Formulating Formulation

This month I have a review in Clinical Psychology Forum of Lucy Johnstone and Rudi Dallos' standard text "Formulation in Psychology and Psychotherapy" (thanks to the editors at CPF for the invitation!). Doing the piece was an opportunity to spend more time than usual thinking about formulation; what it is and how it is supposed to be helpful. However, the resulting article is just 400 words; space enough only to develop a few ideas, and not in much depth. This post is a more extended reflection on the book and the ideas it prompted.

My review is principally positive. I praise the book for being practical (the new chapters especially deal with issues that can only have been identified my psychologists in multi-disciplinary teams in the NHS); for being ambitious (psychologists are given excellent advice on what to expect when formulating in teams) and unique (there is no other book that meets the needs of NHS psychologists in this way). Here's an overview and elaboration of conceptual points in the CPF Review I wished I could extend:

The Validity of Formulation:

The book is not afraid to address concerns about formulation's validity. Formulation is not "evidence-based", (partly because it is difficult to know what that would look like) and although there have been attempts to hold it to an external standard (coding formulations' content for quality), that simply raises the question of how to empirically validate the standard. I wonder if formulation-enhanced therapies would be amenable to validation by RCTs (therapies that are vs. therapies that are not guided by a formulation), but that may be unworkable.

However, as is pointed out, any given formulation is less an entity, diagnosis or instrument (which can be easily tested for reliability etc.) than a process for making inferences and predictions. Inferences are more or less valid, depending on their premises and how they are drawn. Predictions can be more or less valid/useful depending on how testable they are. This brings us to an important role formulation could have in a "local clinical scientist" model of clinical psychology.

Gillian Butler’s statement that a formulation is a set of “hypotheses to be tested” is often cited in the literature on formulation, and this text is no exception. A hypothesis can be tested if it can theoretically be falsified, so in the review I wonder whether formulation based on the principles of “risky prediction and refutation” could play a role in yielding valid psychological knowledge.

"Usefulness vs. Truth"

Another key issue running through the text is the distinction between “usefulness” and “truth”. The implication seems to be that formulation is about the former while diagnosis aspires to the latter. This seems to reflect a leaning toward a "constructivist" theory of knowledge (a respectable enough position) but I don't think we need a strong commitment to that position. On the contrary, I suspect “truth” tends to bring “usefulness” along with it, such that the more we know about a particular individual (and they about themselves) the more able we are to help them.

However, I assume the "usefulness vs. truth" distinction is made because psychologists want to avoid "imposing" their felt certainty on their clients. One way to do this is to say (quite correctly) that there are no basic, easy-to-articulate "truths" about human experience. We make our own meanings for ourselves, and any health professional should have respect for this project. So far I agree. However, things get tricky when we want to extend this way of thinking beyond existential-phenomenological "truths" and into the realm of aetiology. If formulation were just a meaning-making process, a constructivist theory of truth would go most of the distance, but it is not. Formulation also purports to be a description of a person's problems with an inbuilt theory about how they came about. Some stories will be closer to than others to reality, and we shouldn't be afraid of that.

If this sounds like I am advocating an authoritarian view of how to do therapy, I am not. While there is probably something like "the truth" about causes, it is vital that practitioners stay in touch with the tremendous uncertainty we have to face in knowing, for any given person, what that is. Such knowledge as we do have is based on the proportions of variance derived from large N samples. It does not generalise straightforwardly to an individual, whom we have to take with all the idiosyncrasy, and uncertainty, they deserve. This uncertainty is characteristic of science, not inimical to it, so for me the spirit of much of this book is perfectly compatible with a "scientist-practitioner" model.

We can then, believe in the value of multiple constructions AND in a basic underlying reality. The text itself makes a similar point. In the discussion in Chapter 10, "Using Formulation in Teams", the point is made that clinicians can be insufficiently-aware of the prevalence of sexual abuse, leading to the failure to consider this as a factor. Here is a stark fact about reality we are encouraged to face; abuse happens, it has often has devastating consequences and formulation should acknowledge it wherever appropriate. There are obviously limits upon how constructivist to be about abuse (no-one seriously advocates the construction of a truth in which we pretend it hasn't happened), and the same is presumably true for other aetiological factors. The reason all the chapters in this gloriously eclectic text can be helpful is that they can help the clinician formulate different aspects of social and psychic reality in different ways that need not be mutually exclusive.

All White and Predominately Female: Clinical Psychologists

So much for the points in my CPF review. In addition to these ideas, I have some other broad thoughts on the place of formulation within clinical psychology, which I will get to below, but before I do, here are two problems I had with aspects of the book:

1. "Primitive" Defences:

To my mind formulation is at least partly about re-imagining what is going on when people approach mental health services. Whatever you think about diagnosis, it is a fact that it represents an extremely limited way of describing people, and one which alienates many of its recipients. Formulation affords the luxury of a more open and inclusive language; a way to aid the clinician in their understanding of subjective experience. For this reason I am surprised at the continued use, in Rob Leiper's chapter on Psychodynamic formulation, of the term "primitive defences". Sure enough this terminology reflects a longstanding tradition in psychoanalytic discourse, but it seems clearly pejorative, at least as much so as any DSM-diagnosis. What is more, it is increasingly outmoded. In her book on psychoanalytic diagnosis, Nancy McWilliams adopts "primary" defences rather than "primitive", a preferable option to simply placing ironic quote marks around the latter word, as though that were enough to mitigate its influence on the clinician's thinking. Even then the implied developmental trajectory ("primitive" is supposed to mean "developmentally primitive") is probably bogus, as Drew Westen pointed out 25 years ago.

2. "Medical" vs. non-medical:

One of the difficulties in thinking about "psychiatric diagnosis" is to know what we are and are not talking about. DSM-diagnoses obviously fit the bill, but what about other categorisations? Is Judith Herman's "Complex PTSD" a psychiatric diagnosis? What about neologisms that might arise from the patterns which formulations reveal? In Johnstone's final chapter, "trauma reaction" is approvingly suggested as a useful linguistic shortening-but how do we recognise one of those when we see it? If there are criteria for "trauma reactions", and if an aetiology is strongly endorsed, in what way is this different from psychiatric diagnosis? We might say that psychiatric diagnoses are those which posit an "underlying" disease mechanism, except this is false. Psychiatry is plainly interested in entities which are not considered "diseases" by any metric. Perhaps then psychiatric diagnosis is anything which is done by a psychiatrist, but if psychiatric diagnosis is undesirable by definition this seems a little unfair on that profession, whose practitioners are damned whatever they do.

The distinction persists in Johnstone's chapter, which contains a discouragement against using the shortened formulations of psychoanalytic character diagnosis (Obsessional personality; Narcissistic personality etc.). These terms closely resemble formulations in that they put an emphasis on the ways that life events train someone to become the character they are, so why are they ruled out? Because they are "medical" Johnstone claims. Given the extent to which psychoanalytic clinicians lament the rise of the medical model as a challenge to their own approach, calling their system of diagnosis "medical" strikes me as contestable. True enough their descendants, the personality disorders, have found their way into the DSM, but that tells us more about the powerful influence psychoanalysis still had in 1980s American psychiatry than it does about the impact of the "medical model". There are good reasons to be wary of character diagnosis (alongside the usual questions of validity are very real concerns that the terminology is rather insulting) but its putative "medicalness" is not one of them.

Those two issues both get to the heart of why I started this blog, but perhaps they are marginal when it comes to the business of really understanding and articulating the role of formulation. Here is my final section, containing a reflection on the book and the thoughts it prompted about why psychologists should "formulate", and why this book can help them:

Generative Thinking and Eliminative Thinking: In Defence of Formulation:

There is a distinction to be made between the generation of ideas, and their validation. It is a distinction that Herbert Feigl talks about in his 1949 paper Philosophical Embarrassments of Psychology:

The clearest way I can think of to articulate the value of formulation is in terms of how it helps the clinician to generate ideas. When we do therapy with people we would like to know when our pet theories are right and when they are wrong. The standard caricature of Freud is that he believed everything came down to sex. If you took seriously his ideas about the dynamic unconscious (that an idea could stand for its opposite and denial by an analysand can really represent confirmation), you could accommodate virtually any evidence into his scheme. Hence Karl Popper's famous idea about theories needing to be refutable in order to be testable.

However, much as we need a system for eliminating possibilities, we also need a system for generating them. This is what struck me about the variety of this book, which references multiple psychological frameworks, including one (Personal Construct Psychology) that I hadn't heard from since it was referenced in an undergraduate social psychology class. As people who work with people, we need to be able to think our way into the experiences of others, and to be imaginative in understanding why they have the problems they have. Minds are weird and elusive, experiences often half digested or unformulated. Out of this confusion, a therapist seeks to draw some order.

However, our personal frame of reference, our own system of metaphor and interpretation will almost certainly be entirely inadequate to this task. What we need is to listen to people with the utmost respect for their own "construction" of their lives, and with a willingness to jointly forge sense. Where people struggle to make sense of their own experiences, we need to have the flexibility and imagination to frame things in ways that can help. We need to be open to being wrong, and we need to be open to putting things differently. This sort of task is something this book is superbly set up to encourage, and is surely its great strength.

In Favour of Objectivism about Psychotherapy Outcomes

A lot of people in my field dislike the use of quantitative measures to determine the value of what they are doing. Some of the emotional intensity of this view can be seen in the conversation I had on Twitter after posting a link to an article by Richard Gipp. The idea seems to be that objective measures "miss" something that can only be framed in language. People's psyches are fragile and complicated, and using a numerical scale is somehow riding roughshod over this, or doing to violence to the subjectivity of the other. This post is a response to (though not straightforwardly an argument against) Gipps' piece, which contained many fascinating points I don't really speak to here. My aim is to persuade you that "objectivist" approaches to measuring psychotherapy outcome are a good thing.

I am not making a rational argument (though I hope it's not irrational); that has been done consistently over the past 60 to 70 years and it is fairly widely acknowledged that numerical information has its own self-contained logic. This is an attempt at polemic. I want to convince you at a gut level that the use of numbers works for people who use services, that it can actually be quite noble in all the ways that certain forms of clinical writing claim to be. Furthermore, I want to suggest that, for all its value, the most poetic and optimistic clinical writing can act to conceal reality in important ways, potentially giving a veneer of respectability to processes which have little meaningful impact. Conversely, just because there is something rather prosaic about the notion of an "evidence-based-therapy", it is in fact perfectly compatible with all the beauty and subtlety we see in more "subjectivist" approaches.

In regard to the first point, the promise of insight, self knowledge or a deep connection with someone else does not necessarily carry along with it the promise of "feeling better" in important ways. I have learned a lot about myself through experiences of psychodynamic therapy or supervision, but to some degree one can separate the process of self-understanding (itself, in my view, an extremely valuable thing) and the process of feeling substantially less rotten about one's life. Although I find it invaluable to have the head-space to wonder about my relationship to my desires and my personal history; to think about the way they rebound in the minutiae of my social interactions in the present, I am not always sure how essential these are for my capacity to continue getting out of bed in the morning or avoid feeling like I want to kill myself.

Getting somehow "better" (and better is necessarily a vague word, in psychotherapy outcome research. It has ended up meaning whatever is indicated in the questionnaire you choose: "less anxious"; "less depressed" by a certain number of points on a scale) is not always a beautiful process. Something could be good for you psychologically without necessarily reaching your subtlest places. Some of the times I have made the most important changes in my well-being or happiness have been of the "crass" variety; events I could re-describe as "behavioural activation" or "cognitive restructuring". These are not Orwellian portents of a psychotherapy devoid of the human factor, they are clumsy attempts to describe ways we can change ourselves (in the right context) rapidly and  effectively .

More importantly, there is simply no incompatibility between a blunt (but objective) measure of psychotherapy outcome and a fine-grained "appreciation" of the unique and subjective aspects of the experience. Something could be useful in the crass numerical sense (indexed by a clinically interesting drop in an Anxiety measure) while also being moving, poignant, invigorating, thought-provoking, inspiring and so on.

To assume otherwise is to place a peculiar store in one narrative version of events as though "the truth" about someone's subjectivity resides in one place and one place only. Truth is trickier than that. What is "the truth" about a person's experience of a depressive breakdown; their narrative of events? The narrative of the people they love? A measure of their moods on a series of psychological questionnaires? Or does the truth lie somewhere beyond all these, in an inarticulate mesh which can only be variously approximated by different representations?

One critical response to the varieties of a treatment like CBT is to regard them as a form of authoritarian "training" or "brainwashing". This is to take the language of CBT too seriously, to imagine that your experience of that language is identical with the experience of the therapeutic relationship in which its techniques are deployed. "Cognitive restructuring" sounds quite mechanistic, but depending on how it is conducted it can mean something closer to "helping someone consider alternative readings of their situation" or "expanding a person's psychic possibilities". One could experience cognitive restructuring and find the experience elevating and beautiful. One could experience it as an intrusion by an idiot who has no appreciation for how difficult your life is. Or your reaction could lie somewhere between those extremes.

Precisely the same is true for any modality of therapy. Psychoanalysis (for some reason the most poetically described form of therapy) can be constructed as a disintegration and reconstruction of the ego through a profound subjective attunement between analyst and patient. It can be beautiful (it can certainly be written about beautifully by clinicians). It can also be experienced as sadistic, or as pointless (For some reason this perspective is more frequently seen in the narratives of patients like Susanna Kaysen in "Girl Interrupted", or Jenni Diski than in the writing of psychoanalysts themselves).

These disconnects are why I like numbers and think you should too. Efforts to represent the truth are all around us. Although patient narratives ought to be taken very seriously, in clinical professional circles the narratives of practitioners (every clinical psychologist has a copy of Irving Yalom somewhere at home) are generally given more credence. Given this fact, it is nothing short of wonderful if a healthcare system can be organised around a system of aggregating numerical indices of individual experience. We can, if we want, say that quantitative measures are an impossible attempt to objectify the subjective, but their value doesn't rely on this vaunted ambition. Numbers are a simple and systematic language. You indicate how you feel at time 1, and when time 2 comes along, any observable difference can easily be registered. It is much harder for clinicians to deny the fact that more people report feeling worse at time 2 than it is for them to re-construct the whole encounter so as to undermine any testimony that doesn't fit with their own.

None of this is to say that quantitative measures should "trump" patient narratives, but the way that different decisions are made will mean that different sorts of information are prioritised. When a healthcare organisation decides to fund a treatment, it has to know whether said treatment represents a meaningful improvement over something cheaper. Numbers are easier in making this determination. However, when deciding if mistreatment is occurring in a care setting, a single person's narrative ought to all that is needed for substantive concerns to be raised. In the quotidian process of navigating psychotherapy, a clinician cannot hope to proceed without attending, to almost everything done or said by the person with whom they are working.

When it comes to statistical data, it is hoped that we can learn to love the numbers which many people find so alienating. Behind each number lies a person, so used correctly they are an excellent way of finding out important information about real lives and real experiences. Ideally we would be able to think in two registers at once; deploying the skills of the most technically competent statistician and the empathy and interpretative nous of a novelist or poet.