Psychoanalysis and psychotherapy: 

a revision

Merton M. Gill

---

Vai alla versione italiana / Go to the Italian version

---

Part 1 of 2 parts

Part 1: Introduction; A review of the present situation; Transference and its analysis; Illustrations of inadvertent influence on the transference; Neutrality

Part 2: The regressive transference neurosis; New experiences in the therapeutic process; Controlling the transference; Applicability of analytic technique in varied settings; Appendix (the history of this paper); Summary; References

Introduction by Paolo Migone

This classic paper by Gill of 1984, often quoted also in discussion lists, suggests a revision of the traditional way of seeing the difference between psychoanalysis and psychotherapy, trying to solve some of the contradictions that were left unsolved in this age-old debate. Gill, who made this revision on the basis of his own redefinition of the concept of transference formulated in the last 15 years of his life (Gill, 1979, 1982, 1983, 1993, 1994),suggests a widened conception of psychoanalysis, which can be applied to the most varied settings. According to this revision, he modifies not only what he calls the "extrinsic" criteria of psychoanalysis (high frequency of sessions, couch, patients' selection, etc.), but also the traditional "intrinsic" criteria (i.e., internal to the theory)that Gill himself had suggested in another paper he had written in 1954,exactly 30 years earlier, already considered a cornerstone of the classic theory of technique. The intrinsic criteria he had suggested in 1954 were the following four: analysis of transference, analyst's neutrality, induction of a transference neurosis, and privileged use of interpretation. In this1984 paper, instead, Gill maintains only the first one of these criteria, analysis of transference (the other criteria are strictly dependent on this one), but he conceives the transference in a different way, according to his new theoretical revision that could be defined "interpersonal" or "relativistic". As we can easily understand, the theoretical, clinical, and even sociological and institutional implications of this revision are noteworthy, still widely discussed by the psychoanalytic community (consider, for example, the possibility of defining "psychoanalysis" what was previously called "psychoanalytic psychotherapy", and what this might imply for training and psychoanalytic organizations).

This paper first appeared in The International Review of Psychoanalysis, 1984, 11: 161-179, with the title "Psychoanalysis and psychotherapy: a revision" (this article, as well as all the articles published since 1920 in The International Journal of Psychoanalysis, The Journal of the American Psychoanalytic Association, The Psychoanalytic Quarterly, The Psychoanalytic Study of the Child, The International Review of Psychoanalysis, and Contemporary Psychoanalysis, are available for consultation and downloading from a CD-ROM of Psychoanalytic Electronic Publishing [PEP]). The Italian edition appeared in the book edited by Franco Del Corno & Margherita Lang, Psicologia Clinica.Vol. 4: Trattamenti in setting individuale (Milan: Franco Angeli, 1989,pp. 128-157; second edition: 1999; translation by Franco Del Corno e Angelina Spinoni). We thank Ilse Judas (Merton Gill's widow), Franco Angeli Editore of Milan, and The Institute of Psycho-Analysis of London of the permissions. Two parts are added to this POL.it edtion, translated by me ("Appendix: the history of this paper", and "Summary"), which were not included in the Italian edition.

For a biographical and scientific profile of Merton M. Gill (who was a leading figure of psychoanalytic Ego Psychology, and one of the most prominent members of the prestigious research group led by David Rapaport), see the chapter by Irwin Z. Hoffman "Merton M. Gill: a study in theory development in psychoanalysis" in the book edited by Joseph ReppenBeyond Freud: A Study of Modern Psychoanalytic Theorists (Hillsdale, NJ: Analytic Press, 1985, ch. 6, pp. 135-174). For a discussion of the implications of Gill's theoretical revision, with a detailed clinical example, see my chapter "A psychoanalysis on the chair and a psychotherapy on the couch. Implications of Gill's redefinition of the differences between psychoanalysis and psychotherapy" in the book, edited by D.K. Silverman & D.L. Wolitzky, Changing Conceptions of Psychoanalysis: The Legacy of Merton M. Gill, Hillsdale, NJ: Analytic Press, 2000, pp. 219-235. Further discussions, in Italian, can be found in my book Terapia psicoanalitica (Milan: Franco Angeli, 1995), ch. 4 pp. 69-90 (synthesis of the history of the debate and Gill's position), and ch. 13 pp. 221-229 (Gill's biographical profile), while a detailed clinical vignette is in the Internet web site of the Italian journal Gli argonauti.

Merton M. Gill

The gulf between psychoanalysis and psychotherapy is so broad in the thinking of psychoanalysts that any searching re-examination of the relationship between the two techniques is easily suspected of seeking to undermine the uniqueness of psychoanalysis. I have found this to be true even for the re-examination I propose despite the fact that its ultimate conclusion is that the gulf between the two, as I will suggest they be defined and practised, is even greater than it is usually considered to be. The reason for this may be that I am risking serious confusion by dealing with two separate major topics in this paper. The first is reconsideration with suggested changes of the intrinsic criteria by which analysis is ordinarily defined. The second is the argument that the changes in my conceptualization of the intrinsic criteria significantly extend the range of the extrinsic criteria within which these intrinsic criteria can be fulfilled. So I may be considered to be first abandoning psychoanalysis because of the changes in the conceptualization of the intrinsic criteria I propose and then to be abandoning the distinction between psychoanalysis and psychotherapy because for many that distinction is based on the necessary though not sufficient difference in the extrinsic criteria of the two techniques.

I cannot confine myself to a discussion of my changed view of the relationship between psychoanalysis and psychotherapy and my view that psychoanalytic technique can be employed in this extended range of extrinsic criteria because these two propositions depend upon the changes I suggest in the conceptualization of the intrinsic criteria. I hope to minimize such possible confusion by dealing first with the changes proposed in the intrinsic criteria and then the repercussions of these changes on the extrinsic criteria.

By the intrinsic criteria by which analysis is ordinarily defined I mean the centrality of the analysis of transference, a neutral analyst, the induction of a regressive transference neurosis and the resolution of that neurosis by techniques of interpretation alone, or at least mainly by interpretation [Footnote 1]. By extrinsic criteria I mean frequent sessions, the couch, a relatively well integrated patient, that is, one who is considered analysable, and a fully trained psychoanalyst.

Footnote 1: These intrinsic criteria could be divided into techniques-neutrality, analysis of transference and interpretation as distinguished from what they are designed to accomplish a regressive transference neurosis and the resolution of that neurosis mainly by interpretation but I believe the differentiation is unnecessary for the clarity of my argument and hence to attempt to maintain it explicitly would only be distracting. Stone (1954) uses the terms functional and formal for what I am calling intrinsic and extrinsic respectively.

I emphasize at the outset that I am distinguishing between analytic technique and an analysis. That is, I am suggesting that analytic technique can be employed even if a complete analysis is not carried out. The contrary idea is that the issue is one of all or none, that is, unless the technique is designed to bring about a complete analysis, it is not analysis.

Although I will propose that psychoanalysis (i.e. psychoanalytic technique) as I will define it is applicable across the whole range of psychopathology, my convictions are the strongest for its application to patients ordinarily considered analysable for whom issues of time and money preclude the usual setting of an analysis. I stress that point at the beginning because I do not want my discussion to be overshadowed by the question of the applicability of the psychoanalysis I will describe to graver psychopathology, though I do believe that experimentation with it in such psychopathology is warranted.

Today, as in 1954, I am writing about psychoanalysis and psychotherapy from the vantage point of psychoanalysis. Some years after Freud developed psychoanalysis, psychotherapies began to be progressively infused with psychoanalytic practices to the point where it became progressively urgent to keep the differences between the two clear and yet at the same time recognize that they overlap.

Those analysts who contrasted psychoanalysis and psychotherapy most sharply were likely to use the formula that the transference is analysed in psychoanalysis but manipulated in psychotherapy. Those who were impressed by the overlap said that this formula opposes the two methods too sharply. They said that psychotherapies vary in the extent to which the transference is interpreted, sometimes being interpreted a great deal, and in the extent to which the transference is manipulated, sometimes being manipulated not at all. To this latter argument, which brings psychoanalysis and psychotherapy closer together, those who insist on the distinction reply that the intrinsic criteria, as usually stated, apply only to analysis.

The thrust of my 1954 paper was to insist on the difference and at the same time to recognize that the two are on a continuum. My conclusion that there are "relatively autonomous" conflicts which can be resolved in psychotherapy and that therefore the distinction between the two should not be exaggerated led me to say at the end of my paper in 1954 that I feared I would be misunderstood to be saying on the one hand that psychotherapy cannot do what psychoanalysis can and on the other hand that it does in fact do so. I urged that we should import more of the non-directive spirit of psychoanalysis into our psychotherapies, and that while the difference between the two with regard to how they deal with transference could be diminished, it would not be eliminated. I also agreed that only psychoanalysis is characterized by the intrinsic criteria which I accepted as usually stated. I did leave open the door that analytic work might be possible in less than optimal extrinsic criteria.

The question of the relationship between psychoanalysis and psychotherapy is even more important in practice today than it was in 1954 because of the practical difficulties in maintaining the ordinarily accepted extrinsic criteria of analysis. Many analysts are perforce seeing patients less frequently because they cannot afford to come more often as well as because of the competition of the briefer therapies which have become popular. For any particular analyst the urgency of the topic is related to the conditions of his practice. Training analysts, for example, whose practice is more likely to include patients coming 4 or 5 times a week, feel less urgency than other practitioners. The question becomes: How widely can the range of extrinsic criteria be expanded before the analyst must decide for psychotherapy rather than psychoanalysis?

There is even more questioning now among analysts than there was in 1954 as to how rigidly the extrinsic criteria must be maintained for the feasibility of the employment of analytic technique, although less question has been raised about the intrinsic criteria.

What is considered the necessary frequency is gradually dropping. Freud started with six times a week but found that he could work at five times a week when the 30 hours he had expected to divide among 5 patients had to suffice for 6 (Kardiner, 1977). Four is commonly accepted these days. Freud said that

for slight cases or the continuation of a treatment which is already well advanced, three days a week will be enough. Any restrictions of time beyond this bring no advantage either to the doctor or the patient; and at the beginning of an analysis they are quite out of the question (Freud, 1913, p. 127).

Nevertheless many experienced analysts begin a case in analysis with one or two hours a week, either because the patient can afford no more or the analyst has only that much time available, with the hope of later increasing the frequency.

The issue of frequency has been discussed relatively little except in relation to the proposal by Alexander (1956) and some of his associates to influence the transference by decreasing or increasing the frequency of sessions. Whatever value such a technique may have, I agree with the consensus that it violates the central tenet of analysis the analysis of the transference rather than its manipulation and should therefore be classed as psychotherapy rather than psychoanalysis. As a foretaste of what is to come, however, I ask two questions. First: is it possible that it is compatible with analytic technique to alter frequency if at the same time the meaning of the alteration is thoroughly analysed in the transference? Alexander (1954, p. 699) spoke of analysing the transference which in his view necessitated the alteration of frequency but he did not say anything about analysing the repercussions on the transference of the manipulation itself. Second: is it possible that the very insistence that a certain frequency is necessary to conduct an analysis without the interpretation of the meaning of that insistence is a manipulation of the transference which bears some similarity to the technique for which Alexander has been so roundly criticized?

The issue of chair or couch is often glancingly referred to. Various points like these are commonly made: a patient should not be forced on to the couch. "Couch-diving", that is, a patient's excessive eagerness to use the couch, may bespeak serious resistance rather than otherwise. An analysis is not ruined if the patient gets up sometimes (the Rat Man paced the office in early hours) but it is important to analyse discomfort in either lying down or sitting up. The former is more likely to be investigated than the latter. It is recognized that the couch may be preferred for the analyst's comfort rather than the patient's. Critics of analysis are fond of pointing out that Freud said he could not bear being gazed at for eight hours a day (1913). In general it is agreed that while the couch has advantages and an inability ever to lie down bespeaks an unresolved problem, its use may be defensive, and it is not absolutely essential.

The issue of the patient's pathology has received extensive discussion in our literature. A minority of analysts believe that a strict analysis can and indeed should be conducted with even the sickest of patients whereas most analysts believe that, at the very least, important alterations of strict technique are mandatory.

The issue of training of the therapist would seem to be beyond discussion. How can someone untrained in analysis conduct an analysis? But how much training is necessary? Are there certain basic elements of technique which it might be better for the therapist to employ as well as he can rather than to attempt to avoid altogether? I will return to this question when I have laid a better basis to discuss it. I will argue that with the definition of analytic technique at which I will finally arrive, it should be taught to all psychotherapists and that how well it will be employed will depend on their training and natural talent for the work.

The changes I will propose are more radical than a simple extension of the recommendation I made in 1954 that we carry more of the non-directive spirit of psychoanalysis into our psychotherapies. To confine the changes to that would be to imply an acceptance of the prevailing conceptualization of the intrinsic criteria of psychoanalytic technique. The attempt to fulfil the prevailing intrinsic criteria while expanding the range of the extrinsic criteria would actually exaggerate the defects of the prevailing conceptualization of the intrinsic criteria. In fact such practice is what we sometimes see in beginners who are aping what they consider the correct withdrawn posture of the psychoanalyst. The recommendation to broaden the range of the extrinsic criteria within which the intrinsic criteria can be fulfilled is therefore bound up with the changes I will recommend in the intrinsic criteria.

I suggest that with the changes in the conceptualization of the intrinsic criteria I will propose, on the one hand psychoanalysis and psychotherapy become more sharply opposed, and on the other hand the range of applicability of psychoanalytic rather than psychotherapeutic technique broadens. I mean that analytic technique as I will define it should be employed as much as possible even if the patient comes less frequently than is usual in psychoanalysis, uses the chair rather than the couch, is not necessarily committed to a treatment of relatively long duration, is sicker than the usually considered analysable patient and even if the therapist is relatively inexperienced. In other words, I will recommend that we sharply narrow the indications for psychoanalytic psychotherapy and primarily practise psychoanalysis as I shall define it instead.

The length and nature of these introductory remarks are a result of what I consider to be the misunderstandings which have met earlier presentations of my views and which I am attempting to head off or at least to gain a longer suspension of judgment on the reader's part.

I believe the prevailing theory and practice of psychoanalysis and psychotherapy and the relation between them have not changed significantly since 1954. The reconsideration I am proposing is an outgrowth of my changed views on transference and its analysis which I shall soon summarize. This reconsideration has led me to propose changes in the conception of each one of the intrinsic criteria as I had accepted them in 1954. I am not attempting to review in this paper the contributions of my many predecessors who have similarly reconsidered one or more of these intrinsic criteria.

There is a broader perspective than that of transference, however, within which these changes can be conceptualized. It is that the interpersonal interaction between patient and analyst requires reconceptualization. While the usual psychoanalytic perspective on the interpersonal aspect of the analytic situation is how the patient's view of it is distorted by his intrapsychic organization, I suggest instead that the integrate arising from the patient's intrapsychic organization and his experience of the interpersonal interaction should initially be treated as a rational formulation in a relativistic, perspectival framework of interpersonal reality.

The issue relates to an old one in psychoanalysis often discussed as the difference between a two person and a one person view of the analytic situation. What I believe I am contributing is a spelling out of the repercussions of taking the two person view seriously in the context of the classical view of analytic technique. One way of stating the changed view in its application to the analytic situation is that the setting and the analyst's behaviour exert an influence, ranging from a minor one to a major one, on the manifestations of the potential intrapsychically organized patterns of interpersonal interaction and in that sense co-determine the transference. Another way of stating the issue in terms of analytic technique is that without this view of transference the setting and the analyst's behaviour can become vehicles of inadvertent suggestion which if not attended to exert their effects without being recognized or altered.

The first intrinsic criterion of psychoanalysis which I will take up is the centrality of the analysis of the transference. That psychoanalysis aims at as complete an analysis of transference as possible while psychotherapy does not, remains correct, but this formula acquires new meaning through the changes in my concept of transference and its analysis. I have presented them in an article (1979) and in an expansion of that article into a monograph (1982). Here I can only summarize my conclusions:

1. The notion of an "uncontaminated" transference is a myth because the expression of the transference is always influenced by the here-and-now interaction between the analyst and the patent. But even more, the nature of interpersonal interaction is such that the transference will always have some degree of plausibility in terms of something related to the analyst. This something includes things the analyst has not done as well as what he has done. The usual view is that the patient distorts the situation by constructing it in terms of his intrapsychic patterns. I say instead that the therapist's behaviour lends plausibility to the patient's experience. The relative roles of these two contributions differs from instance to instance. The important thing is not to approach any single instance with a predetermined conviction as to their relative importance.

2. The examination of the transference should begin with a careful clarification of exactly what the patient's experience is, including whatever influences in the current situation are involved. The therapist must not ever assume that he necessarily clearly understands the patient especially if the patient is vague, indefinite, and elusive.

3. It is commonly not recognized that transference is ubiquitous because the resistance to becoming aware of it on the part of both patient and analyst leads to its appearance in disguised form in associations not manifestly about the current relationship. The clarification of what the patient is experiencing requires seeking out these allusions to the present relationship and making them explicit. I suggest that this activity be designated interpretation of resistance to the awareness of transference in contrast to interpretation of resistance to the resolution of transference.

4. The analysis of transference, after the clarification of what the patient's experience of the relationship is, should begin with a search for what makes the patient's experience at least somewhat plausible to him. This amounts to a major change from the usual emphasis on how the patient's experience is a distortion of the situation to an emphasis on how the patient's experience can be understood as a plausible understanding of the situation [Footnote 2]. The compulsion to re-experience and re-enact the past is a major motivation for the selective attention with which the patient experiences the present as he constructs his plausible understanding of it.

Footnote 2: Work in the resolution of the transference analogously should seek the plausibility in the patient's experience of past relationships (Hoffman, 1983).

5. A major role in the resolution of transference is played by the patient's coming to see that this plausible meaning of the situation is indeed no more than only plausible and not unequivocal, that is, that his experience of the situation is based to a greater or lesser degree on determinants within himself.

6. The awareness that there are such determinants will probably sooner or later lead to data from the past which help explain how they came to be. Such explanation falls into the familiar category of the resolution of the transference by the examination and re-evaluation of the past.

7. The patient not only experiences the analytic situation in a way which conforms to his preconceptions, whether conscious or not. He also behaves in a way designed to get the therapist to justify these preconceptions which in turn lends further plausibility to them. The extent to which the therapist is unaware of how he is being experienced may well be a measure of his unwittingly responding to pressure from the patient and coming to behave in a way which increasingly justifies and makes plausible the patient's preconceptions. Sandler (1976) has described this phenomenon as the analyst's role responsiveness. Otherwise expressed, the patient stimulates countertransference.

8. I add a point about countertransference, to which I directed little attention as such in my monograph. It is that the most important aid to the therapist in discerning his countertransference is the patient's interpretation of it, to a large extent in disguised references in his associations. Langs (1978) and Hoffman (1983) have described how the patient can be seen as an interpreter of the analyst's experience. This similarity between Langs' views and the view which Hoffman and I share must not be permitted to obscure a crucial difference. Langs sees the patient as correctly perceiving the analyst's unconscious intent whereas we see the patient as only constructing a more or less plausible view of the analyst's motivations.

A frequent criticism of my view of transference and its analysis is that it is said to be in opposition to what many others believe to be the essence of the psychoanalytic method, namely the recovery of the patient's history. I believe it is not a matter of opposition but of technical priority. When I spoke of the important role played in the resolution of the transference by the recognition that the patient's experience of the relationship is plausibly but not unequivocally determined by the actuality of the analytic situation, I did not mean to derogate the role of the patient's awareness and integration of his history in resolving the transference. I do believe, however, that priority of attention to the recovery of the history can lead to important inadvertent effects on the transference. Such priority is often a defensive flight by either patient or analyst or both from the discomfort aroused by explicating, examining, and interpreting the transference phenomena in the here-and-now. To the extent to which this is true the recovery of the past may exert its effect by way of inadvertent suggestion.

The issue which I am discussing has a long history in analytic technique in another regrettably and unnecessarily polarized controversy over the relative importance of experiencing and remembering. I expect to be considered to be underemphasizing the past as Ferenczi & Rank were in their monograph of 1925 over 50 years ago entitled The Development of Psychoanalysis. While I am not discounting the value of remembering, I believe that an analysis in which priority of attention goes to the transference expressed in the here-and-now, including the analyst's contribution, will be much freer of lasting effects of inadvertent suggestion than one in which priority of attention goes to genetic interpretation which may bypass transference. The analysis of the transference may be defined as attempts to understand the patient's current experience, in relation to the analyst, including its plausible sources in the here-and-now, so that its sources in the past experience, wishes, and conflicts can be illuminated and more conscious, and flexible integration of past and present is brought about.

Much of the analyst's behaviour which from the patient's point of view leads him to a plausible interpretation of the analyst's motivations is from the analyst's point of view inadvertent. For initial illustrations I turn to two of Freud's cases. First, the Dora case (1905). Freud defended himself against the possible criticism that he should not have talked about intimate sexual matters with a young woman by arguing that to do so was not necessarily prurient or harmful. True enough. But would it not have been plausible for Dora to interpret his obvious interest in her sexual life and the possible apparent concomitant relative lack of interest in what was consciously her primary concern, the hypocrisy in her family as a more subtle variant of Herr K's sexual interest in her? Second, the Rat Man (Freud, 1909). Would it not have been plausible for the Rat Man to have interpreted Freud's interest in getting the details of the rat torture to be a form of torture? Freud has been criticized (Kanzer, 1980) for influencing the transference by trying to guess what the torture was. We may infer from Freud's explicit disavowal that he wanted to torture him that he was trying to dispel any such feeling on the Rat Man's part.

I am not saying that Freud should not have inquired into Dora's and the Rat Man's mental content. I disagree with those who believe that all such material can emerge spontaneously in free association, and that they should wait for it to so emerge so that the patient has no rational basis for imputing sexual or aggressive intent to the analyst [Footnote 3]. I am suggesting on the contrary that such plausible imputation is unavoidable precisely because of the multiple interpretations to which human behaviour lends itself and because the patient is primed by his past selectively to interpret the present. Rather than to pursue such impossible avoidance on the basis of a mistaken premise as to the nature of the analytic situation, the analyst should bend his energies to detecting the implicit indications that the patient is making such imputations and bring them into the open by interpretations of their plausibility in the light of the here-and-now with the ultimate aim of elucidating the patient's own contribution. The latter stems from the patient's past, the there-and-then.

Footnote 3: How directly and openly something needs to be expressed, in order to be able to say that it is "emerged"? This has always been a hot topic in clinical discussions, due to the differences in evaluating the role played by inference in assuming that something is presumably present in a masked way.

Here is an illustration from a recent paper by Rangell (1979). He writes that for a period of time a patient's response to every interpretation or achievement of a piece of insight was "So what?" or "So what happens now?" or "So what am I supposed to do?" and such. He does not describe any investigation of the meaning of the "So what?" but says he soon came to feel this was a genuine seeking of information partly from the patient's impatience to have his behaviour improve more rapidly and partly to block the progression to more painful insights. So he made the following intervention: "You ask so what and not so why". He considers that the patient needed this suggestion to stimulate progress in the analysis. He explicitly disavows that the patient was being reprimanded or ordered to think along these lines.

I believe this disavowal betrays Rangell's peripheral awareness that the feeling of being reprimanded or ordered to think along these lines might well have been plausibly experienced by the patient as an inadvertent effect of his interpretation on the expression of the transference. In fact he introduces the vignette to the reader by saying he wants to make the point that an active role on the part of the patient must be enlisted continually. This was therefore probably the implicit message both as intended by the analyst and experienced by the patient. I believe he should not have assumed that he knew what the "so what?" meant. He should have attempted to understand its meaning in the transference. Furthermore, once having made his remark he should have been alert to its possible repercussion on the expression of the transference.

Rangell says the interpretation was successful and cites some subsequent insights about the patient's relation to his mother. He does not provide enough data to say how these insights may have combined disguised references to the patient's past and to the analyst's intervention.

I believe the example illustrates a typical failure to recognize that primary attention should be directed to the examination of the transference in the here-and-now, in this instance first to the "so what?" and then to the response to the intervention, rather than to the there-and-then as implied in the therapist's asking "why?" I suggest that the repeated "so what?" is a sign of a persistent and central issue in the transference being expressed in the here-and-now with roots in the past, and that unless it is elucidated the continuing analytic work may well be essentially intellectualizing and under the influence of suggestion.

The examples I gave all illustrated how the patient experienced interventions. What is also often not focused upon by analysts is the role they have played in the common transference responses to features of the analytic setting themselves. The fact that these features may be experienced very differently by different patients or by the same patient at different times, or even simultaneously, probably increases the likelihood that the analyst will regard the patient's experience as essentially or even entirely self-determined. But the analyst's attitude is also a determinant of how the patient experiences these features of the analytic situation. The couch may be a welcome indication that the patient need not concern himself with the therapist's reactions or it may mean that the patient is deprived of the cues he must have to the analyst's reactions without which he is too frightened to speak. The manner in which the analyst brings it about that the patient lies down will to a varying extent co-determine whether the patient experiences the couch as a relief from fear of meeting the therapist as an equal or as a degrading submission. The way the frequency of sessions is settled will co-determine to a greater or lesser extent whether frequent sessions mean a promise of indulgent unending care or a loss of respite from a relentless invasion of privacy. Similarly, open-ended duration may be a reassurance that there is adequate time but it could also be experienced as an indeterminate sentence without possibility of probation. Jacobson writes:

Many depressives tolerate four or even three sessions weekly much better than six or seven... Daily sessions may be experienced once [sic] as seductive promises too great to be fulfilled, or then again as intolerable oral sadistic obligations which promote the masochistic submission (Jacobs, 1954, p. 603).

The way the therapist exercises restraint will codetermine whether it is interpreted by the patient as aloof austerity, sagacious thoughtfulness, or an anxious effort to avoid breaking some rule of proper analytic conduct (Stone, 1954, p. 575).

While these possible experiences may be divided into a stimulus § lying down for example § and the patient's experience, this distinction is not the same as the division of the patient's experience § possible only conceptually at best § into plausible responses to the present and transferences from the past. For the patient's experience is the transference, an indivisible unity.

The analytic interaction and setting thus inadvertently influence the patient's experience of the relationship. Freud (1919) explicitly based his recommendation of abstinence on the substitute gratifications which the patient may get from the transference relationship itself. One may disagree with Alexander's recommendation to interfere with the dependent gratification sometimes afforded by the analytic setting by diminishing the frequency of sessions but one must applaud his recognition of this phenomenon and his recommendation that it be interpreted. Weigert (1954) discusses well the magical meanings of the rituals of the setting although she does not put enough emphasis on their detailed analysis in the transference. I suggest that a technical conclusion which may be drawn from these illustrations is that in analysing the transference the analyst should first focus on his contribution to the patient's experience of the relationship in the patient's response both to interventions and to the features of the analytic setting.

In the light of this suggested changed view of transference I turn to the second of the intrinsic criteria of psychoanalysis as I defined them in 1954 and as they are generally accepted. It is the neutral analyst. The implication that, whether advertent or inadvertent, the therapist's influence on the transference is avoidable is carried in the concept of neutrality. In my 1954 definition I said that psychoanalysis was conducted by a neutral analyst. I realized then that the recommendation for neutrality is an effort to avoid effects on the transference but that it cannot accomplish that aim because the patient inevitably interprets the analyst's behaviour in ways other than those the analyst had intended. This realization is in fact an aspect of one of the most fundamental propositions in psychoanalysis. It is that the meaning to a person of an external situation can never be determined from the outside alone. The analyst tries to avoid behaving in a way that will be construed correctly and clearly as obviously reflecting some erotic or hostile intent, but even so he can never take for granted that a particular behaviour on his part has a particular meaning to the patient, both because he cannot have unequivocal knowledge of his intentions and because of the transference. To know the meaning he must explore the patient's experience. While investigating the meaning the patient ascribes to his behaviour the analyst does not reveal his intended meaning because such revelation prejudices the exploration of the patient's experience. If the analyst is persuaded that he can indeed behave in a completely neutral way he will think that a patient who experiences him as different from what he consciously intended is distorting his intentions. The analyst will be motivated to search for such inadvertent effects of his behaviour only if he is convinced not only of how commonly such effects occur but also that they will be expressed in disguised ways.

I believe it is a partial awareness of this point which has led to the degree to which analysis is characterized by a reluctance to engage in interaction with the patient. It is as though we try to prevent the patient from mistaking our intent by having no intent. The belief that the analytic process has a kind of automaticity which takes over once it is set in motion fosters this reluctance to interact.

But such reluctance implies a failure to be fully aware that because analysis takes place in an interpersonal context there is no such thing as non-interaction. Silence is of course a behaviour too. Nor can one maintain that silence is preferable for the purpose of analysis because it is neutral in reality. It may be intended to be neutral but silence too can be plausibly experienced as anything ranging from cruel inhumanity to tender concern. It is not possible to say that any of these attitudes is necessarily a distortion. Because I was partially aware of this fact in 1954, I suggested that the analyst could at least adopt a stable relatively neutral attitude so that there would be a baseline against which to measure the patient's attitudes. What I now recognize and did not in 1954 is that these experiences which the analyst inadvertently produces are not distortions of the analyst's behaviour because he cannot assume that these experiences have no rational basis. With this realization the analysis of the transference takes on a new complexion. While inadvertent effects on the transference can be overlooked even with the best of intentions, they are more likely to be recognized if the analyst has become persuaded of the role played by the here-and-now interaction in how the transference becomes manifested.