Monday, 15 October 2012


neither top down nor bottom up

Reading the cognitive neuropsychiatric literature on delusion one is often offered the following two choices, nicely summarised by Lisa Bortolotti in her Delusions and Other Irrational Beliefs (p. 29).
Empiricists argue that the direction of causal explanation is from the experience to the belief. Delusions involve modifications of the belief system that are caused by 'strange experiences', in most cases due to organic malfunction (Bayne and Pacherie 2004a; Davies et al 2001). These accounts are also referred to as bottom-up (from experience to belief).
Contrast the:
Rationalists about delusion formation [who] argue that delusions involve modifications of the belief system that cause strange experiences. This [top-down] thesis, where the direction of causal explanation is reversed, has been proposed for monothematic delusions such as Capgras (Campbell 2001; Eilan 2000; Rhodes and Gipps 2008) and for delusions of passivity, in which the subject experiences her movements, thoughts or feelings as controlled or generated by an external force (Sass 1994; Graham and Stephens 1994; Stephens and Graham 2000).
No doubt these could be perfectly reasonable options for explaining why sometimes certain people come to maintain, at least passingly, some rather strange things. (I don't however think that Campbell, Eilan or Rhodes & YoursTruly are best characterised as being concerned primarily to explain atypical monothematic delusions (although the first two do touch several times on these, perhaps because they have been discussed so extraordinarily frequently in the philosophical cognitive science literature) - we all seem interested mainly in more (proto/)typical schizophrenic delusions - but  this doesn't matter for now; what follows can however be read in part as querying whether my own views are best described as 'rationalist' or 'top down'.) But it is, I think, pretty hard to see how either of them is supposed to work in understanding schizophrenic delusion. Because this discussion about top-down and bottom-up has become somewhat hegemonic, it can often seem that we simply must choose one of these two options, or offer some kind of hybrid account. In this post I will be suggesting that the discussion presupposes too much about the nature of delusion - in particular that it involves some kind of a problem in belief formation and maintenance - and that we need to look elsewhere to understand it adequately.

The empiricist top-down option is usually cashed out as: the patient is either automatically 'endorsing' their odd experience, or they are going on to further 'explain' it using their reason. Endorsing by itself won't explain why anyone cleaves to what their wonkified experience seems to be telling them. Hence the 'two factor' approaches of Max Coltheart and Martin Davies who suggest that we need to supplement naively believing in weird experiences with continuing to reflect weirdly on this experience. The weird reflection that Anne Aimola Davies and Martin Davies propose amounts to a failure of working memory and/or executive function caused by right frontal damage.

One objection to this is that this failure seems too highly selective (to the delusional scheme). Given that the patient spends a lot of time and energy preoccupied by their delusion one might think that only selective deficits would be less, rather than more, likely to show up here. Another is that schizophrenic patients have of course not at all inevitably suffered lesions in their right frontal cortex. But my main objection draws on something I remember Matthew Broome saying to me about 10 years ago when we were talking about those probabilistic reasoning experiments (to do with proportions of coloured beads in a bag) that used to be popular in this field. 'But, after all, why would someone believe in such wonkified experiences?' I asked. 'Well, perhaps it is because they are mad', Matthew said. And something like this seems intuitively right to me. 'You would have to be mad, unhinged, to believe that!' is our intuitive response to the question of why someone with schizophrenia buys into their unusual experiences with their unusual beliefs.

The point of saying this, of course, is not to thoughtlessly throw round derogatory words like 'mad' or 'unhinged', but rather to urge that, if we are honest, we must admit to still being as puzzled about the delusionality of the delusion as ever. Odd experiences and deficits in working memory and executive function are all very well, but 'how can someone really believe something as odd as that!' is the question that still seems to me to remain after the empiricist bottom-up explanations have run their course. (Of course there's a difference between a phenomenological explication of the delusionality of the delusion, and a causal explanation of how it arises, but the latter explanation must be an explanation of how it - this thing with these properties - arises.)

What about the 'top-down' alternative? Well: what is this alternative? The idea that our beliefs shape our experiences is, stated like that, far too vague to really mean anything at all. In one form or other it will be self-evidently true, blatantly false, or just nonsense. So here is one reading of it which I think captures something of what is typically meant. The idea (Campbell 2001) is that we have a set of basic framework beliefs which partly constitute the meanings of the associated terms whilst not themselves facing the tribunal of experience. (I believe that the world has been here for a jolly long time; my believing this is not to be taken as dependent on a separable grasp of the meaning of 'world'.) We need them in order to formulate testable hypotheses, but they are not themselves up for test (at least, not normally). (Philosophers: think of Wittgenstein's On Certainty run through Quine's epistemological mangle and you pretty much get the picture.) Now consider that delusions function as framework beliefs: the reason the patient so intransigently believes something this weird is that they have a deviant framework belief.

The problems with this kind of approach (the problems with the very idea of nonsensical framework propositions) are already well known (Thornton 2008). I'm not going to rehearse them here. Instead I just want to point out the obvious fact that whilst this 'top-down' approach aims to explain why the delusion is maintained, and how it is that certain affects and meanings find their way into the subject's experience, it does nothing to say how delusions come about. Campbell suggests an unspecified brain disorder, and wards off the criticism that appeals to brain disturbance are better at explaining disrupted experience than disrupted belief. But nothing in the top-down 'rationalist' theory itself leads to this neurological suggestion. (Point of interest: Campbell himself appears to ultimately endorse neither the bottom-up nor the top-down approach, instead applauding Gerrans' view that disturbances of belief and experience are effects of a common cause, rather than either being the cause of the other.)

Both Campbell and Rhodes & YoursTruly draw on Wittgenstein's On Certainty to explicate the essential disturbance of delusion. Whilst Campbell refers to the notion of framework propositions, Rhodes & I instead draw on the notion of a non-propositional dispositional bedrock of sensibilities. Disturbances in this subdoxastic background or bedrock throw the subject out of his automatic kinship with others, disrupting his going-on-being-in-the-world. We don't draw on the dubious idea of an unfathomable framework proposition (although note that Campbell only says that delusions are akin to framework beliefs), but instead promote the idea that delusions are compensations for disturbances in the bedrock which are so radically unchallengeable because the subject no longer enjoys a non-psychotic place to stand from which to mount any critique. There is surely something right, however, in what Campbell says: the fact is that the delusional patient tends to take their delusion as an axiom, and the system of their delusional thought is built around it. It itself is not, on the whole, up for grabs, but instead becomes the filter through which other thought and experience is processed. Yet whether we draw on Wittgenstein to explicate the absence (damaged background) or the presence (quasi-framework beliefs) of certainty, in neither case are we offering any kind of causal hypothesis. In this sense the top-down theories are not to be considered theories of how delusions are formed. And in a sense this ought to have been obvious from the get go: top-down theories tell us that the direction of explanation is to be from delusion to manqué delusion-confirming experience - and it is obvious that this presupposes that we already have to do with a delusion.

Throw in a dynamic hypothesis: that delusions are anxiety-reducing compensations for a loss of pre-reflective certainty and we have the beginnings of (what could be called) a causal explanation (so long as explanations in terms of motivation, character and function are allowed to be called 'causal' - alongside those ?perhaps more prototypical examples of 'efficient' precursor-referencing forms of causal explanation).  At the same time we may also reference the traditional idea that delusionality involves an influx of imagination into reality, a local collapsing of the domains of inner (dream/phantasy) and outer (e.g. perceptual) experience (a failure in 'reality testing' as the analysts say). As Sass describes it, the inner world becomes objectified; as the analysts describe it, the outer world becomes subjectified; as it seems to me, these are more complementary than competing notions. In this collapsing, 'symbolic equation' replaces symbolism (Hanna Segal). An anxiolytic purge of intolerable (self-splitting) inner conflict becomes solved for. The lure of paranoia - to locate the conflict outside, even at the expense of depleting the self (cf the major losses of parts of the self to itself in what is called 'massive projective identification') - grows strong.

The 'rationalist' version of 'top-down' theories of delusion, just like the empiricist version of the 'bottom up' approach, want to put the subject's rational sense-making at the heart of the mechanism of delusional thought. This ever-so-innocent and widespread, but to my mind utterly pernicious, idea of the human subject as located in the world fundamentally as a sense-maker takes us further away from, not closer to, an understanding of the delusionality of delusional belief.1 Sense-making takes place under the aegis of the 'reality principle'; delusion, I want to say, precisely does not.

How are we to understand these evasions of madness in psychological theories of delusion? My own hunch is that they constitute largely unconscious motivated deflections from the unbearableness of psychosis. The phantasy may be that if we could explain the delusional condition in terms, say, of a matter of an isolable pathology of belief, then we can reduce the phenomenon. And by reducing (assimilating) it, the hope is that one would not have to accommodate to it, or better - not be frustrated or pained or disturbed by it in our thwarted attempts at such accommodation. But reduction, here, is just what doesn't work; instead it amounts to changing the topic (changing it to that of dealing with various odd intransigent mistakes that people make). Far too many cognitively oriented theories simply carry almost none of the quality of the psychotic mind. And it is not as if we could say 'well the authors are philosophers rather than clinicians', for the dreadful sound of schizophrenic delusion and of the terror it hides is amply available for those with ears to hear in the hundreds of first person and other accounts that are widely available. Instead of the schizophrenic world, theorists offer us tidy accounts of the relation of beliefs to experiences, failures in hypothesis testing, inferential reasoning, and so on. The terror and tragedy of the experience of true madness - and, in case it's not obvious, the terror and tragedy I am thinking of here are precisely those that the patient is often not feeling when they are deluded; the terror being what the delusion displaces, and the tragedy being the ethical assault of all of this on their personhood - is just nowhere on the table.

I don't want to be taken  the wrong way; this is just a hunch of mine, and the value or lack of value of it will depend on what the cognitive theorist has to say by way of acknowledgement: does honest reflection incline you at all to self-ascribe this deflective disposition? Well: does it?



 Take Kelly's 'people are like scientists' cognitivist trope. On the one hand, the obvious retort is that, well, some of them are like scientists; some of them even are scientists. Then again one wants to ask, rhetorically, if the reverse is true: ... but are scientists like people? (one can only hope). There are of course various times in which I am puzzled in my goings about; my situation stops making sense to me; I stop seeing the sense in it that is there to be found. At these times I may pause and have to try to make sense of it. But an iota of phenomenological reflection tells us that this is hardly the normal case: I am not normally having to make sense of the situations I am in: they already make sense. 'Ah', says the cognitivist, 'I'm not saying that sense-making is a conscious activity; instead it takes place subconsciously, or perhaps even subpersonally'. Well, now what we need are criteria for the application of the term 'subconscious sense-making'. It had better not turn out (on pains of circularity) that such grounds for attributing subconscious sense-making are the exercising of those very capacities which that sense-making is posited to explain - my capacity to automatically see the sense in what I encounter, for example. Referring to brain functions at this point would be fine by way of causal explanation, so long as (again to avoid circularity) they are identified independently of the said functions - fine so long as we also acknowledge that we've now left epistemology way behind.