How Psychoanalysis is a Science—and How It is Not*

Arguments concerning the scientific status of psychoanalysis have been going on for some time now and, judging by the vehemence with which positions are held, show little sign of resolution. At one extreme are the positivists, those who judge by what they take to be the standards of physical science. The biologist P. B. Medawar declares that “doctrinaire psychoanalytic theory is the most stupendous intellectual confidence trick of the twentieth century” (1975, p. 17). The philosopher of science, Grünbaum, attempts a comprehensive judgment of psychoanalytic propositions in his forthcoming book, Is Psychoanalysis a Pseudo-Science?, and concludes that, while some psychoanalytic propositions are testable, there is little solid evidence for most of the theory. A more sympathetic conclusion is reached by Kline (1972) and Fisher and Greenberg (1977), who sort through the experimental or “controlled” evidence available on a variety of psychoanalytic concepts and argue that some are scientifically validated while others are not. Whether hostile or friendly, however, these assessments all share allegiance to a set of standards derived from physical science that are, in my view, inappropriate to the data of psychoanalysis. All assume that the primary observations on which psychoanalytic theory is based — observations from the psychoanalytic and psychotherapeutic interaction — are not suitable for the validation of the theory. This can only be done by testing the theory with other methods or on different data. But when the theory is moved to these other spheres — the laboratory, the controlled experiment, the statistical survey — there is