Scientific Freud

In this month’s American Journal of Psychiatry: The Efficacy of Cognitive-Behavioral Therapy and Psychodynamic Therapy in the Outpatient Treatment of Major Depression: A Randomized Clinical Trial. It’s got more than just a catchy title. It also demonstrates that…

Wait. Before we go further, a moment of preaching.

Skepticism and metaskepticism seem to be two largely separate skills.

That is, the ability to debunk the claim “X is true” does not generalize to the ability to debunk the claim “X has been debunked”.

I have this problem myself.

I was taught the following foundation myth of my field: in the beginning, psychiatry was a confused amalgam of Freud and Jung and Adler and anyone else who could afford an armchair to speculate in. People would say things like that neurosis was caused by wanting to have sex with your mother, or by secretly wanting a penis, or goodness only knows what else. Then someone had the bright idea that beliefs ought to be based on evidence! Study after study proved the psychoanalysts’ bizarre castles were built on air, and the Freudians were banished to the outer darkness. Their niche was filled by newer scientific psychotherapies with a robust evidence base, such as cognitive behavioral therapy and [mumble]. And thus was the empire forged.

Now normally when I hear something this convenient, I might be tempted to make sure that there were actual studies this was based on. In this case, I dropped the ball. The Heroic Foundation Myth isn’t a claim, I must have told myself. It’s a debunking. To be skeptical of the work of fellow debunkers would be a violation of professional courtesy!

The AJP article above is interesting because as far as I know it’s the largest study ever to compare Freudian and cognitive-behavioral therapies. It examined both psychodynamic therapy (a streamlined, shorter-term version of Freudian psychoanalysis) and cognitive behavioral therapy on 341 depressed patients. It found – using a statistic called noninferiority which I don’t entirely understand – that CBT was no better than psychoanalysis. In fact, although the study wasn’t designed to demonstrate this, just by eyeballing it looks like psychoanalysis did nonsignificantly better. The journal’s editorial does a good job putting the result in context.

This follows on the heels of several other studies and meta-analyses finding no significant difference between the two therapies, including, another in depression, yet another in depression, still another in depression, one in generalized anxiety disorder and one in general. This study by meta-analysis celebrity John Ioannidis also seems to incidentally find no difference between psychodynamics and CBT, although that wasn’t quite what it was intended to study and it’s probably underpowered to detect a difference.

(other analyses do show a difference, for example Tolin et al, but the studies they draw from tend to be much smaller than this latest and in any case are starting to look increasingly lonely.)

Suppose we accept the conclusion in this and many other articles that psychodynamic therapy is equivalent to cognitive-behavioral therapy. Do we have to accept that Freud was right after all?

Well, one man’s modus ponens is another man’s modus tollens. The other possible conclusion is that cognitive-behavioral therapy doesn’t really work either.

If parapsychology is the control group for science, Freudian psychodynamics really ought to be the control group for psychotherapy. Although I know some really intelligent people who take it seriously, to me it seems so outlandish, such a shot-in-the-dark in a low-base-rate-of-success environment, that we can dismiss it out of hand and take any methodology that approves of it to be more to the shame of the methodology than to the credit of the therapy.

But what about the evidence base for cognitive behavioral therapy over placebo? Or, for that matter, the evidence base for psychoanalysis over placebo?

Part of the problem may be what exactly is used as placebo psychotherapy. In many studies, it’s just getting random people to talk to patients. This makes intuitive sense as a placebo therapy, but it seems vulnerable to unblinding – people usually have some expectation of what psychotherapy is like, and undirected conversation about problems might not match it. Or if the placebo therapists are not professionals, they may be less confident in talking to people about their mental health problems, more awkward, less charismatic, or otherwise not the sort of people who would make it in the therapy profession. So now a lot of people are coalescing around the idea that all therapy studies done against these kinds of placebo therapy are fundamentally flawed.

Studies that compare what are called “bona fide psychotherapies” – two therapies both done by real therapists with real training – tend to have a lot more trouble finding differences. This has led to what is called the Dodo Bird Verdict, after an obscure Alice in Wonderland reference I feel vaguely bad for not getting: that psychotherapies work by having a charismatic, caring person listen to your problems and then do ritualistic psychotherapy-sounding things to you, but not by any of the exercises or theories of the specific therapy itself.

Then the question becomes: if the Dodo Bird Verdict and the active placebo problem and so on are equally true of all psychotherapies and all psychotherapy studies, how come everyone become convinced that cognitive behavioral therapy passed the evidence test and psychoanalysis failed it?

And the answer is the CBT people did studies and the psychoanalysts didn’t.

That’s it. It may be, it probably is, that any study would have come back positive. But only the cognitive behavioral people bothered to perform any. And by the time the situation was rectified and the psychoanalysts had (positive) studies of their own to hold up, “everyone knew” that CBT was evidence-based and psychoanalysis wasn’t.

This seems like another case of doctors not understanding that there are two different types of “no evidence”.

I should qualify this sweeping condemnation. I believe a few very basic therapies that address specific symptoms in very simple ways will work. For example, exposure therapy – where you treat someone’s fear of snakes by throwing snakes at them until they realize it’s harmless – is extremely and undeniably effective. Some versions of CBT for anxiety and DBT for borderline also seem to just be basic coping skills about getting some distance from your emotions. I think it’s likely that these have some small effects (I know a study above found no effect for CBT on anxiety, but it was by a notorious partisan of psychoanalysis and I will temporarily defy the data).

But anything more complicated than that, anything based on an overarching theory of How The Mind Works, and I intuitively side with the Dodo Bird Verdict. And I think the evidence backs me up.

EDIT: Do not stop going to psychotherapy after reading this post! All psychotherapies, including placebo psychotherapies, are much better than nothing at all (kinda like how all psychiatric medications, including placebo medications, are much better than nothing at all).