A Neglected Problem About Clinical Evidence
By Dale Boesky, M.D.
Outcome studies demonstrating the effectiveness of psychoanalytic treatment are rightly regarded as a very useful corrective to the critics of the usefulness of psychoanalytic treatment. The growing number of these studies demonstrating the highly useful effect of psychoanalysis is to be applauded. But there is a paradox about this particular form of research that deserves further consideration. When evaluating the outcome of an individual case that may have required 1500 hours or more of treatment it is easier for the independent evaluators to agree on the effectiveness of the entire treatment than it would have been for them to reach an agreement about the major issues that should have been addressed in many of the individual sessions during that treatment. At least I believe that this is a reasonable inference based on my experience during some four decades of participation in a variety of study groups where individual sessions were considered in extensive detail and where it was regularly the case that serious differences emerged among colleagues within the “same” theoretic model of comparable experience and ability. Why should it be that we can agree that a treatment was successful on the basis of a macro-evaluation of many hundreds of sessions but have so much trouble agreeing about any one particular session? Isn’t that just another way of saying that we know psychoanalysis works but we don’t know how? Our ability to know how it works would be much advanced if we had a better methodology for single hour process research. It should be more worrisome than it seems to be that we can seem to agree about the final outcome of hundreds of sessions without being able to reliably agree on the optimal clinical interpretations for many of the individual sessions comprising any one psychoanalytic treatment.
This is a neglected problem about our methodology for evaluating clinical evidence. There are many diverse types of psychoanalytic research but only a few authors deal with this problem which is called process research, [e.g., see Waldron (1997); Bucci, (1997); Luborsky, (1996); and Spence, D., Mayes, L, and Dahl H. (1994)]. And these are usually reports of empirical research rather than single case studies reported by the treating analyst.
It is often observed that most of our visible theoretic disputes are on the higher levels of theoretic generalization and abstraction. But we tend to ignore the very important confusion in our understanding of clinical events that are “experience near”, on the lowest levels of theoretic abstraction. When a psychoanalyst attempts to “understand” (or to organize) the raw data of the patient’s associations she or he is integrating selected portions of these associations with selected portions of the psychoanalyst’s theoretic expectations. The criteria that the analyst utilizes are correspondence criteria. The term “correspondence” denotes the match between the data and the theory, not the match between the theory and the “truth”. It is well known that we lack a canon of inference. That is another way of saying that we do not yet have a consensual agreement about our correspondence criteria.
Since this problem is integrally related to a clear understanding of the pivotal role of the diverse levels of theoretic abstraction in psychoanalysis, no matter which model, it will be useful to remind you of Waelder’s classic definition of these levels. What follows is a mixture of quoting and paraphrasing of Waelder’s (1962) views.
1. In speaking of psychoanalysis…one can distinguish between different parts which have different degrees of relevance. First, there are the data of observation.
2. These data are then made the subject of interpretation regarding their interconnections and their relationships with other behavior or conscious content. This is the level of clinical interpretation.
3. From groups of data and their interpretations, generalizations have been made, leading to statements regarding a particular type such as, e.g., a sex, an age group, a psychopathological symptom, a mental or emotional disease, a character type, the impact of a particular family constellation, or of any particular experience, and the like. This is the level of clinical generalizations.
4. The clinical interpretations permit the formulation of certain theoretical concepts which are either implicit in the interpretations or to which the interpretations may lead, such as repression, defense, return of the repressed, regression, etc. This is the level of clinical theory.
5. Beyond the clinical concepts there is, without sharp boundaries, a more abstract kind of concept such as cathexis, psychic energy, Eros, death instinct. Here we reach the level of metapsychology.
This will make it clearer that my comments are primarily devoted to the first two levels of abstraction in Waelder’s schema. It is at the second and third levels, the levels of clinical interpretation and clinical generalization that we have so much confusion about our correspondence criteria. It is at the higher levels of abstraction that we have our most visible disputes between the various theoretic models.
It is commonly accepted that one of our major problems in comparing the competing claims for alternative views of a single session is that we have no canon of inference that will guide our evaluation of clinical data. A canon of inference would not imply a manual of standards, but would represent an evolving refinement of the correspondence criteria underlying clinical interventions. The assumption has been that if we some day developed such a methodology that we could substantially improve the present dismal state of our evidential support for our theoretic and clinical claims. That is because we could then state explicitly just what it was the patient said or did that was the basis for a specific intervention. The reasons for our failure to develop such correspondence criteria (another term for a canon of inference) are enormously complex and have been discussed in a very large literature by friends and foes of psychoanalysis. I wish to suggest here a conjecture about just one of the many reasons why this problem has remained neglected. The relative absence of attention to this problem may represent an institutionally sanctioned agreement to protect ourselves from the narcissistic wounds of admitting our ignorance and uncertainty. The polemics of disputes on higher levels of abstraction between models allows for the projective relief of having allies and enemies. In sharp contrast to these group phenomena are the wounded feelings and anger of many analysts (myself included) when a cherished demonstration of theoretic skill is challenged by an alternative explanation. The heady altitudes of the highest levels of theoretic abstraction protect the individual analyst presenting clinical material from the flak of dissent.
These problems have remained vexing since the very beginning of psychoanalysis. The disagreements among analysts have fortunately become more visible as a consequence of the proliferation of so many diverse theoretic psychoanalytic models in the past twenty years. It is commonly recognized now that it is a serious error to say that any member of a group of psychoanalysts who are adherents of a theoretic school necessarily agrees across the boards with all the other members of that group. We analysts are far more eclectic than that. We know that it would be a mistake to lump together all adherents of any one theoretic school and to homogenize their views. There are major differences between classical and contemporary Kleinians and between some Self Psychologists and others. Certainly there are important intra-model differences as well as inter-model differences. But it is the inter-model differences that are the most visible basis for the majority of our disputes. And the most familiar intra-model disagreements also tend to be on the higher levels of abstraction as illustrated by the differences in the recent writings of Irwin Hoffman and Jay Greenberg (Greenberg, J. 1995)
So there is a selective and partial uniformity of agreement that does indeed exist among the members of any “school”, but such agreements (as well as the noted disagreements) tend to occur at the higher levels of theoretic generalization. Just as there is no monopoly on truth or knowledge in any of our numerous theoretic models, so there is no monopoly on intellectual honesty or scientific curiosity in our various diverse theoretic groups. For that reason if one were to attend a clinical presentation of process material no matter what the theoretic group sooner or later disagreements would arise about the most advantageous way to understand the patient. Even in the days of the famous supremacy of the ego psychology model there were widespread disagreements about clinical material in study groups, in our literature and in scientific meetings. Then just as now if an analyst presented detailed process notes to colleagues in a study group or in a scientific meeting there were “spirited” disagreements about how best to understand the clinical material. The much touted theoretic unity of that era was at the higher levels of abstraction, e.g., about the unconscious and the oedipal complex, it was not about what was going on in an hour with a patient. And these very abstract theoretic agreements were very useful for political solutions to very complex problems (Schafer, 1990). This can be compared to the platform statements that are issued by both the Democratic and Republican parties every four years. The lofty principles espoused in those platforms can be espoused by a broad coalition of members of each party who actually would be at each others throats if the practical application of the abstract platform statements were to be discussed.
The misunderstanding I wish to clarify is this. Our current theoretic pluralism is widely viewed to emphasize the disagreements between analysts at higher levels of theory and masks the universal inter-model and intra-model neglect of our confusion at the lower levels of theoretic abstraction.. And this is where the canon of inference is most visible by its absence. We actually do know a great deal more than is commonly recognized about how we are applying our correspondence criteria. This can successfully be taught in our Institutes but for the most part is not (Boesky, 2002). If it can be agreed that this is an important problem it will be useful to stimulate discussion about why it has been neglected and how it can be remedied.
Boesky, D. (2002) Why Don’t Our Institutes Teach the Methodology of Clinical Evidence? Psychoanal. Q. 71: 445-476
Bucci, W. (1997) Patterns of Discourse In “Good” And Troubled Hours: A Multiple Code Interpretation. J. Amer. Psychoanal. Assn., 45:155-187
Greenberg, Jay R. (1995) Reply to Discussions of Oedipus and Beyond. Psychoanal.Dial., 5:317-324
Luborsky, L. (1996) Theories of Cure in Psychoanalytic Psychotherapies and the Evidence for Them. Psychoanal. Inq., 16:257-264
Schafer, R. (1990) The Search for Common Ground. Int. J. Psycho-Anal., 71:49-52
Spence, D., Mayes, L, and Dahl H. (1994) Monitoring the Analytic Surface. J. Amer. Psychoanal. Assn., 42:43-64
Waelder, R. (1962) Psychoanalysis, Scientific Method, and Philosophy. J. Amer. Psychoanal. Assn., 10:617-637
Waldron, S. (1997) How Can We Study The Efficacy Of Psychoanalysis? Psychoanal. Q., 66:283-322