Bulletin of the Psychoanalytic Research Society, Volume IV, Number 1, Spring, 1995

How Psychotherapy Works: The Findings of the San Francisco Psychotherapy Research Group

Joseph Weiss, MD, Harold Sampson, PhD, and Lynn O'Connor, PhD
San Francisco Psychotherapy Research Group

Jospeh Weiss, MD and Harold Sampson, PhD began a research collaboration in 1964 with the purpose of studying by rigorous, quantitative research methods how psychoanalysis and psychotherapy work. Their research was inspired by earlier informal studies that Weiss had carried out over a number of years using process noses of psychoanalyses. Weiss's approach to these initial informal studies was empirical. Specifically, he investigated what events precede significant therapeutic progress (e. g., patients becoming aware of previously repressed ideas, experiencing new affects, developing insights, or relinquishing symptoms).

Weiss observed in his process notes that patients often made significant progress without any preceding interpretation by the analyst. Weiss inferred, from studying sequences leading to this progress, that patients made progress when they felt safe enough to do so. Moreover, patients work actively, often without awareness, to increase their sense of safety. One way they do so is by unconsciously testing beliefs about themselves and their interpersonal world in relation to the analyst. For example, a patient who believes that any disagreement with the analyst may hurt him (as she believed that disagreements with her father in childhood had caused him harm) tested this belief over a long period of time in analysis. She did so by disagreeing with the analyst's interpretations, at first timidly, and later forcefully. When the analyst's reactions tended to disconfirm her belief (that is, when the analyst "passed" her tests), she became more relaxed, more bold, and more insightful.

Weiss gradually developed a new psychoanalytic theory of the mind, of psychopathology, and of treatment. He proposed that psychological problems stem from pathogenic beliefs acquired in traumatic relational experiences in childhood, that these beliefs tell the individual that it is dangerous to pursue certain important goals, and that therapy is a process in which therapists work to change these beliefs through experiences with the therapist as well as through knowledge acquired in therapy.

A distinctive aspect of Weiss's theory is the assumption that patients are highly motivated to solve their problems and that they work actively throughout treatment (by testing the therapist) to obtain experiences and knowledge that will help them do this. Also distinctive is the assumption that the patient works in accord with unconscious plans as to which problems to tackle first, and which ones to defer until later. Patients decide unconsciously how they may work with their therapists to get help. They unconsciously coach their therapists with the aim of guiding them, so that they may provide the experiences, display the capacities, or convey the knowledge that patients need to disconfirm their pathogenic beliefs. Patients unconsciously monitor their therapists' attitudes toward the pathogenic beliefs they are working to change.

Because Weiss's new theory was based on observations of how change takes place, and appeared to have considerable explanatory power, Weiss and Sampson decided to subject it to rigorous empirical scrutiny. They met daily over a period of years to plan how to test Weiss's hypotheses about the therapeutic process. They figured out precisely what Weiss's theory asserted about nature, deduced specific empirical consequences of the theory, identified situations in which alternative psychoanalytic hypotheses predict different clinical observations. selected or developed reliable measures, and finally devised research designs with appropriate controls for sources of error (Weiss & Sampson, 1986; Weiss, 1993). In effect, they developed over time the intellectual capital required to undertake a formal program of research.

Pilot studies were conducted in 1967, and the first research paper was published several years later (Sampson, Weiss, Mlodnosky & Hause, 1972). In 1972, Weiss and Sampson established the Mt. Zion (now called the San Francisco) Psychotherapy Research Group. This group began a series of interrelated studies of the therapeutic process, using verbatim transcripts of psychoanalyses and psychotherapies.

Research findings of this group support the idea that patients want to overcome their problems, that they exert unconscious control over their repressions, and that they make progress in therapy when they decide unconsciously that it is safe to do so. For example, one study showed that repressed mental contents frequently become conscious in treatment without prior interpretation, that they are experienced vividly yet without stress or conflict, and that they are kept in consciousness and used by patients to increase their understanding of their mental life (Gassner, Sampson, Weiss & Brumer, 1982). Moreover, therapy events that increase patients' sense of safety, such as the therapist passing the patients' tests or giving patients interpretations that help them to understand and disconfirm their pathogenic beliefs, are typically followed by immediate patient progress (Broitman, 1985; Bush & Gassner, 1986; Davilla, 1992; Fretter, 1984; Fretter, Bucci, Broitman, Silberschatz & Curtis, 1994; Kelly, 1989; Linsner, 1987; Silberschatz, 1986; Silberschatz, Curtis, Sampson & Weiss, 1991; Silberschatz, Fretter & Curtis, 1986; Weiss, 1993a, 1994; Weiss, Sampson et al., 1986).

The consistent finding of immediate patient progress following a therapy event that increases a patient's sense of safety supports the idea that patients unconsciously and continuously monitor the therapist's behavior and attitudes and are immediately influenced by indications that support or challenge their pathogenic beliefs. Patients unconsciously are in touch with their environment and are guided by their appraisals of their reality.

We have also accumulated a great deal of evidence to support Weiss's concept that patients work in treatment in accord with unconscious plans as to how to use their therapy to change their pathogenic beliefs and solve their problems. For example, studies of the patient's testing of the therapist demonstrate that patients carry out trial actions to test their pathogenic beliefs in relation to the therapist (Kelly, 1989; Silberschatz, 1986; Silberschatz & Curtis, 1993). They become less anxious, bolder, more insightful, and more productive in treatment following "passed tests" (i. e., tests to which the therapist responds in a way that tends to disconfirm the belief being tested). These studies lend support to the idea that patients work in accord with unconscious plans to change their pathogenic beliefs.

The unconscious plan concept is also supported by studies of the effects of interpretations (Broitman, 1985; Bush & Gassner, 1986; Caston, 1986; Fretter, 1984). Patients respond with progress to interpretations that are compatible with their unconscious plans. However, they do not show progress following interpretations that are incompatible with their plans. These findings demonstrate that the unconscious plan concept predicts whether or not an interpretation will prove helpful to a patient. This finding cannot he predicted or explained except by the plan concept.

Studies investigating changes in the patient's level of insight in psychotherapy further support the plan concept. In these investigations, the level of "pro-plan" insight -- that is, insight which helps the patient go where he or she wants to go -- was determined in each session of five psychotherapies that the patient knew in advance would be limited to 16 sessions. In each case it was found that the patient had a relatively high level of insight in the intake interview and the first therapy session, then appeared to lose insight during the middle of therapy, and regained it toward the end of therapy. In each of the five cases a graph of the changes in the level of insight was best fit by a parabolic curve. These findings may be explained by the assumption that the patients each had an unconscious plan for the therapy. They unconsciously controlled their behavior in accordance with the time allotted to them, and their level of insight was a reflection of this unconscious assessment and planning (O'Connor, Edelstein, Barry & Weiss, 1994; Weiss, 1993b). They showed high insight during the first sessions in order to provide their therapists with the information that the therapists needed to help them. Their insight dropped when they began testing their therapists, and rose only toward the end of treatment, when they were no longer actively testing their therapists because therapy was coming to an end.

In a study investigating changes in the level of insight in the first 100 hours of an openended psychoanalytic case, a different pattern emerged. In this case, the initial level of insight dropped more slowly than in the time limited therapies and it rose and dropped again in several places throughout the first 100 hours of treatment. A graph of the changing level of insight was best fit by a fifthorder polynomial curve. This suggests that the patient devises a different unconscious plan in an openended treatment. In this instance, we assume that the patient, having unlimited time, allowed himself more time to carry out his initial testing. We assume that when time is not limited, a patient's plan for use of his therapy is not driven by time considerations, and thus differs from that found in the brief therapies.

Another line of research supporting Weiss's theory of psychopathology has developed through empirical studies of guilt and shame in larger populations. According to Weiss, pathogenic beliefs are inhibiting and constricting, particularly because they connect the pursuit of certain developmental goals with feelings of guilt, shame, fear and anxiety. For example, as a result of traumatic experiences in childhood, people may develop the belief that if they pursue certain adaptive goals, they will harm their parents or siblings. These beliefs may continue into adulthood when efforts to pursue these goals, or even consideration of such efforts, evoke feelings of guilt. Thus, Weiss's theory emphasizes interpersonal guilt-that is, guilt that arises out of altruism and a concern about harming others-and its importance in the development and maintenance of psychopathology. In therapy, a patient's unconscious plan often involves changing those pathogenic beliefs that give rise to irrational interpersonal guilt.

In 1989, O'Connor, Berry and Weiss initiated a series of pilot studies which led to the development of a questionnaire-the Interpersonal Guilt Questionnaire (lGQ)-designed to operationalize and measure interpersonal guilt. This instrument includes subscales of Survivor Guilt, Separation Guilt, Omnipotent Responsibility Guilt, and SelfHate Guilt. Survivor guilt is characterized by the belief that a person experiences good things at the expense of others, and that by attempting to further their own cause, they may harm others. Separation guilt is characterized by the belief that one is harming one's parents or other loved ones by separating from them, or by being different from them. Omnipotent responsibility guilt is characterized by an exaggerated sense of responsibility for the well-being of others. Self-hate guilt is an extreme form of guilt that occurs in compliance with harsh and/or neglectful parents, and is characterized by people punishing themselves with negative thoughts and feelings.

Initial pilot studies using the IGQ have shown significant correlations between these types of interpersonal guilt and various problems such as depression, a pessimistic explanatory style, negative automatic thoughts, and child abuse and trauma. In several populations studied thus far, survivor guilt and selfhate guilt appear to be especially associated with depression. And in each sample we have found an especially high correlation between survivor guilt and shame (O'Connor, 1994; O'Connor, Berry, Weiss, Bush & Sampson, in preparation). Furthermore, these pilot studies have demonstrated that a clinical group -- drug-addicted clients -- were higher than a nonclinical population in interpersonal guilt (Meehan, O'Connor, Berry, Weiss, Morrison & Acampora, in preparation). Thus, the significance of interpersonal guilt to psychopathology, suggested by Weiss's theory, is supported by these studies.

As seen in the body of work described in this brief history, Weiss's theory has, from its inception, maintained a grounding in empirical data. Both in studies of the psychotherapy process, and more recently in studies of emotion and psychopathology, the basic assumptions of this theory -- that psychopathology stems from pathogenic beliefs derived from childhood trauma, that people are striving for health, and that people think and plan unconsciously -- have been shown to be testable, and are supported by empirical research.


About the authors

Editor's note

Back to Spring, 1995 Bulletin

Psychoanalytic Research Society Home Page