This is a report on a five-year research program [***] at the University of Wisconsin and Mendota State Hospital. The research was originated and sponsored by Carl Rogers. It attempted a number of radical things and employed research rigor and measurements that are extremely rare in the field of psychotherapy.
The difficulties and the many discouraging aspects of doing psychotherapy with schizophrenics as reported by others were also found by us. We did not walk in and empty out the hospital. But we did have some successes in the therapy. Similarly, we found the research difficulties almost as hard as we were told we would. The attempt at rigorous measurements of psychotherapy in a hospital was said to be impossible. We found that it was not impossible, but, at times, it was extremely close to that.
Our schedule called for testing each patient regularly, but the human condition plus the conditions of a hospital prevent anything from being quite regular. Whereas we wanted to test the patients every six months, we were fortunate to obtain at least one good instance of adequate testing early in therapy (not always in the beginning) and at least one instance late in therapy. Hence we can compare these patients early and late in therapy.
We had a no-therapy control group. That is also a very difficult matter. How can one arrange to prevent therapy for the control group without making them worse by such arrangements? If one makes the control group worse, naturally the psychotherapy group will look better. What will then seem like a good finding will mean little. Ideally, one wants a group of con-
[*] Presented at the Second National Meeting of the Association for the Advancement of Psychotherapy, St. Louis, Mo., May 5, 1963.
[**] Department of Psychology, University of Chicago.
[***] The project was supported by the National Institute of Mental Health, directed by Carl R. Rogers, Ph.D., Eugene T. Gendlin, Ph.D. and Charles B. Truax, Ph.D., at the University of Wisconsin Psychiatric Institute and Mendota State Hospital, Madison, Wis.
[Page 5]trols with whom nothing like therapy is done so that one can later estimate whether improvements in the psychotherapy group are really due to psychotherapy, or whether they happen just as much by spontaneous remission without therapy. But in our treatment-oriented hospital, group therapy is common and individual work not infrequent. To isolate the control group in some special way would have singled them out for worse treatment. So we merely tested them and otherwise let them get whatever the hospital gives. Most controls had group therapy, and two were in individual therapy.
Our main concern was to test Rogers'  original hypotheses—that the patients will get better to the extent that the therapist is genuine (really himself as a person), to the extent that he is empathic (understands what is going on in the patient), and to the extent that he shows the patient unconditional positive regard (likes the patient regardless of all of the patient's difficulties and bad feeling). I will rename these three conditions realness, understanding, and liking. We took on the task of measuring the extent to which the therapist is really himself, understands and likes his patient. This involves measuring in-therapy behavior itself, not only before-and-after tests outside of therapy. We are new at measuring psychotherapy itself, as it goes on during the interview hours. But, if psychotherapy research is to make real progress, one has to begin to define and measure what occurs in the psychotherapy hours. So by realness, understanding, and liking of the therapist, we meant what he does during the psychotherapy hours.
In addition to these three aspects of the therapist's interview behavior we are also measuring the patient's behavior. To what extent does the patient behave in ways that indicate an involvement in an experiential psychotherapy process? Beginning with Rogers' Process Scale [2, 3] we have refined an Experiencing Scale with indices of the extent to which the patient avoids feelings and felt meanings, and uses verbalizations that do not involve his experiencing, externalizes, and the extent to which he is really involved in a psychotherapy process, his reactions and phrasings being fresh expressions of experiencing .
We hypothesized that therapists who measured highest on realness, understanding, and liking would have patients who not only showed the most improvement at the end, but who during the course of therapy would show the highest and most increasing degree of this experiential therapy process.
To put it all in one sentence: Will the cases that show highest therapeutic behavior by therapists also be those that show the highest degree of patient therapeutic behavior—and will these be the patients who are most improved at the end?
We developed quite specific rating scales to apply to the tape-recorded [Page 6] interviews. Except for interviews in seclusion rooms or on the ward, we tape record all the psychotherapy hours. We have walls lined with stored tape recordings. We randomly took from each of the tape recordings little four-minute bits and re-recorded them onto small separate, coded reels. We instituted some little researches to determine whether we should use four minutes, ten minutes, or two minutes. Four minutes was most reliable and valid [5, 6]. Despite the long psychotherapy case and the full hour's psychotherapy tape, listening to a four-minute excerpt gives a surprisingly definite impression of what sort of relationship the two people have. Of course, we use very many such excerpts.
To define the therapist and patient behavior which we measure on these tapes, we first began with simple general descriptions. For example, we had one general description of how a therapist is genuine. He does not use false fronts, does not present only a professional image, a mere exercise of his doctoral role. Instead, he is the person he really is, responding as a human being to the patient. Then, below such a description we had numbers: 1, 2, 3, 4, 5, 6, 7. Number 1 was "very little," number 7 was "very much." That was our rating scale. We asked raters to rate this one general concept. That is how we began.
We keep our raters separate so that they do not talk to each other. We code all these bits of tape recording so that their only identifying aspect is a code cipher such as "segment #3245J." The rater does not know if this is an early bit or a late bit, or if this is thought to be a success case or not. If these raters, rating separately, give approximately the same ratings (applying a good deal of higher mathematics to estimate the extent to which they are rating the same way), then we know we are measuring something. If they disagree to much, then we know we are failing to measure anything.
Raters did not show sufficient agreement until we moved from this very vague rating scale, to a much more exact specification of exactly what tape-recorded observations were to be given exactly what number on the scale. Here are some examples of our many specific definitions of various scale points: Stage 2: The therapist sounds very uninvolved as though nothing that happens here could possibly affect him. Stage 3: There is artificiality and stiffness in the therapist's behavior, indicating that he is having trouble with the patient. Stage 4: He sounds quite personal, but this is a style that always remains the same throughout; a stereotyped repetitious professional style controls his response. Stage 7: In responding to the patient there are spontaneous expressions of the therapist's. Of course he responds to the patient's meanings, but what he says is a direct expressive outcome of his own experiencing of these meanings. Stage 8: In addition to saying whatever seems necessary according to his method of therapy, he also intends to, [Page 7] and does, express some of his own inner processes, steps, and feelings at the moment. And finally Stage 9: There are times when he "rolls out" with some very potent feelings of personal involvement or closeness or care. This series of examples gives some of the rating scale definitions of what we expect the raters to observe on these tape recordings.
Our research has brought science about half-way toward really defining observations well enough so that rating scales become scientific instruments rather than subjective impressions of the listener, as they were before. We have not yet reached the point where these observations are so clearly defined that if you listen to a tape recording, you can simply say "I heard that behavior" or "I didn't hear it." We may very well reach that point of precision in another decade. If we do, we will then have scientifically defined observation. Even now, the scale definitions—not subjective impressions, are being measured. We know that because the scales can be used reliably by undergraduates who know nothing of psychotherapy, clinical language, or concepts. Such raters now show more agreement in their ratings than do sophisticated clinicians (who usually project their own subjective views and attitudes into what they hear). This indicates that we are succeeding in precisely describing observable events so that they can be picked out on these tape recordings. On the other hand, we are succeeding only to the extent of reliabilities of .6 to .8—not yet precise enough.
The prediction was that the more the therapist behaves with realness, understanding, and liking, the more the patient will get better—as I originally called it. By this we meant that on our before-and-after tests the patient would show improvement. Outcome tests, on the whole, do indicate whether the patient has gotten better or not. However, the various tests do not always agree with each other. An individual may improve on one test and not on another. Furthermore, even such tests correlate somewhat better with each other than with the judgment of the therapist, the patient, and the hospital staff. These tend to be different. A somewhat different group of patients improved according to the judgment of the ward staff than those who improved on the tests, or those of whom the therapists call improved. But, there is considerable overlap, fortunately. Hence, we are able to define fairly clearly which group of patients are successful and which are not.
It is already certain that the patients did a great deal to us. I might say that our own improvement has been remarkable. It is well known that schizophrenic patients have a way of changing their therapists. They give their therapists powerful experiences, producing growth in them. I think also that if the patients fail to change us, if we hold up our professional [Page 8] images to such an extent that we do not let them change us, then I doubt very much if we are giving them the kind of relationship in which they can get better. Thus, for example, genuineness has changed for us from the mere absence of a false front, to a very active, self-expressive mode of making an interaction [7, 8, 9, 10, 11].
It used to be common that if the client was silent and had nothing to say, the client-centered therapist also sat in a receptive, respectful silence, and waited for the inner process of the client to produce something. I still advocate silence when it is the kind where a person (who has been talking) goes deeply into himself. I still feel strongly that psychotherapy needs periods of silence. These permit concretely felt depth. But, with schizophrenic people we met a different kind of silence, one in which a patient is simply cut off, in which little is happening, clearly an impasse. Here the patient does not know what to do, and I do not know what to do. In that kind of silence I have now learned always to do something. I may talk about the feelings I have. I may talk about being puzzled or wishing I heard something from the patient. I may express a very personal sense of what I wish were happening, what I wishfully imagine he might be doing; I may wonder aloud whether he might be thinking about this or that; I might tell him that I know it is hard to talk about some things. Perhaps that is some of what he is feeling.
I do not say many different things at one moment. I stand or sit for a few silent minutes. I have many feelings. I express one feeling that seems all right to express. A few silent minutes later, I may again say something of what is going on in me. I find that when I am not getting anywhere with a patient, quite a lot is going on in me . I can be very frustrated, I can be very concerned to do something with him, and I can feel very badly that I do not know what to do. I can be very curious, personally interested. I can get quite angry because so much of my own welcoming for this patient is wasted. He is not getting any of it. He has no sense of my waiting for him. All of this makes a very rich "stew" from which I can always take something that feels quite real to me and might give him a sense of what kind of a situation this is. I am perfectly willing, at first, for the patient not to know how to take that, what to do with it, not to be able to respond overtly. I know that when I express myself in this fashion, it constitutes a relationship we both have, even if he remains silent. If my expressions are quite personal, that kind of interaction is going on for both of us, even though he can say nothing. I have the power all by myself to make the kind of relationship I want to happen. I used to wait for it somehow or other to happen, but now I find that nobody can really stop me from making it happen.
We have quite a number of tape recordings in which a very, very moving [Page 9] interaction is going on, but you never hear the patient speak. Instead of T (therapist) C (client)—T C T C T C T C—the typed transcription is T - T - T - T. The entire tape is silences and therapist speaking. One can see how these patients have changed our basic notion of genuineness. The principle is the same as it was: the therapist is the real person he is, rather than a professional front or a professional model. The principle is the same but we behave quite differently, in an active, positive, self-manifesting interaction-making.
A similar development occurred with "empathic understanding." This used to mean, and still means, responding to what is going on in the client. But we used to limit ourselves to what the client said or conveyed, what we knew to be going on. Now we find we can respond to "what goes on in the client" even if he does not say anything. Of course, we may not know what it is we are responding to, but that does not mean we cannot respond! "You look like you feel very sad. I wonder, are you feeling sad? You haven't said anything, so of course I don't know." Or, I can say "I know you are having a rough time" (because I do know that). That is saying much, but yet it is responding to, referring to, what happens in him. I can always refer to his concrete feelings, although crudely, roundly, vaguely, stupidly, sentimentally, imaginatively, foolishly, roughly, inconsiderately, wrongly, but I can point at whatever is going on in him. It may be altogether different than I guess or very vague to him, but I refer to (that in him,) whatever he does feel, and I show in what spirit I view such feelings.
I have described this situation with the most difficult example, when the patient says nothing, shows no expressive behavior, not even with his face or body. But, even when people are speaking quite articulately, we want to respond not to the words we hear and know, but to the felt experiencing, the felt referent, the mass of inner momentary felt meaning, which we do not fully know. This felt experiencing is not what people say but rather what they talk from. And only as they work with this experiencing, and as its felt meanings evolve, does change happen in any psychotherapy [12, 13, 14]. We can respond to, and follow, not only the logic of what is said or the external situations people talk about. Even when people make perfect realistic and logical sense, it is still vital to respond to their felt meanings rather than only to abstract conceptual meanings. For this reason, psychotherapy is not really so different when the patient gives us no concepts. We know less, but we can still respond to the concretely felt events going on in him.
With schizophrenics, the therapist must often not only initiate the interaction, but also keep on making it happen all the time. It does not just develop and continue of its own momentum. When it gets started, that does not mean it will be there next time. At least it may not be at all visible next [Page 10] time. Over and over again the therapist must make a moment of interaction. If he does not, nothing happens.
What is the nature of that "illness"? First of all, "schizophrenia" is (and in our research we took it as) the catch-all category in hospitals, a label attached to anyone who is not clearly manic-depressive, alcoholic, epileptic, or something else one can define. This means it includes about half the hospital's population, and consists of just anyone. Some of these people are no different from anyone else, except that things recently happened to them which made life impossible and pushed them out of the world, so to speak. If someone can help them back into the world, they are not fundamentally different from other people.
Another group in that mixed population were perhaps pushed out of the world very early, never quite fully got into the human interpersonal world. These people may be much more difficult to help. However, I use the same words about them. I think schizophrenics suffer from being disconnected from the world. Being in a hospital, particularly a state hospital, is a late, visible, physical dramatization of their being disconnected from the world. And this is the disease we try to treat in the hospital! At first, these people were abandoned and isolated as persons and often lived in situations which seemed externally all right. Other persons could have existed interpersonally in such a situation, but this person could not. His inward isolation explains why finally he could not last . Being isolated in a hospital in physical space is at least the second sense in which he has been abandoned. First he was abandoned many many times in interpersonal space.
The point I want to make is that human beings are not machines who have loose wires in them or burnt-out tubes. There is not in us the kind of broken machinery that an ideal surgeon can reach and fix, or readjust, or take out the thing that is wrong, or reconnect something inside this machine. We are interactive, experiential organisms [14, 16-23]. When I respond to what goes on in a person, then something goes on in him. Of course, something goes on in him also before I respond. He is in pain, anxious or dulled, he has lost his sense of himself, he does not have any feelings, everything is flat. When I respond (or let us say, when I succeed in responding, because I often try and fail for weeks and months) then something more is suddently going on, he does feel something, there is a surprising sense of self and he feels "Gee, maybe I'm not lost." He does not say that. On the contrary, only then does he first say he feels lost. That is when he first says "There is no place for me in the world." A person can feel and express anything only as he is in an ongoing process. Without any place or world he feels nothing, only weird and selfless. With me there is enough of a place and world so that he feels [Page 11] interactively ongoing. Then he feels lost. It is not the inside that is sick. The "illness" is not internal pieces we have to eradicate. The "illness" is not "in" the human being as if he were a separated, boxed, packaged machine. We live as interactive processes.
How we live toward the world and others, how we sense ourselves in situations and referred to by others, that is us. If there is nobody there to refer to me personally, and if I have not somehow learned in other relationships to respond to myself personally, or cannot now do so, then I am not there, and everything gets very flat, very strange, and very weird. If you have ever spent five or six days by yourself without talking to anyone else, then you know something of the quality of feeling it is. But many people can live well toward nature, or with their own responses to themselves. Others find only stoppage and weirdness when intolerable events and feelings have been ground into dullness and inner isolation has long been permanent.
What kind of an illness is that? We talk of "resolving the symptoms and not reaching the basic illness." This would be the case when there are no more hallucinations but the person is still miserable, cut off, alone. It is then said that "the basic personality trouble" has not changed. Thus, "schizophrenia" is not really the "crazy" symptoms as such. Then again, other people talk of just the opposite: "I know many schizophrenics who are out there in the streets, who are working, and they are all right, but they still have the same crazy experiences," says one well-known therapist. Here the personality difficulty seems ameliorated, but that is not what schizophrenia is, either. Despite solutions in personality difficulty, these people still have "schizophrenic" experiences. It is the symptom-mode which is "crazy." But, we say that the symptoms also are not quite what schizophrenia is. These symptoms can go on or off within minutes. When we cure the symptom we are not content. The overt psychotic manifestations do not really define schizophrenia.
A third factor is indicated in the evidence that schizophrenia is really a relationship [17, 23]. It is a sick way of being married, or a sick family, it is an untenable way of being with another person. One is "isolated" from the world by reacting always within a given single intolerable relationship. Within this relationship one's experiential feeling processes can not be interactively ongoing—yet one is stuck within that relationship and not in the world . Not the bad relationship as such, but the stoppage of experiential process in it, is the "illness."
The policy of many hospitals (in Wisconsin, for example) is to send patients back to the same relatives that signed the patient in. This policy sends him back exactly to the relationship in which he can be no more than his sickness. We are tending in two directions with that problem. One is to treat the whole family, which gives some recognition to this interactive na-[Page 12]ture of the illness . The other direction is to try and make a new life possible for the patient (protected workshops, halfway houses, new lodgings, and work). But, the possibility of a new life for the patient should be held out to him right at the beginning, when he is sitting there silent, has no hope, and nothing to say . I can say to him "I think I can help you get out of the hospital and, if you want to, you can live in the city instead of going home. I suppose you don't believe that you could get out of here, but I do. First you work upstairs, and then we will help you find a job outside, and then I'll help you find another job and a room in town, if you want one. I'll stick with you and get you out of here. I know you don't think that's possible now." If that is held out to patients when it still makes little sense, then the fundamental cut-offness can yield to a beginning interaction process into the world. We must begin by overcoming the break that has happened between the patient and the world, his sense that he is not in the world and can't be. Inside himself he is not feelingly alive to think about this, or feel and express himself about it, hence we must begin by restoring the possibility for such feelings and thoughts.
My conception of the illness: It is not so much what is there, as what is not there. The interactive experiential process is lacking, stuck, deadened in old hurt stoppages, and in disconnection from the world. It cannot be ongoing, except in and toward someone and in the world. If a toaster is unplugged, would you take it apart to find out what is wrong inside of it?
The concrete reality of humans is the experiential process, and this is no purely internal thing, but a feeling-toward others in situations. If it is not ongoing, then it cannot be made ongoing, except as we respond empathically to make interaction happen, as well as reconnect the person at least to a promised and imagined outside situation in which he might be able to live. And only if he can later actually try such situations long before he is objectively well enough to do so, can he usually become well enough to do so. Later, we really must help him with job and room, be available for calls at night and meetings in odd places. It was through what some released patients taught me in this respect that I came to promise such things to other patients at the start.
Of course, we do not yet really understand what schizophrenia is. We cannot claim to know. In addition to symptoms, personality difficulty, and experiential interactive stoppage, there may be physiologic conditions both etiologic and accumulated results of long isolation. If pharmologic help is found, it may greatly speed the recovery. But, someone must respond. Only in being responded to does the patient then seem to have ongoing feelings and therefore, the ability to "be aware" of them. It seems likely that the absence of this experiential interaction process is schizophrenia.[Page 13]
The above talk was given in the Spring of 1963. Since then, some preliminary findings have been reported [24-28]. The project has been completed, and the findings are in press . Here is a brief summary of the findings:
The only test measure which showed statistically significant differences between the therapy group as a whole and the control group was the TAT. The therapy group appeared significantly more improved than the controls on Barrington's  TAT analysis (rated "blindly," that is, coded, so that the TAT analyst could not know early from late or success from failure cases).
Both therapy and control groups as a whole showed some degree of improvement. However, if psychotherapy is effective, the group, all of whom received it, should have shown clearly greater changes on most measures.
But, to which patients in the therapy group was effective therapy given? In past researches one had no way of knowing who of the "therapy group" received "psychotherapy" as the researcher would have defined it, and who did not. In the present research we defined "psychotherapy" in terms of both patient and therapist behavior during the interviews. The group, all of whom did receive this psychotherapy, should do clearly better than the controls (even granting that some of them might also have received it).
In the present research it was our main aim to define actual in-therapy behavior. The hundreds of randomly selected, coded segments were analyzed along three variables of therapist behavior (empathy, genuineness, and positive regard) as defined by Rogers , and as operationally defined on our rating scales in terms of specific aspects of therapist behaviors. On the patient side, with developments from Rogers Process Scale [2, 3], we measured the patient's level of "experiencing" [4, 12, 13, 14].
Those patients who showed the highest level of experiential engagement in the process of therapy also (with statistical significance) showed the greater improvement on most of our diagnostic tests, and on the MMPI, than either the low process group or the controls.
Similarly, those patients whose therapists were high on empathy and genuineness showed significantly greater improvement both on diagnostic measures and MMPI changes, than either controls or patients whose therapists were low on empathy and genuineness.
Finally, the patients who showed high levels of the process of therapy were also significantly those whose therapists showed the highest levels of empathy and congruence (as already implied above, since both variables correlate significantly with the positive outcome changes).
These findings raise the question: To what extent is belonging to the high group due to therapist characteristics, and to what extent might it be [Page 14] due to patient characteristics? In one part of our project  therapists regularly visited one ward so that many patients could see the same therapist, and each therapist could see many of the patients. As a result, we could later measure the extent to which (in that part of the project) the differences between therapists affected the process level of the same patient, as well as the extent to which different patients receive different therapeutic levels from a given therapist. Both effects were found , but the effect of the therapist is stronger.
The above findings concern the level (throughout therapy) at which the patient engages in an experiential therapy process. Originally, we had predicted that successful patients would increase this level over the course of therapy. Since then we have found, both with schizophrenic and with more usual neurotic clients [30, 31] that there is only a slight increase during therapy. If the level of engagement in experiential process is sufficient, the individual succeeds in therapy even without a great rise in this level later in therapy.
The Experiencing Scale enables us to measure the extent to which effective psychotherapy is actually ongoing in a given tape-recorded psychotherapy interview (that is, whether that sort of patient behavior is now ongoing which is associated with later successful outcome changes).
There may be considerable implications, both for research design and for practical clinical policies  in the fact that, on both patient and therapist behavior variables, we can measure whether or not effective psychotherapy is now taking place, without having to wait some years for outcome measures to tell us so.
Considering these many findings, one should not miss the fact that the degree of success, and the length of time involved (six months to three years) in our work with these patients was broadly speaking not much better or swifter than has been reported by others. Only approximately half of our patient group were really greatly improved. On the other hand, that was the group in which both therapists and patients showed measurably higher levels of that sort of interview behavior we defined as psychotherapy.
. ______. A Tentative Scale of the Measurement of Process in Psychotherapy. In Research in Psychotherapy. Rubenstein, E. A. and Parloff, M. B., Eds. American Psychological Association, Washington, D. C., 1959.
. Gendlin, E. T. and Tomlinson, T. M. The Process Conception and its Measure-[Page 15]ment. In The Therapeutic Relationship and its Impact: A Study of Psychotherapy with Schizophrenics. Rogers, C. R., Ed. University of Wisconsin Press, Madison, Wis. (in press).[Page 16]
. Gendlin, E. T. The Social Significance of the Research. In The Therapeutic Relationship and Its Impact: A Study of Psychotherapy with Schizophrenics. Rogers, C., Ed. University of Wisconsin Press, Madison, Wis. (in press).