Focusing Home > Gendlin Online Library > What comes after traditional psychotherapy research?
ABSTRACT: The different therapies cannot be well studied or integrated as wholes. Defining subprocesses such as free association, two-chair techniques, desensitization, and focusing (and the microprocesses that are part of these) can lead to more replicable findings and an eventual clinical integration. Seventeen other new strategies concern outcome, a new view of controls, the confounding of patient (Experiencing Scale) and therapist variables, the choice of operational indexes, and the use of research variables in clinical training. The individual-difference approach is questioned. Social and chemotherapeutic variables alter the usual definitions of psychotherapy variables. Some strategies are adapted to research in hospitals, popular psychology networks, and other settings.
Much that could happen only in a therapist's office 30 years ago has now been incorporated into many contexts of our society. Not the whole of therapy, but specific processes, are used in many settings. These more specific processes need research, and their specificity can also lead to more significant and replicable findings. There are now enough outcome studies of psychotherapy as a whole, and for this reason also, much more specific research variables are indicated for the future.
This article proposes 18 strategies for changing psychotherapy research to make such research more productive and meaningful. One central thrust of this article is to propose more specific psychotherapy research variables—not psychotherapy per se but its subprocesses and microprocesses. These smaller units cut across the different methods of psychotherapy, and they are also relevant in settings other than psychotherapy.
Strong success cases have naturalistic outcome indexes: Work and love, impossible before, now happen. There is a sense of well-being, and new dimensions open. The therapist, the patient, and questionnaires from friends agree the therapy is a success. I propose that therapists routinely tape-record, use one set of pre/post measures, and send strong successes to one bank (such as the American Psychological Association or the National Institute of Mental Health). A bank of clearly successful cases would yield clusters of correlated change measures. Not every desirable change goes with every other. Some routinely used tests have items chosen for test-retest stability under many circumstances. Change does not register on some of them. Also, what amount of change defines success on a given test? A bank of genuine success cases would address these issues.
A study's data must be compared to the general distribution before dividing "high" and "low." The usual habit ignores this. It is wrong to call the upper half of one's cases "successes" and the lower half "failures." Even without the needed bank of change measures, genuine successes can be known by patient and therapist reports. They can also be noted from the distribution of one's data. What if among 25 cases two stand out and look genuinely successful? The proper way is to compare the two with the other 23. As therapists grow in competence, more success cases will occur. Meanwhile, honor the distribution.
Suppose you test artistic ability before and after therapy. Should you predict a difference between treatment and control groups? Not at all! Predict that your measure will increase only for the successful subgroup. After all, you do not want to predict an increase for the failure cases.
Outcomes of "treatment groups" sometimes differ from the controls, sometimes not. Why are the findings so irregular? The therapy/control design does not check for whom therapy actually occurs. The "treatment group" is too often defined only by spending time in a room with someone who intends to make therapy happen. The irregular findings show that therapy intention sometimes leads to effects, sometimes not. That finding should be reported to the public, not that therapy happened and was ineffective.
To test the effects of something, one must know when it is present and when not. Therefore, a better strategy is [Page 132] to measure therapy process. From tape recordings, standard process notes, and other observations, one can measure the degree to which a therapy process happened in the sessions. High process should differ in outcome from low process (as well as from controls).
There is no agreement, however, on defining psychotherapy process. Therefore, an outcome finding would be limited to the measured process variable. But this has an advantage: Such findings are likely to be replicable.
The competing methods or schools of psychotherapy have been taken as research variables. Which is more effective, psychoanalysis or client-centered therapy? Different methods do feature different techniques, but these are only part of what therapists and patients do. All therapies have much in common. Most previous psychotherapy research studies also leave undefined how often the featured procedure occurs.
One current direction is to study just the specific techniques. I call them "subprocesses" of psychotherapy. Their short-term results may be studied in single interviews or outside of therapy. They offer a very promising research direction. Some examples are (a) free association (Bordin, 1966a, b, 1980); (b) systematic desensitization (Lazarus, 1968, 1971, 1984); and (c) The two-chair gestalt reversal. Four successive studies (Greenberg, 1983, 1984) were good models of the research direction I advocate. This subprocess can be studied alone or in sessions where a recognizable client statement marks when therapists assign this "patient task" (as Greenberg called it). The client's behavior in each pair was measured separately. The process was divided into two phases. The second phase (called "resolution") was defined on several instruments. Promising research possibilities were thereby opened.
The subprocess I know best is the client's attending to a directly sensed inward referent. This subprocess has two main measures. One of them is the Experiencing Scale (Klein, Mathieu, Gendlin, & Kiesler, 1969), applied to tape recordings. On the low end of the scale there are not even any adjectives to indicate inward reference. High on the scale, clients can be heard checking what they say against an as yet unclear physically sensed referent we call "felt sense." There is a silence. Then: "No, that's not it; what I said isn't it. It's right there, but. . . I don't know what it is, yet." More silence and bodily sensing of this unclear quality. Then: "Oh. . . yes. . . breath]. . . yes, that's it all right, whew!" Currently we not only measure that, we also teach it. As taught, it is called "focusing" (Gendlin, 1981). A checklist after instruction provides a second measurement avenue (Alemany, 1985). These variables seem to be patient processes.
Although one speaks of "patient tasks" separated from "therapist techniques," each variable implicitly includes both sides. For example, "free association" includes therapist interpretations that are possible only if associations come. Conversely, interpretations are validated and guided by further associations. Similarly, behavior therapists assign a later step only if the client did the previous one. They can reinforce only what was done. The "therapist procedure" happens only with the client behavior. If empathy is defined only as the reflective mode of responding, it depends only on the therapist but does not predict success or failure. On the other hand, if empathy is accuracy with intricate content, a therapist's empathy varies with the depth of client communication.
When gestalt reversal is measured only as a patient "task," the interaction variability remains hidden. Greenberg said (but cannot know) that therapists provide a "constant environment." In the first phase of the two-chair procedure, his clients did not differ. In the second phase, some clients became "resolvers," some did not. He did not know why. I think there were undefined differences in the first phase too, probably involving both therapist and client. The name of our "Experiencing Scale" is similarly misleading. It measures interaction, not just the client. Large time-units obviously include both sides. But even short units do not separate therapist and client. A person is a differently ongoing process depending on who else is present. Both sides are interaction indexes. Working back from phase-two "resolvers" might show phase-one differences in therapist procedure. These might then point to client differences in phase one.
The subprocess is often defined only by type. One assumes it is always done well. Instead, let us define how it is done when it is effective. Then we can also study how therapists engender or participate in these differences. An undefined mix of well-done and poorly done procedures is not replicable. In the next study, the proportions will not be the same. But to define the effective versions requires outcome criteria.
It is often, and very wrongly, said that "process" is outcome bit by bit. No, not at all. By "process" we mean whatever produces outcomes. Our process measures must not include small-scale outcomes. For example, the frequency of improvement reports during the interviews is not process—not what brought about the improvements. [Page 133] It is not surprising that eventually successful patients report more good changes during therapy, such as "I feel more whole now," or "Oh, now this problem feels resolved." Such positive statements are not hard to count. Their appearance is a useful small-scale outcome measure. Similarly, Greenberg's "resolving" is outcome. Of course, an effective process produces bits of outcome. But if we want to test whether the process produces the outcomes, we must not mix the outcomes into the process measure. We must keep the measures distinct so that if there were much process without bits of success our measure could show it.
Specifying one's own style could keep others from using a subprocess. To specify the universal, ask two questions: (a) What in this subprocess would all therapists want? (b) What cannot be had by any other method? In gestalt reversal the client is asked to sit in another chair and be, feel, and act an opposite role (e.g., aggressor, a dream figure) toward the empty chair (the client as usual). Some therapists might not want the chair drama. There are also other ways to sense two known sides of a conflict. What would all therapists want that cannot be obtained in another way? It is the spontaneous acting from a side usually experienced as alien by the conscious personality. Specify what is universally desirable and special in your subprocess.
I propose that the greater proportion of clinical research time be spent "playing in the laboratory." For social scientists, many common situations can be the laboratory. One can try out one's questionnaire items, hypotheses, subprocesses, and other small research steps with friends, co-workers, classes, or three or four research subjects. One gets instant feedback and much fuller information than in formal research.
Physical scientists tinker. They try out many things in a day and follow little leads. When something is found, complex methods of verification are instituted.
Our research methodology is excellent for verifying hypotheses. But how does one come upon good hypotheses? In our field, both students and mature researchers often invest in hypotheses they have not explored. The design cannot be changed for years. At that rate, a lifetime is not enough for the steps a physical scientist takes in a day. Education and research habits should make for constant exploration. Our training facilities should make such exploratory work possible, and it should occupy most of the time.
The exploring (and "tinkering" with subprocesses) that I advocate can be created only in exploratory studies. These are not careless versions of what must be done well to have meaning. We cannot use an N of 3 for conclusions requiring 25. Rather, for example, we get feedback on each item from three subjects. That lets us reject some items and specify others. Three subjects can vividly disconfirm a hypothesis and give new leads. Less than one day can reveal what no formal studies can show. Let us also regularly write such exploratory ministudies, make them available, and discuss them. Formal psychotherapy research will be improved through such exploration, communication, and discussion, and much more information will be systematically developed.
Fiske (Fiske & Shweder, 1966) has shown that different ways of measuring "the same trait" fail to correlate (e.g., projective tests, self-reports, and direct observations). He did not go so far, but I conclude: Different measurement-interactions are different variables. In measurement, as in therapy, only interactions exist.
This has several implications. First, if some patients predictably fail in therapy, let us not think of that as due to their individual traits. Rather than predicting prognosis, let us change therapy to be more regularly successful. For example, we had predicted that success cases would increase on the Experiencing Scale. Instead we found (Gendlin, Beebe, Cassens, Klein, & Oberlander, 1968) and continue to find (Mathieu-Coughlan & Klein, 1984; Klein, Mathieu-Coughlan, & Kiesler, 1985) that scores on the Experiencing Scale are significantly higher also in the first period of successful therapy. We might have used the finding as a prognosis measure. The challenge, however, was to get failure-predicted people to do what we found success-predicted clients doing.
We devised instructions (Focusing Manual; Gendlin et al., 1968) for teaching just that, which the Experiencing Scale had found correlated with success: attention to an unclear sensed referent. Failure-predicted people have now been raised to a success-predicting experiencing level in their therapy (Clarke, 1980). But the individual-difference habit is very strong. Our first brief Focusing Manual and the postfocusing measure were widely used to divide "focusers" from "nonfocusers." If some portion do not learn focusing from the 10-minute instructions, is that their trait? Is it not an interaction? We must keep improving our procedures to increase the proportion who learn. The strategy is not just to predict success from individual traits but to bring success about more regularly and to measure it more regularly.
Fiske's findings will also help me formulate my next proposal.
A concept—for example, "anxiety"—may be "operationally defined" as assent to items on some "anxiety scale." But nothing more than the name promises that these items relate to what the researcher had in mind, which led to the hypothesis. If no extant measure gets at [Page 134] that anxiety, it saves time to devise a new measure. Why test what can be known in advance not to correlate? Do not assume a reality of verbally defined things. It is not odd that measures having the name in common do not correlate. "Anxiety" is not one "general" thing, which must unfortunately be measured by test items or galvanic skin resistance, by ratings of facial expression, projective tests, or in other ways. One word does not ensure one thing. Language is not composed of thing-labels. The same words often make new "things." Operational measures are inherently more specific than concepts. Therefore, they cannot be chosen at random from measures sharing a name. The measure must specify exactly that which makes the hypothesis probable.
I next propose a new level of specificity: microprocess variables. Here is an example. Lazarus had thought that his systematic desensitization was a well-defined research variable. The procedure consisted of graduated imagery instructions. Subjects lifted a finger when they completed an instruction and were ready for the next.
But when he invited his subjects to describe what they actually imagined, he was shocked. He told me one such story (A. A. Lazarus, personal communication, 1971). To desensitize her sexual anxiety, one subject's first step was to imagine being home alone. The next instruction asked her to imagine her husband entering at the front door. When she raised her finger, Lazarus asked her what she imagined. She had indeed been home alone, but she was standing at the head of the stairs. Her husband came in the door all right, but with her mother-in-law. Then she fell down the stairs and wrestled with her mother-in-law, and many more scenes ensued.
Lazarus then studied many processes in systematic desensitization. Small-scale researches defined what I call "microprocesses," new variables in themselves. One was the "runaway imagery" just described (Barrett, 1970). For Lazarus, "runaway imagery" was only an obstacle, but other therapists would know how to use it. So far they have not taken up this and his other specific researches. I propose that we adopt each other's microprocesses, so that a whole therapeutic avenue (for example, imagery) can become specified.
Similarly, focusing and the microprocesses found with it can be used in any therapy. We would like to hear from anyone wishing to collaborate with us in such studies. During therapy, bits of focusing instructions are used in response to certain kinds of content. Immediate effects can be studied. Here are some of the other microvariables found in studying focusing. Would not some of these matter in any therapy method and research?
We find that about one third of the population (in Western countries) is unable to feel the stomach or chest from inside, in the ordinary way (not focusing). This inability (which we now remedy) must affect their behavior in any therapy.
Focusing requires some preliminaries: A mix of concerns is usually carried as one physical tension. We find that people do not experience a physical release of such an unsorted mix even if they turn their minds away and relax. But if they sort out these bodily sensed preoccupations, one by one, a positive body-energy appears This preliminary to focusing (we call it "clearing a space") is now being studied in its own right (Gendlin, Grindler, & McGuire, 1984).
We also find two kinds of "depth." One kind brings suggestibility and eliminates the bodily feedback on which focusing depends. The other does not.
We differentiate the known "gut feeling" (sad, glad, scared) from the as-yet-unclear referent, the "felt sense." A certain kind of attention brings the "felt sense" of a given problem. There are ways to check if it does indeed belong to just this problem.
Different people locate a felt sense in the throat, chest, solar plexus, or abdomen. We do not yet know the reason or the differences that makes. A problem-related tension in a peripheral part of the body (skin, shoulders, arms, legs) can be moved to one of the center locations. We observe but need to verify that therapeutic change-steps come more readily from the center.
We have defined ways to keep one's attention on a felt sense. There are specific ways to help a change-step come. There are specifically defined ways to keep what comes and also to defend it from superego attacks.
The felt sense makes differences in imagery (Kuiken, 1981). Focusing is easier and deeper in the presence of an attentive, totally silent person. Silent attention is a measurable variable.
But most of these microspecifics are still only definitions on tape, not studies. Now I turn to strategies of wider scope. I want to show that microspecifics are wider in application than traditional psychotherapy research.
When a process is specifically defined for research, the same specifics can be used also in training. Then they can be used again to assess the practice. The measures are separate, but the microspecifics are the same. Such specifics enable us to instruct paraprofessionals and laypeople. Research precision enables wider applications.
Therapy-derived subprocesses are important in many settings, both clinical and ordinary. They can also be studied in these settings. Psychotherapy research may take years, but the effects of microprocesses can often be measured after some minutes.
For example, we predicted that cancer patients would not be able to do our preliminary "clearing a space" or would not find the resulting positive body-energy. Of four patients, three easily did both. Later, with [Page 135] our most experienced trainer, the fourth patient did as well (Kanter, 1982). That was enough to disprove what we had thought. This microprocess was much welcomed by the patients. Such studies also allow immediate further exploration.
Microprocesses can be studied in many contexts and can improve them. For example, focusing has been found useful in mental hospitals (Ohta, 1982) and also in creative writing classes (Perl, 1980). Experiencing Scale level has been measured in groups (Lewis & Beck, 1983). It is stable for individuals' dream reports (Hendricks & Cartwright, 1978), leading us to ask next: Does the dream Experiencing Scale level increase for success cases, or after focusing training, or in other ways of working on dreams? Focusing is also useful in working with dreams (Gendlin, 1986). Aged people high on the Experiencing Scale have a higher survival rate (Sherman, 1984, p. 94). Whether focusing-teaching also has this result is now being tested.
Society at large is adopting therapy-based processes. Thirty million people belong to some "network" using one or another process. Is this quackery or is it the social change that therapists since Freud have always wanted? It depends on whether the quality can be high. This social process needs research. Microprocess research can fit these settings.
But how can these many processes ever be put together in one effective psychotherapy? Let me first take three seemingly separate dimensions:
Body, psyche, and social interaction are not three variables that we put together. They are already always together. For example, from 1950 to 1960, the drug LSD regularly produced paranoid psychosis. Today we know that this was due to the interactional conditions: The subject was tested alone in a whitewashed cell. The experimenter observed through a peephole in the door. With different interactional conditions one gets different results from the same drug administered to the same individual! We must not assume that the tested effects of one factor remain the same if one of the others is changed. The same drug taken by the same individual has different effects under different social conditions. It does not simply have more of the same effect. Caffeine has no effect on one rat alone, but if the caffeine-injected rat is put with other rats, its behavior is markedly more active.
The social-life conditions as well as chemistry must be considered. Hospitals are still full of patients, although the patients get drugs and therapy. Under those social conditions, chemotherapy has not cured them. For example, everyone agrees that an interpersonal support system may enable a reactive depression to lift quickly. In relative loneliness, depression can continue for years. Traditional research ignores this, but some social conditions do obtain for all research subjects. We do not know which subjects have which conditions.
When we combine currently separated methods, we must allow each to change the other so they can find their natural unity. For example, we cannot just add chemotherapy to psychotherapy, both in unchanged form. Chemotherapy is not one unambiguous variable. Are drug and dosage constantly readjusted to each person's unique reports in the psychotherapy? Or is the passivity and helplessness of drug regimes simply added to psychotherapy's attempt to maximize active control? We must also develop ways to handle or limit the times when the drug clouds and covers what psychotherapy needs to open. Can psychotherapy use the drug intermittently to help people tolerate what must be opened bit by bit? Both variables must change also in relation to social conditions. There is considerable research on this variable in Japan, where it has been related to the Japanese psychotherapies (Lebra, 1976). Let us allow it to reshape our psychotherapy and chemotherapy.
The hundreds of competing therapies need not be put together, because they are redundant. What are genuine differences? One is their avenues: Some work only with cognition, others only with feelings, and some only with dreams, imagery, chemotherapy, dance, body work, family, or interpersonal interaction. Behavior therapists work only with actions. These are genuine differences.
Can these avenues be put together? In the human individual, they are already together! Everyone thinks, feels, dreams, and imagines; has a body; has a family; acts in situations; and interacts with others. We therapists select one or a few of these human dimensions to work with in "our" therapy. The rest seem to make us uncomfortable, probably because we are not trained in them. But, whatever the therapy, these dimensions all occur together in a person.
We cannot combine the advertised therapies. Each claims full time. Let us go by how the avenues are already together — and change the therapies. Each avenue develops certain sensitivities and responses to certain observations. Therapists trained in a second avenue retain their sensitivities for the first. They do both in a better way. The synthesis develops naturally. For example, in operant behavior therapy, the target is often chosen at the start after a few minutes, in a crude way. But behavior therapists do not wish be superficial. If the therapists also know a feeling therapy, the same clients will choose more meaningful behavior targets. Conversely, psychotherapy clients can choose a small behavior-step for each week. Then they encounter what they avoid by always trying for one jump. Such advantages cannot be had by adding whole therapies together. If micro-processes are added, the methods improve each other.
No conceptual synthesis of orientations is intended [Page 136] here. I propose a synthesis through a renewed clinical research strategy.
Alemany, C. (1985). A short version of the post-focusing check list. Focusing Folio, 4, 89-97.
Barrett, C. I. (1970). Runaway imagery in systematic desensitization and implosive therapy. Psychotherapy: Theory, Research and Practice, 7, 233-235.
Bordin, E. S. (1966a). Personality and free association. Journal of Consulting and Clinical Psychology, 30, 30-38.
Bordin, E. S. (1966b). An experimental analogue of the psychoanalytic situation. In L. A. Gottschalk & A. H. Auerback (Eds.), Methods of research in psychotherapy. New York: Appleton-Century-Crofts.
Bordin, E. S. (1980). Of human bonds that bind or free. Presidential address, Society for Research in Psychotherapy Research, Pacific Grove, CA.
Clarke, D. B. (1980). EXP levels and focusing with therapy patients. Unpublished doctoral dissertation, Georgia State University, Atlanta.
Fiske, D. W., & Shweder, R. A. (1966). Pluralisms and subjectivities in social science. Chicago: University of Chicago Press.
Gendlin, E. T. (1981). Focusing. New York: Bantam Books.
Gendlin, E. T. (1986). Let your body interpret your dreams. Wilmette, IL: Chiron.
Gendlin, E. T., Beebe, J., Cassens, J., Klein, M., & Oberlander, M. (1968). Focusing ability in psychotherapy, personality, and creativity. In J. Shlien (Ed.), Research in psychotherapy (Vol. 3). Washington DC: American Psychological Association.
Gendlin, E. T., Grindler, D., & McGuire, M. (1984). Imagery, body, and space. (With Grindler, D., & McGuire, M.) In A. A. Sheikh (Ed.), Imagination and healing (pp.) New York: Baywood.
Greenberg, L. S. (1983). Toward a task analysis of conflict resolution in gestalt therapy. Psychotherapy: Theory, Research and Practice, 20, 190-201.
Greenberg, L. S. (1984). A task analysis of intrapersonal conflict resolution. In L. N. Rice & L. S. Greenberg (Eds.), Patterns of change (pp. 67-123). New York: Guilford.
Hendricks, M., & Cartwright, R. D. (1978). Experiencing level in dreams: An individual difference variable. Psychotherapy: Theory, Research and Practice, 15, 292-298.
Kanter, M. (1982). Clearing a space with four cancer patients. Focusing Folio, 2, 23-37.
Klein, M. H., Mathieu, P. L., Gendlin, E. T., & Kiesler, D. J. (1969). The Experiencing Scale: A research and training manual. Madison: University of Wisconsin Extension Bureau of Audiovisual Instruction.
Klein, M. H., Mathieu-Coughlan, P., & Kiesler, D. J. (1985). The Experiencing Scales. In W. P. Pinsof & L. S. Greenberg (Eds.), The psychotherapeutic process: A research handbook. New York: Guilford.
Kuiken, D. (1981). Self-consciousness as a component and correlate of focusing ability. In E. Klinger (Ed.), Imagery: concepts, results, and applications (Vol. 2). New York: Plenum Press.
Lazarus, A. A. (1968). Variations in desensitization therapy. Psychotherapy: Theory, Research and Practice, 5, 50-52.
Lazarus, A. A. (1971). Behavior therapy and beyond. New York: McGraw-Hill.
Lazarus, A. A. (1984). The specificity factor in psychotherapy. Psychotherapy in Private Practice, 2, 43-48.
Lebra, T. S. (1976). Japanese patterns of behavior. Honolulu: University of Hawaii Press.
Lewis, C. M., & Beck, A. P. (1983). Experiencing level in the process of group development. Group 7, 18-26.
Mathieu-Coughlan, P., & Klein, M. H. (1984). Experiential psychotherapy: Key events in client-therapist interaction. In L. N. Rice & L. S. Greenberg, (Eds.), Patterns of change. New York: Guilford.
Ohta, T. (1982). A psychological method where the East and the West meet. Focusing Folio, 2, 1-10.
Perl, S. (1980). Understanding composing. College Composition and Communication, 31, 363-369.
Sherman, E. (1984). Working with older persons. Boston: Kluwer-Nijhoff.
