The Empirical Status of Empirically-Supported Psychotherapies
Two recent journal articles stridently discuss different aspects of this issue. The first is by Drew Westen, Catherine M. Novotny and Heather Thompson-Brenner, “The Empirical Status of Empirically Supported Psychotherapies: Assumptions, Findings, and Reporting in Controlled Clinical Trials,” (2004), Psychology Bulletin, 130(4), 631 – 663. The second is by Timothy B. Baker, Richard M. McFall and Varda Shoham, “Current Status and Future Prospects of Clinical Psychology – Toward a Scientifically Principled Approach to Mental and Behavioral Health Care,” (2008), Psychological Science in the Public Interest, 9(2), 67 – 103.
Westen et al. review the assumptions and findings of studies establishing psychotherapies as clinically valid. They characterize these as “empirically-supported therapies” (ESTs). The distinguishing feature of an EST is that it has been or can be corroborated using randomized controlled trial (RCT) methodology. An example of an EST is CBT. RCT methodology, however, imports a number of problems, such as: limited sample size, participant self-selection, iatrogenesis, attrition and the impossibility of eliminating confounding variables. It assumes psychopathology is highly malleable, that patients are not comorbid and can be treated for a single problem or disorder, that personality is irrelevant or secondary in the treatment of psychiatric disorders and that experimental protocols are the only way to evaluate whether a therapy works. Based on these and other considerations, Westen et al. conclude most EST’s are not quite as empirically-supported as they may seem. In fact some psychotherapies typically thought of as lacking empirical support (such as IPT or even psychotherapy) are or can be just as effective as those backed by EST claims, despite their methodological shortcomings (such as non-random assignment of patients and lack of experimental control).
Baker et al.’s paper takes the opposite point of view. From a public policy standpoint, interventions must be efficacious, disseminable, cost-effective and scientifically plausible. The only treatment meeting these criteria is CBT, for example, when used to treat tobacco addiction or depression. All other treatment’s are “pre-scientific” and therefore of dubious validity. In fact the discipline of clinical psychology itself may be conceptually suspect because as an applied science it is insufficiently grounded from an empirical standpoint.
Both Westen et al. and Baker et al. are asking the wrong question. They assume psychology should aspire to a deductive nomological model of explanation, along the lines of physics or chemistry. Rational accounts of human behavior, though, are not deterministic in form; that is, they are not logically deductible from a specified set of causes.
For example, I voted for Obama in last year’s Presidential election because I thought (maybe wrongly in retrospect) he would do a better job of solving problems with the economy. Suppose there is an absolutely exceptionless universal generalization about people like me to the effect I invariably will vote for the Democratic party candidate. This law doesn’t explain why I voted for Obama. I just as easily could have voted for his opponent. All it does is state a regularity, not explain anybody’s behavior. This situation is fundamentally unlike (say) Boyle’s Law or Charles’s Law, which actually establish the causal relationships between pressure, temperature, and volume of gases.
One might consider a proposed course of action in advance, weighing the pros and cons, but doing so doesn’t transform them into “causal” factors. Even though one has justified convictions about what to do, one easily could have acted otherwise. One acts on the basis of reasons, but these reasons are not a “vector” of forces.
One also might argue a certain decision is “caused” by neuron firings and the neuro-chemical transmission of information along axons to dendrites, etc. While this trivially is so, no social or psychological phenomena perfectly mirror molecular movements. There is an indefinite range of stimulus conditions for any psychological state. Brain anatomy doesn’t “map” onto psychological outcomes or systematically correlate with them, and there are no “bridging” principles to get from one to the other. To continue with the Obama example, there never will be laws of elections like there are laws about gases.
More broadly, psychology has a radically different explanatory style than does physics or chemistry (best characterized as “hermeneutic,” that is, grounded in history and context). It is not a propositional calculus or series of logical inferences and it shouldn’t aspire to be something it’s not. For example, early psychoanalysis was unsure of its scientific standing and therefore tried to ground itself in pseudo-scientific concepts such as hypnosis. Freud constantly was trying to defend psychoanalysis as a science because he wanted to give a supposedly scientific account of human behavior. See, e.g., George Makari (2008), Revolution in Mind – the Creation of Psychoanalysis, p. 298. These efforts were misconceived because of the intrinsically mental character of psychological phenomena. Freud’s explanations were not scientific, rather, common-sensical.
This critique certainly is not original with me. One of the best explanations of it is in John Searle’s book Minds, Brains and Science (1984), esp. chapter 5. Searle builds on the work of Donald Davidson (“Philosophy of Psychology”) and Charles Taylor (“Interpretation and the Sciences of Man”). Westen et al. and Baker et al. arrive at such polemical and mutually incompatible outcomes because neither of them consider this fundamental problem.