|
|
| |
On the Clinical Psychoanalytic Theory and Its Role in the Inference and Confirmation of Particular Clinical Hypotheses
Commentary by Robert R. Holt
It seems likely that most of the papers chosen in this series earn their landmark status because of important new clinical advances or theoretical innovations with rather direct implications for practice. Since the paper I have chosen deals with the clinical theory, one might assume that it, too, lends itself to rather direct use in the analyst's daily work. While its clarification of theory may help in that way, in my view it is chiefly notable for its long-range implications for psychoanalytic research. In the long run, however, I believe it will make such research more helpful to the working clinician than has typically been the case.
Benjamin Rubinstein, who died in 1989, was a practicing psychoanalyst of considerable clinical experience and skill. In that respect, he did not differ from scores of his contemporaries. But he stood out among them all in one critical way: he devoted several decades to the intensive study of the philosophy of science, and a good many of his last years to directing a cooperative empirical investigation of the processes of making and testing clinical inferences. As a result, he was uniquely qualified to address the scientific grounding of psychoanalysis.
I see four ways in which his 1975 paper breaks new ground. First, in it Rubinstein restates the major propositions of the clinical theory without using any metapsychological terminology-a difficult feat indeed. Second, this paper presents a clear, hierarchical organization of the theory. Third, it brings out one fundamental point about the clinical theory, which I had never seen made by anyone else: he focuses our attention on the probabilistic nature of its propositions, a fundamental property it shares with much social and behavioral science, with notable but usually overlooked implications for research. Fourth, here he puts his finger on one other way in which psychoanalytic theory differs from the model of physics, implicitly assumed in standard methodology: the fact that it deals not with specific actions but only with classes of them.
It will probably surprise many readers to discover that, stripped of its pseudo scientific language, most of the clinical theory resembles common-sense psychology. Quite independently of Rubinstein, however, I had come to the same conclusion in a contemporary, much less ambitious effort to disentangle all the theoretical assumptions underlying the case of Dora from their metapsychological formulation (Holt, 1975). In this new guise, however, the clinical theory proves to be quite testable.
The next of Rubinstein's major contributions in the present paper, his hierarchical organization of the propositions of the clinical theory, is a boon to anyone who wonders just what the clinical theory consists of. That should help any clinical practitioner who wants to be clear about the theoretical underpinnings of his work, but it also helps to give basic research a coherent direction. For if this observational heart of psychoanalysis is to guide research in such a way as to make it cumulative, it has to be ordered and systematized. That task, in turn, requires insight into what the clinical theory's first principles are, and a rational scheme for arraying its bewildering proliferation of propositions into manageable and comprehensible categories. So far as I am aware, no one else has undertaken this daunting but necessary task.
Rubinstein begins by specifying the three types of general clinical hypotheses: the motivational, situational, and genetic hypotheses, plus a fourth class of miscellaneous general clinical hypotheses; for example, a set dealing with unconscious fantasies. Then there is a large class of special clinical hypotheses. "Some of them represent more or less extensive generalizations of particular clinical hypotheses," he wrote. These special hypotheses too may be subdivided into motivational, situational, and genetic subtypes. Examples of the special motivational hypotheses are the familiar strategies of resolving conflict which we usually call defense mechanisms. And, "the theory of the Oedipus complex and of its dissolution or persistence in an unconscious state... represents a set of special clinical hypotheses, motivational, situational, and genetic." Finally, lowest in the hierarchy of abstraction come the specific clinical hypotheses: those that apply to particular persons. Interpretations offered to patients are examples of such hypotheses; so are generalizations about particular patients in written case histories.
The third major point he put this way: "the more we subject them to scrutiny, the more evident it becomes that most psychoanalytic statements are probabilistic." I know that I had never quite realized this property of psychoanalytic theory before reading his paper, though I greeted this last statement with a shock of recognition: it seemed very similar to a central point of a seminal paper by Paul Meehl (1978). In Meehl's terminology, one reason for the slow progress of soft psychology (which includes clinical, social, and personality psychology) is that the regularities it deals with are statistical; therefore, rigorous, nomothetic laws are impossible.
Often we conceal from ourselves this probabilism by stating generalizations overconfidently. Freud habitually did so; he tells us what "the boy" does in his development, not what boys of a particular type are likely to do. Other psychoanalytic authors uncritically follow suit. One embarrassing result is constant implicit contradiction: Generalizing from his or her clinical experience, each author asserts that the development of "the boy"-that is, implicitly, of every boy-proceeds in quite different ways, which are often mutually incompatible; likewise for "the girl." Not only is that confusing, but because of it we fail to profit from the real discoveries clinicians make. If they were taught from the beginning that all clinical generalizations are probabilistic until proved otherwise, they could make use of their precious opportunity to observe special behavioral patterns and begin to formulate hypotheses about regularities in them with specifiable parameters.
It may not be evident that Rubinstein's most general clinical hypotheses are probabilistic, partly because he did not formulate them as assertions. For example, the hypothesis of motivational determination: Doesn't that mean that all behavior is motivated? Yes; as Rubinstein puts it, "all activities in which a person engages...are motivated even if on the face of it they may not seem to be." But Rubinstein does not reject the evidence that some parapraxes, for example, may be adequately accounted for by nonmotivational determinants. He adds, also, that the behavior of dreaming does not occur because of wishes. We get here into rather sticky matters of definition; some people maintain that a non-motivated slip or accident is not behavior; it's an event but not an activity. Thus, if someone stumbles in the dark and breaks an ankle, we need not assume that it happened because of a repressed self-destructive wish. Even though it is plain that the activity of walking was motivated, the stumble may not have been unconsciously directed; a cigar is sometimes only a cigar, and an accident is sometimes truly accidental. Hence, even such an apparently universal statement as that all behavior is motivated can be seen to have a probabilistic aspect: not every aspect of any act or every one of its outcomes is necessarily determined by wishes, fears, or defenses against them; it just may be.
In another respect, Rubinstein noted, psychoanalytic theory does not lend itself to the standard experimental method, in which logical manipulation of theoretical propositions generates specific predictions. Instead, it permits us only to predict that one or more of a large class of actions will occur. (A little thought shows that this is another aspect of the basically probabilistic nature of the theory.)
Let's take a concrete example. Suppose a psychoanalyst uses the classical technique, being quite passive and permissive and encouraging his patient to say anything that comes into her mind. The patient, however, constantly assumes that she is dealing with a harsh taskmaster who demands certain kinds of information and who becomes angry when she deviates from his strict rules. The analyst forms the hypothesis that the patient is acting transferentially, out of a disposition to perceive authority figures as rigid, unempathic, and demanding. Here is a specific clinical hypothesis, evidently based on several general and special clinical hypotheses. It guides the analyst's treatment by setting up several expectations. Some are genetic: the analyst implicitly or explicitly postdicts that evidence will emerge that in her childhood the patient had a series of experiences with a father figure who treated her in the indicated fashion. Other expectations deal with the patient's contemporary life: she will report difficulties in dealing with authority figures, like teachers or bosses.
Notice two ways in which these predictions lack specificity. First, the theory permits the specification only of a class of father figures in her past and present, and second, it does not zero in on any particular interaction between those persons and the patient. Until all the details of the case history are known, the analyst has no way of knowing whether the original authoritarian figure was her biological father or some other person in the paternal role. The details of specific incidents that galled the girl and shaped her transferential anticipations about males in superior roles cannot be foreseen either. Nevertheless, it is possible to describe typical classes of persons and actions precisely enough to enable reasonably sophisticated judges to select relevant episodes from a transcript of this patient's analysis, and to do so with satisfactory reliability.
In an earlier discussion of Rubinstein's analysis of the clinical theory (Holt, 1989, Chapter 13), I drew the following implications for research. We should build on his promising start in collecting and ordering the propositions or hypotheses of the clinical theory, searching the literature for clinical generalizations. They should be freed from their usual admixture of metapsychological concepts and expressed, so far as possible, in terms of observable behavior (using that term broadly to include emotions, thoughts, images, and verbal reports of all kinds). I can imagine teams of workers in different locations going over all of Freud's clinical writings and extracting not only his explicit clinical generalizations but those that are implicit in the way he interpreted the patient's productions. Ideally, then, the teams could get together, comparing and discussing their findings and trying to reach consensus. Others, meanwhile, might do the same thing with their favorite contemporary clinical writers from whatever school. As a way of organizing the task, I recommend sorting the resulting collections of propositions into the hierarchical system set up by Rubinstein.
Next comes the attempt to confront theory with data. The probabilistic nature of the clinical theory implies asking, for each hypothesis, just how probable is the behavior described? How often does it occur in people of various types, and under different specified conditions?
Then I took a concrete example of an intermediate proposition of the clinical theory, one of Rubinstein's special motivational hypotheses: When insulted, a person is somewhat likely to feel hurt. I imagined a program of research to collect comprehensive information about the reactions of all kinds of people to all kinds of insults. While there was some value in this exercise, I was taken aback when I began to contemplate the end result. It was nothing less than a set of actuarial tables, telling just how frequently people of specified types in specified types of situations behaved in specific ways. 'Wait a minute,' I said to myself, 'that looks more like a set of data than a verified theory. And why should psychoanalysts be interested in it?'
On a little reflection, I realized that of course one major result of any program of research is a set of data-the findings-from which one reaches conclusions about the theory. What struck me as odd, since I too was raised on a non-probabilistic set of assumptions about science, is the fact that the table of results itself constitutes the conclusions about the theory: not that the proposition is true or false, but just how far it is true under various circumstances.
A further part of the problem was my having begun with one of those propositions from commonsense psychology, which summarize the ways people usually act. Rubinstein began the present paper by noting that we ordinarily do not ask for a scientific explanation of what seems obvious, as when people behave in perfectly normal, expectable ways. Surely Freud was not interested in developing a total psychology, a theory of ordinary as well as unusual behavior. As a clinician, he was concerned with the abnormal, surprising, and unexpected things people do, which challenge ordinary explanations and call for new theories. There is value in making explicit what he took for granted, the psychology of normal behavior, but a research program to investigate that might logically be left to social and personality psychologists. They are traditionally concerned with the majority who follow expectations. Psychoanalysts still have plenty to catch their interests, however: the deviant minority.
If we follow the logic of probabilism to its end, it appears reasonable not to throw away any of the findings. In this kind of research, we are less inclined to accept or reject hypotheses than to keep score on them, to build up statistics on how frequently (or infrequently) predicted events occur in specifiable populations and under given conditions. The psychoanalytic researcher, I am suggesting, would look for, and then study carefully, instances where the puzzling, more probably neurotic behavior appears, e.g., when the insulted person does not feel hurt. To begin with, what kind of person, under what circumstances? If the raw data of recorded psychoanalyses are available on them, the findings will become more specifically psychoanalytic.
Even when testable hypotheses derived from the clinical theory are repeatedly confirmed to a satisfactory degree, the theory contains so many theoretical terms and is so loosely coupled that the high-level, general hypotheses cannot be definitively confirmed by the use of psychoanalytic data only. One telling point is that the same general hypotheses we wish to test must be assumed to be valid so that we can formulate the testable predictions themselves.
There are two possible conclusions, Rubinstein wrote (in a part of the paper not reproduced here). On one hand, we can treat the set of general hypotheses as axioms or presuppositions, not attempting to test them. The resulting "theory may be regarded as essentially a system of rules of interpretation, a hermeneutic system...[which is] neither falsifiable nor confirmable as such" (his emphasis).
If they have been framed carefully, however, the general hypotheses are in principle "confirmable neurophysiologically." When we become able to make such tests, using nonclinical data, then-and only then-we will have a valid means to decide among rival clinical theories.
The decisive steps forward in science do not come from amassing data and summarizing them as empirical generalizations. Someone (preferably a genius) has to examine the a body of factual knowledge and discern within it a simple, ideally mathematical, way of expressing the order contained therein. While we wait for our Newton, however, we have to make it possible for him to work by amassing the necessary bodies of knowledge in the way Rubinstein has pointed out.
For additional reading:
Eagle, M. N. (1985). Benjamin B. Rubinstein: Contributions to the structure of psychoanalytic theory. In Beyond Freud: A study of modern psychoanalytic theorists, ed. J. Reppen. Hillsdale, NJ: Analytic Press. Also in Holt, 1997, pp. 23-42.
Holt, R. R. (1975). The past and future of ego psychology. Psychoanalytic Quarterly, 44(4), 550-576. Also in Holt, 1989, Chapter 8, pp. 199-217.
Holt, R. R. (1989). Freud reappraised. New York: Guilford.
Holt, R. R. (ed.) (1997). Psychoanalysis and the philosophy of science : The collected papers of Benjamin B. Rubinstein. Madison, CT: International Universities Press. Also in Psychological Issues, Monograph No. 62/63.
Meehl, P. E. (1978). Theoretical risks and tabular asterisks: Sir Karl, Sir Ronald, and the slow progress of soft psychology. Journal of Clinical and Consulting Psychology, 46, 806-834.
If you have comments on this course that you want to share with other section members, please visit our listserv.
|
|
|
