artifacts in psychotherapy research. The elimination of placebo groups in psychotherapy research is recommended. The negative connotation of psychotherapy as a placebo is discussed.

No abstract