Free Association: Is It Still Relevant?

Joseph Schachter

Joseph Schachter, M.D., Ph.D., retired, was formerly Pittsburgh Psychoanalytic Center training and supervising analyst, former chair of the Committee for the Evaluation of Research Proposals of IPA and author of five books and more than 60 papers.


Joseph Schachter

Freud’s formulation of free association seems to have been derived from Francis Galton’s previous discovery of associations. Lewis Aron (1996) believes that “Freud was eager to substitute something for hypnotic suggestion that was more his own, and the very specific form that Freud’s psychotherapy was to take was highly distinct and reflected his own personal complex synthesis.” Conceived by Freud in 1892, free association is one of the most long-lived elements of traditional psychoanalytic treatment, while receiving less critical attention than other recognized facets of psychoanalytic treatment. For the period 1920-2016, PEP lists 828 papers about countertransference, 1227 papers about transference, while free association lags behind with 380 papers. Numerous aspects of psychoanalytic treatment have been thoughtfully discussed and modified substantially, some even dramatically over time, but the concept of free association remains virtually unquestioned and unchanged since its origin.

Unable to deeply hypnotize Miss Lucy R of the Studies in Hysteria, Freud abandoned hypnosis and carried out the treatment with the patient in a more or less ordinary state of mind. The patient was told to lie back, close her eyes and concentrate uncritically on the thoughts and images that arose. When the patient could not answer a question, “I [Freud] placed my hand on the patient’s forehead or took her head between my hands and said, ‘You will think of it under the pressure of my hand. At the moment at which I relax my pressure you will see something in front of you or something will come into your head. Catch hold of it.’”

Note in this initial conception it was not the patient’s spontaneous verbalizations that were utilized, but rather, the patient was asked to respond to a question. Further, the patient was interacting not with a neutral analyst, but with a magical, powerful figure who had presented her with a suggestion, reminiscent of the hypnotic treatment which Freud had been utilizing; the required response arrived after Freud removed his touch.

Thus, the origin of free association was the analyst adopting an omniscient role and using suggestion to enable the patient to provide the “correct” answer. Only later, when Freud became concerned about the negative impact of the analyst’s use of suggestion upon the evaluation of the psychoanalytic profession, did he radically revise his conception, proposing that the patient’s free associations might be considered to be produced independently of the analyst’s thoughts and feelings.

By 1925, however, Freud returned to his initial conception with the recognition that the patient’s productions were actually not produced completely independently of the analyst’s behavior and feeling by an enhanced statement: “We must, however, bear in mind that free association is not really free. The patient remains under the influence of the analytic situation even though he is not directing his mental activities on to a particular subject. We shall be justified in assuming that nothing will occur to him that has not had some reference to that situation.”

Defining the Concept

Reviewing the conception of free association, Merton Gill wrote he agreed with Samuel Lipton (1982) “that a correct statement of Freud’s view of free association is that it is whatever the patient says in response to the request to follow the fundamental rule”—to tell us everything that is in his/her mind. Gill notes that Freud in a number of places says patients should include, even if they are disinclined to, thoughts that are apparently trivial, irrelevant, nonsensical and embarrassing. Harry Stack Sullivan’s (1962) formulation was the patient should “Say every little thing that comes to your mind, that you notice in your consciousness.”

“We must, however, bear in mind that free association is not really free.”


Leopold Bellak (1961) noted, “The analytic concept of free association was originally closely related to the early topological model. In its simplest form, it posited that repression of traumatic events is the cause of neurosis. It was in this context that the ‘basic rule’ was substituted for hypnosis as a superior tool for filling in childhood amnesia.”

However, in the 1950s and 1960s the conception of the patient’s free associations was carried to an extreme. Analysts were considered to be so neutral and anonymous that they were thought to be interchangeable. Switching a patient from one analyst to another was not acknowledged as a disruption of treatment. The analytic process was thought to go on just as it would have if the transfer to another analyst had not taken place. Recent substantial empirical data demonstrates, however, that each psychoanalytic dyad is unique, though changing over time. The implication is if one analyst is replaced by another, this new analytic dyad would be different and unique, and the course of treatment would be substantially modified.

Valuable for Revelations

Leo Rangell (1987) believed, “Free associations also reveal and refer directly to the etiologic traumata of life.” Ernst Kris] (1996) declares, “For me, the central point in psychoanalysis is the commitment to the free association method.” Christopher Bollas (2003) considered that “through the chain of ideas revealed in free association, deeply meaningful and often conflicted thoughts will be revealed.”

More recently, Axel Hoffer (2006) agreed, maintaining “the pillars of psychoanalysis remain the analysis of resistance, transference and countertransference by the use of the free association method.” He added, “Enhancing their freedom to associate remains, in my opinion, the best we have to offer them as they decide for themselves how much they can and want to change.” Basically, he concurred with Rangell’s (1968) earlier conception that psychoanalytic treatment takes place in the mind of the patient—not in the interaction between patient and analyst.

Cautions and Criticisms

Several analysts had earlier delineated criticisms of and problems with free association. Anna Freud (1936) had asserted that even if this ideal [of free association] was realized, it would not represent an advance, for after all it would simply mean the conjuring up again of the now obsolete situation of hypnosis, with its one-sided concentration on the part of the physician upon the id. Otto Fenichel (1945) observed, “There are individuals (obsessional neurotics) who never learn to apply the basic rule.” Gregory Zilboorg’s (1952) early conclusion was “It becomes quite clear how little we know of the processes and laws of free association.” Sounding a cautious note about assessing free association, Rudolph M. Loewenstein (1963) questioned, “if it is possible at all for a patient to free associate, really, in the sense of telling everything that occurs to him. When the question is couched in these terms, an answer is not easy to find.” Loewenstein observed, “As the process of analysis unfolds, we do not expect the analysand to remain continuously in a condition of calm self-observation.”

“Free association is still one of the sacred cows of the psychoanalytic tradition…”

—Irwin Hoffman

A chronological review continues, beginning with Donald P. Spence (1982) who asserted “the associations of a patient have no one-to-one correspondence with his memories and dreams, much less with his unconscious thoughts.” Further, if the patient “is truly free and in his reporting, he cannot be understood; if he is understood, he is not freely reporting.” Seymour Fisher and Roger P. Greenberg (1985) conclude “… free associations cannot be considered entirely free. The verbal conditioning studies indicate that patient response may be strongly influenced by the values and expectations of the analyst.” Robert R. Holt (1989) concluded: “The analyst steers and shapes what the patient reports in his or her mislabeled free associations.”

“The analyst,” Gill (1994) agrees, “inevitably influences the patient’s flow of associations by everything he or she says and does.” He adds, “The very existence of the analytic situation makes the analyst a participant in the patient’s associations.” Clearly, the patient’s free associations are never free either of the analytic situation or of the person of the analyst no matter how “neutral” the analyst strives to be. According to Gill, “If the analyst feels that material has been withheld, rather than attempting to ferret it out, as is unfortunately often done, he or she should analyze the resistance to expressing it.”

“Free association” wrote Irwin Hoffman (2006), “is still one of the sacred cows of the psychoanalytic tradition; it is a term one tampers with at peril of his or her psychoanalytic identity. Can you claim to be a psychoanalyst if you do not believe in free association?”

Hoffman delineates a set of three fundamental criticisms of free association entailing various kinds of denial:

  1. Denial of the patient’s agency. The assumption is that what “comes to mind” is not “chosen” so much as it surfaces as an unbidden thought. However, the patient’s thoughts might emerge as a function of the patient actively thinking them. Further, if several thoughts come to mind simultaneously, only by choosing will it be possible to speak at all.
  2. Denial of the patient’s and the analyst’s interpersonal influence. The relatively silent analyst deprives the patient of an object relationship, a deprivation that was designed to induce regression and, with it, affective states associated specifically with object-related frustration and loss.
  3. Denial of the patient’s share of responsibility for the analytic relationship. The analyst is viewed primarily as a technical instrumentality, rather than a person, and the patient’s provocations will have little serious influence on the analyst’s experience. Why wouldn’t the patient be called upon to exercise some greater degree of judgment in considering the impact of his or her words and other aspects of his or her behavior on the analyst?

While Otto Kernberg (2015) still agrees with Rangell, Kris, Bollas and Hoffer that free association remains at the core of psychoanalytic treatment, he acknowledges and describes in clinical detail his understanding of the factors that limit the ability of patients with narcissistic personality disorders to free associate. The instructions given to the patient explaining the “fundamental rule” of free association include the invitation to try to say whatever comes to mind, in whatever form that occurs, whether this includes thoughts, fantasies, observations, relationships, fears and dreams without attempting to order all these contents in any way—regardless of whether, for instance, that seems easy or difficult, something to be proud of or ashamed about, something important or trivial.

The narcissistic personality patients’ fear, however, that complying with these instructions will expose them to becoming dependent upon the analyst. This fear is intolerable; defensive operations then are geared to protect the patient from any authentic dependency on the analyst. This degree of avoidance of true dependency is frequently matched by a complementary defense of omnipotent control in order to avoid both the emergence of feelings of inferiority and a complete devaluation of the analyst. In this context, the patient’s capacity to free associate becomes distorted by narcissistic pathology, to the extent that a suggestion to the patient to associate to any apparently significant subject matter fails to lead to deepening awareness of emotionally significant material.

Although not remarked by Kernberg, it seems likely that many patients with lesser degrees of narcissistic disorder would have similar, though less intense, difficulties with free association.

Michael Hölzer and Horst Kächele (2010) have argued that free association has never really existed. Neil J. Skolnick (2015) asserts that aside from the consensus that there is no consensus as to what free association actually is, there is no agreement as to which patients are capable of producing free associations. Others, including Kachele and me (2010), have added the clinically important conclusion that free association itself does not necessarily lead to therapeutic gain.

Gill’s summarizing critique of free association (1994), however, goes further, and ultimately leads him to raise the fundamental question, “Is free association necessary?” He reiterates, echoing Freud, “the very existence of the analytic situation makes the analyst a co-participant in the patient’s associations.” Gill indicates he would not begin treatment “by proclaiming the fundamental rule, even if it were worded to make it seem like less of a command.” He adds, “Free association cannot be forced by a demand. It is not possible to ward off resistance by a prophylactic maneuver.” Gill says nothing when beginning with a new patient, except perhaps, “I would like to hear what’s on your mind.…What is important about any instruction or lack of instruction is what it means to the patient.” He quotes Edgar Levenson (1991) speaking of an “extended deconstructive inquiry” rather than of free association.

Previously (2002) I concluded, “In the postmodern period of inter-subjectivity there is less concern about free association and more concern about patient-analyst interaction and relationship.” Andrew Gerber (2009) sounds a similar note, “Or perhaps it would be more appropriate … to set aside the extremes of ‘free association’ versus ‘resistance’ and think instead of the infinite variety of how individuals attend to and spontaneously report thoughts and feelings.”

More than 20 years ago, those two outstanding analysts, Levenson (1991) and Gill (1994), discarded free association as a useful clinical and theoretical concept. Meanwhile, over these same two decades, conceptions of relational/interpersonal psychoanalytic treatment have grown, while interest in free association has diminished significantly, apparent in the 2010-2016 PEP totals of 93 papers about countertransference, 111 papers about transference, but only 11 papers about free association. Given this declining interest, plus the numerous criticisms and problems with free association, and the lack of any empirical evidence that it enhances therapeutic benefit, perhaps it is time for the 124-year-old original concept of free association to be retired.

In Conclusion

Freud believed free association would help him attain his goal of curing neuroses. Now, psychoanalysis strives to help patients develop satisfying and rewarding personal lives and human relationships, often beginning with a trusting patient-analyst relationship. Acknowledging that free association may no longer be useful for many of our patients is the first step toward developing an alternative conception that will facilitate modern psychoanalytic treatment for most patients—perhaps a task for the next Freud.

I would like to thank Judith S. Schachter, M.D., for her valuable suggestions and editorial comments.

Editor’s Note:
References are available upon request to the author at