For example, someone who claimed that what really did the work in CBT was changing pathogenic beliefs might be confounded by a study which showed the relevant changes in mood occurring after the behavioural activation phase, but before the cognitive restructuring phase, was introduced. ... In fact, once you start thinking and talking in the inhuman way I contrivedly did in the previous sentence, the idea of empirical research into 'how it actually works' starts to seem perfectly straightforward. What might not be straightforward is the answer, or the methods needed to reach it - but the question, that at least, supposedly, is perfectly clear.
Well, I don't think it at all clear. I think the appearance of clarity to be an illusion undergirded by latent conceptual confusion. In this post I spell out why I think this. My claim will be that for the question to find intelligible application we must surely be able to separately specify process and product. But when it comes to psychotherapy - in fact, when it comes to most everything of interest in psychology - we can't actually do this. Sure, we can create impoverished measures of success which don't make explicit reference to the therapeutic process, or banal measures of the process which don't make explicit reference to the product. However these, I claim, only give us the appearance of the kind of externally related relata which can cogently function as independent variables in a causal explanation. (Just because you give a rather thin description of something which doesn't overlap with another rather thin description of what is only allegedly a separate thing doesn't mean that you can then intelligibly relate the two through a causal explanation.)
Example. Marjorie's existence is misery-saturated. What does this misery consist in? When we and she get to know her better what we both come to understand is that she rather moralistically doesn't allow herself certain of her ordinary human feelings. Feelings of illness are morally judged as indulgent and as in any case unsafe to acknowledge since they may betoken serious unmanageable disease. Feelings of both her reasonable and her somewhat childish anger are tacitly judged as simply too shameful; no understanding is shown towards herself in them. Her hopelessness is a function of her sucking the imaginary poison out from fate's sting before the real, and inevitably more benign, deal comes anywhere near her. And so on, you get the picture. In all these ways she suppresses herself, offers herself no understanding or encouragement, and takes no courage in facing the challenges of life without having first denatured life in her imagination. Her misery is a function of this unwitting auto-restriction of her life-energy and her hope and her spontaneous emotionality.
Marjorie's therapy looks like this. When she automatically acts in one of the above self-defeating ways in the therapy session, or when she reports having thoughts and feelings during the week which, on reflection in the therapy session, can be seen to be fit one of the above self-defeating ways, then the therapist gently and firmly calls her on it. Then Marjorie can see what she has been doing. But this isn't just a bit of intellectual self-knowledge that she gains. It is itself a liberatory experience: in being called out for her relentless fusion with her miserabilist superego, she experiences emancipation from it. She de-fuses from it. If she didn't de-fuse from it then she wouldn't be having a living understanding of what was being talked about. She would just be knowing that a sentence composed of certain words was true of her, without really understanding their significance for her. (Sometimes this pseudo-knowing is called 'intellectual insight'; my own view is that there is no such thing as merely intellectual insight.)
Do these de-fusings take? Does Marjorie come to fully internalise her therapist's way of relating to her (Marjorie's) feelings - and thereby cultivate a more benign and unstressy superego? Well, let's suppose, to some degree: yes they do, yes she does.Now, what would it mean to question, here, how the therapy works? What we'd need to do, to get our question off the ground - if, that is, we're pursuing it in a hypothesis-testing mode - is to parcel off our understanding of therapeutic action from our understanding of therapeutic boon. Thus we could think of the therapeutic action in terms of a 'therapeutic intervention' (the word 'intervention' is already nice and causal-sounding; that should rhetorically help the cause-effect game get into play), and we could also think of the therapeutic boon as a remission of 'depressive symptoms' (the word 'symptom' is already nice and distal-sounding, just ripe for taking the place of a product, an effect, of some otherwise-to-be-specified cause). We could measure these therapeutic 'interventions' (for example, we could count the number of interpretations that are offered!) and these depressive 'symptoms' (using the BDI, for example). And then we could see if these 'interventions' and these 'symptoms' are correlated. And if they were correlated then we could posit a causal relationship between them. ... ... ...
But why on earth would we want to do any of that - apart from to try and ape the procedures of natural science? For we already know full well that life isn't in any way like this. We know that it isn't the quantity of the interpretations given which makes them mutative. I can take in what you say to the degree that I can trust you, to the degree that I'm prepared to risk climbing out from under the thumb of my superego, to the degree that I can be touched and moved. This can sometimes take a long time to develop. It can take patience, perseverance, and courage. Sometimes the less you 'interpret', the more effective you are - it depends on what's live in the relationship between the two of you. But the point of what I'm saying is not to plead for complexity - not to say 'but there's all these other factors we will need to take into account in the causal model'. It is rather to question the very idea of the intelligibility of a causal model of therapeutic action. What I am saying is: the only meaningful description of the so-called 'intervention' is of a therapist saying something that actually touches the patient in such a way that they do start to see things differently, do step outside of their enmeshed relationship with their own inner critic, etc. And the only meaningful description of a life without depression is of a life without these vitality-stultifying defences ruling the interior roost. After all, life can be hard, painful feelings need to be felt, losses mourned, hopelessness endured. It's not the affects but the defensive relation to the affects which matters in stepping outside of depression and into authentic relationship with whatever is in the offing in one's inner and outer life.
I don't really expect to convince anyone with this post. My purpose in writing it has however been to start to articulate a very different vision of what it might mean to ask 'how does psychotherapy work?' The answer we can give which is actually intelligible is, I believe, one which includes within it an apt characterisation of the nature of the problems the patient faces, and includes within it a characterisation of the nature of the relationship with the therapist within which understanding is reached. Here understanding does not mediate change - for understanding is change. (If I truly come to understand myself anew, this is itself always-already transformational - and there's no need to hold onto a here-inappropriately third-person-type conception of understanding or knowledge as extrinsic to the being of that which is understood or known. 'Know thyself' is not an instruction to accumulate more true beliefs about yourself! It's already a matter of not being alienated from yourself, already about becoming more self-possessed.) The therapeutic relationship - i.e. the therapeutic aspect of the relationship between therapist and patient - is also not a vehicle for or mediator of change, but rather its locus. Being touched by the therapist, internalising (not, nb, learning facts from!) what she is saying, this is 'therapeutic action'. And in therapeutic action we don't have a cause followed by an effect. We have a transformational moment within a relationship.
Sure, it's important to go to therapy. But going to therapy does not by itself cause therapeutic improvement. You can go to therapy all you like and not get anything out of it. Once again it would be meaningless to correlate going to therapy with going better and say that the former must have caused the latter. We already know that going there isn't going to do anything except accidentally ('behavioural activation' or 'getting off your arse'). For you have to engage with the therapy. You can't have a change of heart unless your heart is in it to start with.
So, as I see it, the question 'how does psychotherapy work?' does not receive an empirical answer - it turns out to not have been an empirical question. At the end of our quest we might reach a better phenomenological explication of just what therapy is. That I'll take any day. But the component analysis or what have you - that we can hand back to the natural scientist.