In a memorable scene in Fiddler on the Roof, the main character, Tevye, pretends to have been awakened by a nightmare that he concocts to convince his wife to change her mind about who their daughter should marry. As he describes this “dream,” dancers and singers act out the story, accompanied by a small band of strolling musicians. In the original Broadway production, this scene was always a crowd-pleaser. In the revival, however, Zero Mostel (as Tevye) found a way to turn it into a showstopper. In the new version, as he recounts his tale, the audience watches him become increasingly distracted by the deafening cymbal crashes of the nearby percussionist. Suddenly eyeing a solution, Tevye grabs for the nearest bed pillow and hurls it at the musician. It lodges between the cymbals just in time to stifle the next crash. This improvised bit of comedy elicited such howls of laughter from the audience that it was permanently incorporated into the show.

Such consummate pieces of stagecraft are the lifeblood of a theatrical production. However, because they develop organically out of the “conversation” of actor, role, and audience, they’re virtually impossible to plan. (Mostel didn’t “find” the pillow bit until he’d played Tevye hundreds of times.) For similar reasons, effective psychotherapy interventions can’t be fully scripted in advance. Like Mostel’s innovation, therapy’s most effective moments are improvisations that arise out of the conversational flow between client and therapist. Indeed, every therapeutic journey is unique, involving unexpected twists and turns. These aren’t signs of the therapist’s lack of skill or treatment failure; they constitute the very heart of the process.

From this perspective, the growing emphasis on treatment manuals and empirically validated methods is a step in the wrong direction. Yes, the public needs to be protected from quacks, and managed care organizations certainly want some assurance that their money is being spent wisely. In the final analysis, however, the effectiveness of a client-therapist pairing is a function of their collaborative dialogue–a process that resists standardization. Undoubtedly, one can specify general principles and guidelines, and therapy can be anchored in a contract that defines roles and sets boundaries. However, therapy also requires a certain creative ambiguity that can’t be reduced to stock exercises or “bottled” like an antidepressant.

The past three decades of research on psychotherapy have added little to Lester Luborsky’s famous 1975 “dodo bird verdict,” that “everybody has won, and all must have prizes.” In other words, with only a few circumscribed exceptions, every empirically tested therapy has shown roughly the same level of (some would say mediocre) results. As Mark Hubble, Barry Duncan, and Scott Miller note in The Heart and Soul of Change in 1999, the vaunted list of empirically supported approaches amounts to little more than a recognition that some methods have been tested and others haven’t.

Even in the case of heavily researched treatments, such as Marsha Linehan’s Dialectical Behavior Therapy (DBT), we still know almost nothing about how particular components of the approach operate. All we really know is that the potpourri of methods Linehan assembled–skills training, mindfulness techniques, phone contacts, group sessions, etc.–works somewhat better for certain types of emotional problems than the ragtag assortment of treatments generally available. It’s possible, even likely, that the advantages in outcome the DBT studies report are mainly attributable to the skills and enthusiasm of handpicked, crackerjack research teams and the comparative abundance of resources available to them. Thus far, we haven’t seen the kinds of dismantling studies that would enable us to understand more about how DBT works.

Unfortunately, manualized treatment protocols, designed to make treatments replicable, create a false impression of objectivity. For instance, we know many who’ve worked on manual-driven projects and who report, off the record, that they were only able to adhere to the manual for the first few sessions–during the intake phase. After that, the content of their sessions became increasingly variable and idiosyncratic, tailored, as one might have predicted, to the exigencies of the developing client-therapist dialogue. In some of these studies, the therapists dutifully continued to administer the prescribed exercises, but they also felt obliged to slip in side conversations that more directly addressed their client’s needs. Unfortunately, such procedural deviations rarely make their way into the published literature. Therefore, readers are left with the false impression that the manualized procedures are easy to implement and reliably produce the desired result. Manuals may not be entirely worthless, but they certainly don’t tell the whole story of what transpired between client and therapist.

Several years ago, we watched a manual-driven group treatment from behind a one-way mirror. Time after time, group members were actively discouraged from discussing issues that concerned them because these didn’t fit the session’s prescribed agenda. The leader kept complaining that the group had to move on because they were “falling behind the schedule.” Ironically, it was those forbidden “disgressions” that group members later remembered as the most valuable aspects of the sessions. At a feedback session with an outside evaluator, members were quick to say that they liked being in the group and profited from meeting the other members, but had little use for the leader’s heavy-handed attempts to administer “the treatment.”

Even the best manuals tend to devolve into a series of vague instructions, such as “continue checking the client’s homework,” “review the previous week’s gains,” or “identify other instances of catastrophic thinking.” Such directives hardly achieve the goal of insuring standardization. Most such studies emphasize group statistics, ignoring the therapist variability that exists even though each client is supposed to be receiving the same treatment. In fact, much of the field’s scientific sweat equity has been invested in studies of therapy’s lowest common denominator–group findings from hothouse projects of canned therapies, using inexperienced graduate-student therapists and highly selected populations. The shame is that these studies convey an erroneous message about what works, focusing attention on particular techniques and away from the kind of conversational improvisation that good therapy requires.

We can trace the current conundrum we’re in–over the difficulty of making real therapy fit into a scientific paradigm–to the “slow-acting poison pill” that former American Psychological Association president George Albee says the mental health profession ingested several decades ago. With this pill, we swallowed the deeply flawed medical/psychiatric assumptions about diagnoses and dosages, culminating in the unrealistic expectation that forms of psychotherapy can be administered with the reliability of, let’s say, a surgical protocol. The belief that this level of consistency can be obtained derives from a serious confusion of models–what philosopher Gilbert Ryle called “a category mistake.” In other words, psychotherapy has been misclassified; it should never have been considered a treatment in the first place. Rather, it’s a specialized form of inquiry–more philosophical journey than medical procedure.

In fact, if Jungian James Hillman had his way, the therapy enterprise would be categorized “as an art form rather than a science or a medicine.” At root, therapy is just two people conversing. That would be evident if you peeled back the layers of mystification and simply listened to a therapy tape. “Consciousness,” says Hillman, “is really nothing more than maintaining conversation, and unconsciousness is really nothing more than letting things fall out of conversation.”

The derivation of the word conversation is worth examining: it comes from vertere (to turn) and con (with). This is a perfect metaphor for the therapeutic process. As Hillman describes it, you “walk back and forth with someone . . . turning and going over the same ground” from a variety of directions until “what we already feel and think [has been converted] into something unexpected.” To be effective, therapeutic talk must have an edge: “It opens your eyes to something, quickens your ears, . . . and keeps on talking in your mind later in the day,” adds Hillman, and, hopefully, for days to come. Instead of talking of cures, the therapist’s job is to “cure our talk.”

A week ago, a client came in for her first session. She described the frustrations of having engaged in years of self-improvement efforts–meditation classes, body disciplines, empowerment groups–only to find herself as confused as ever about whether to stay in her marriage and if a career shift might bring greater fulfillment. At that moment, I found myself contradicting Socrates, telling her that sometimes the unexamined life is worth living. That statement struck a chord. She lit up, laughed, relaxed into her chair, and said, “Thanks for saying that.”

My remark couldn’t have been preplanned and wouldn’t necessarily have been appropriate for either a different client or even the same client at a different time. Like a thousand other such bits of conversational ingenuity, it worked because of its positioning in the ongoing dialogue–it was exquisitely responsive to the several layers of meaning contained in the client’s communication.

It’s practically impossible to explain how such comments are generated or exactly what clients make of them, yet the immediate reaction and subsequent discussion in this case made it clear that comments such as these catalyze important shifts in perspective and advance the collaboration inherent in therapy. Like Mostel’s bit of chicanery, such spontaneous remarks are not learned from manuals and can’t be dispensed on demand, yet they’re the essence of the therapist’s conversational craft.

Our recent informal survey of real-world practitioners–the folks who actually make their living seeing clients–suggests that most therapists don’t use cookbooks, don’t place their faith in techniques, and don’t pay much attention to what’s on the latest list of validated treatments. Moreover, the longer they’ve been in practice, the less their treatments resemble the rule-bound procedures they learned in school. Experienced clinicians intuitively follow Hillman’s advice to avoid fixed positions, realizing that any prejudgments can “stop conversation dead in its tracks”–leading to a sterile monologue rather than a productive dialogue. When that happens, you might just as well send a memo.

This article first appeared in the November/December 2005 issue.

Jay Efran

Jay Efran, PhD, professor emeritus of psychology at Temple University.  He received the Pennsylvania Psychological Association’s 2009 award for Distinguished Contributions to the Science and Profession of Psychology and is co-author of Language, Structure and Change and The Tao of Sobriety.


Mitchell Greene

Mitchell Greene, PhD, is in private practice in Wayne, Pennsylvania.