Cookies

We use cookies to ensure that we give you the best experience on our website. You can change your cookie settings at any time. Otherwise, we'll assume you're OK to continue.

Durham University

Department of Philosophy

Staff

Publication details for Professor Andy Hamilton

Hamilton, Andy (2006). Against the Belief Model of Delusion. In Reconceiving Schizophrenia. Chung, M., Fulford, W. & Graham, G. Oxford: Oxford University Press. 217-234.

Author(s) from Durham

Abstract

Psychotic and non-psychotic delusions
The central aim this article is to criticise the received opinion that delusions are beliefs. I will argue that in many psychotic and non-psychotic cases, the basic level of description of delusion falls short of the ascription of belief. In monothematic, behaviourally inert cases at least, I maintain that although the delusion shares some features of belief, the disanalogies are sufficient to justify withholding a clear belief-attribution. My thesis is not quite that in many cases delusions are not beliefs; rather, it is that there is no fact of the matter concerning whether S believes that p.
 In this opening section, however, I will also question the usual concentration - by philosophers as well as psychiatrists and psychologists - on psychotic delusion. Psychotic cases are the most fascinating, but in ignoring non-psychotic delusion, one loses a wider perspective which, I believe, tends to support the central thesis of this article, that many delusions are not beliefs. In a range of non-clinical cases we speak of people being deluded. These include conditions such as self-deception and wishful thinking, where in contrast to many psychotic delusions, there is nothing bizarre about the proposition that is allegedly believed. The self-deceiver is deluded when, against the evidence, she apparently gets herself to believe that her partner is not being unfaithful. The thirsty, exhausted desert traveller is deluded when wishful thinking makes him disregard the likelihood that the oasis is just an optical illusion; even someone unfamiliar with this kind of optical illusion, and therefore epistemically relatively blameless, could be described as deluded. When Lord Archer, whose confabulations have so enlivened British public life in recent decades, says that he gained a degree from an American university, or denies that he was ever arrested for shop-lifting, there is the feeling that he is not just lying for reasons of self-advantage, but is deluded—he is a fantasist.
 Non-psychotic delusions mostly fall under the heading of motivated irrationality. But in colloquial usage the scope of non-psychotic delusion may appear broader still. People are ready to describe as delusory many religious or political beliefs which they reject—for instance the Roman Catholic belief that the bread and wine in the Eucharist are converted into the body and blood of Christ—but this is a metaphorical use. For the description to be more than metaphorical, an ideological critique along the lines suggested by Freud or Marx would have to be endorsed. A delusion is not simply a groundless framework principle that one does not accept; to call something a delusion is to imply that it is more than a simple error. It is essential to distinguish delusions, especially psychotic cases, from mere mistaken belief. J.L. Austin remarked that the presence of delusion indicates that something is wrong with the subject, whereas the explanation for illusion lies in the world (Austin 1962, pp. 20–25). However, non-perceptual illusions are often assimilated with delusion; an illusion, except when perceptual, is always a positive belief, something that one wishes to believe, while delusion is neutral. Hence Freud in ‘The future of an illusion’ refers to religious beliefs as illusions rather than delusions: ‘not precipitates or end-results of thinking: they are illusions, fulfilments of the oldest, strongest and most urgent wishes of mankind’ (Freud 1953, p. 30). Ian Kershaw describes Hitler in 1944 as believing unshakeably that ‘the strength to hold out would eventually lead to a turning of the tide, and to Germany’s final victory…he expressed his unfounded optimism through references to the grace of Providence…the self-deception involved was colossal. Hitler lived increasingly in a world of illusion, clutching as the year wore on ever more desperately at whatever straws he could find’ (Kershaw 2000, pp. 609–610). Both of the latter cases could be described as delusion.
 In contrast, psychotic delusions are never described as illusions. They include Cothard’s and Capgras’ delusions, and schizophrenic delusions; under this heading would be included Descartes’s examples of subjects who believe that they are made of glass, that they are a pumpkin, and so on. Some psychotic delusions are caused by brain injury, or are otherwise organic, others are not.1 Capgras’ delusion seems to fit into the former category, since it occurs after right-hemisphere brain damage; while many drugs cause psychotic episodes involving delusions. The hallucinating subject is deluded if they believe that there really are pink rats, that is, where the hallucination is not ‘lucid’ in the manner of lucid dreaming. (Most cases of hallucination involve either psychosis, or drug-induced psychotic episodes; but hearing voices may, in some cases, be a non-psychotic hallucination.)
 The account presented here allows for a vague boundary between psychotic and non-psychotic cases. But the very distinction between psychotic and non-psychotic may be tendentious, since it implies a medical model of mental illness. Although the anti-psychiatry debate is not addressed directly here, the arguments presented will have implications for it. In particular, in criticizing the assumption that all delusions are beliefs, and offering a deflationary analysis of the phenomenon of delusion, the present account may undermine the standard medical model of psychosis as involving ‘poor reality-testing’. It should also be noted that the analysis offered here treats work in scientific psychology and psychiatry as useful data rather than theoretical support for philosophical conclusions.
 It might be argued that psychotic delusions, as the results of mental illness or brain damage, are the only genuine, objective cases of delusion. This is a tempting position, especially for those who regard delusion as loss of reality-testing and thus as defining madness or psychosis, since those who are deluded in the non-psychotic sense have lost this contact with reality only in a relatively mild way. The position seems to have wide support. For Lacan, ‘the clinical characteristic of the psychotic is distinguished by this profoundly perverted relation to reality known as delusion’ (Lacan 1993, p. 44); G. Lynn Stephens writes that ‘delusion does more than serve as a sign of madness: it is constitutive of madness, or at least of some forms of madness’ (Stephens 1999, p. 25).2
 Proponents of this view regard non-psychotic delusion as secondary. Psychiatrist Anthony David evidently assumes that all delusions are psychotic. He provides an interesting list of false beliefs found in anorexics, sufferers from Chronic Fatigue Syndrome, anosognosia, and other conditions, which he claims illustrate our inconsistency in applying the concept of delusion; claiming to be overweight and denying the fact that one is dangerously thin is regarded merely as an ‘overvalued idea’, and claiming that Elvis is still alive is regarded as daft, and not a delusion (David 1999, p. 18). But if one allows that there are non-psychotic delusions, then most of the cases that David cites will count as delusions in this broader sense. (The anosognosic patient who denies that they are paralysed on their left side is an interesting case, perhaps closer to ignorance than delusion.) It is hard to judge which position is correct, and here I leave the question open. The most important criterion of delusion is that in neither the psychotic nor the non-psychotic sense does it involve a simple error; the attribution of delusion implies that there is something fairly serious wrong with the subject.
 The status of non-psychotic delusion bears on the question of whether delusion is something that is not accepted by others in one’s culture or sub-culture, and the related question of the contrast between bizarre and more mundane delusions. DSM-IV contrasts mundane delusions—the delusion that one is being followed or that one’s spouse is having an affair—with bizarre delusions such as that one’s internal organs have been removed or that one’s partner has been replaced by a double. According to DSM-IV, bizarre delusions are generally impossible, non-bizarre delusions are merely improbable. (Presumably the impossibility could be either physical or conceptual.) Clearly there is a vague boundary between these categories. However, DSM-IV also claims that a delusion is a belief not ‘ordinarily accepted by other members of the person’s culture or subculture (e.g. it is not an article of religious faith)’ (APA 2000, p. 765).
 The apparent conflict between these criteria—if something is merely improbable why could it not be accepted by others who are sane?—is resolved when one recognizes that it is the perplexing relation to a ground as well as any bizarre propositional content which makes something a delusion. Thus the propositional content of a mundane delusion may very well be accepted by others in one’s culture or sub-culture, but on a rational basis. What DSM-IV should say, therefore, is that in psychotic cases, if the delusion is bizarre, it is not accepted by (normal) others at all; if mundane, it is not accepted in the way that the psychotic subject appears to accept it, that is without grounds. It may be felt that the bizarreness of a bizarre delusion has to be culture-relative; but the essential claim is that the delusion is bizarre relative to the prevalent beliefs of the subject’s own culture. It might be argued that Capgras’ delusion, where the patient says that their partner has been replaced by a double, is not intrinsically bizarre; recall the films The Return of Martin Guerre and Sommersby, where the long-lost husband turns out to be an imposter. The question is whether the subject says that their partner is an imposter or a double (a replica).3 Bizarre delusions such as ‘I am the Virgin Mary’ may be characterized as those which cannot properly be acted on, while mundane delusions, which might be acted on, are often curiously inert—or so I will argue.
 The DSM-IV claim that the delusory belief is not accepted in one’s own culture remains problematic, however. It is not just that the belief, if it is a belief, is not accepted by others—it is not even understood by them. Moreover, it is not clear that culture-wide delusion is impossible. Collective hallucinations have resulted in collective delusion, for instance when troops in battle report that angels appeared from the clouds and fought on their side. The idea of a collective delusion seems more plausible when a physical cause is implicated, for instance ergotamine poisoning from eating bread made from rye flour infected with ergot, which on one analysis contributed to causing the First Crusade; mass hysteria and other infectious delusions are also exogenous, if not within the sphere of reasons. Culture-wide delusions such as ‘The end of the world is nigh’ will not be bizarre in the sense of ‘I am the Virgin Mary’; and a culture cannot be collectively interpreted as ‘insane’.
 The diverse range of delusions implies differences in philosophical treatment. Non-psychotic delusions, I will argue, fall within the sphere of reasons, broadly construed, and only in these cases can one speak of a bias in reasoning—the subject’s beliefs are motivated by their desires or wishes. Psychotic delusions, in contrast, do not fall within the sphere of reasons, and so a reasoning bias explanation is not appropriate. However, this claim does not imply that the deluded subject does not really understand what they are saying. That would be one possible thesis, and I will discuss it shortly. The alternative thesis to be defended here is, to reiterate, that in many psychotic and non-psychotic cases, the basic level of description of delusion falls short of the ascription of belief. Such cases include self-deception and monothematic delusion, where apparently only a single belief is affected, and the delusion tends to be behaviourally inert; in contrast, wishful thinking and polythematic delusion, where patients tend to be delusional about anything that attracts their attention, are closer to belief-status. I argue that at least in monothematic, behaviourally inert cases, although the delusion shares some features of belief, the disanalogies are sufficient to justify withholding a clear attribution.
 I have argued elsewhere, in an analysis of self-deception, that it is not clear-cut that the self-deceiver believes both that p and that not-p, where p is for instance ‘My partner is being unfaithful’; rather, there is evidence that they believe that p, and evidence that they believe that not-p. In such cases one should take as basic the conflict of evidence concerning whether X believes that p or not-p, and not assume a further, possibly evidence-transcending fact of the matter concerning their belief (Hamilton 2000). Here I argue that in the case of monothematic psychotic delusion also, there can be no clear ascription of belief. Rather, what one should take as basic is that there is evidence that X believes that p and evidence that X does not believe that p, in contrast to self-deception where there is evidence that X believes that p and evidence that X believes that not-p. The evidence for the ascription of a (delusory) belief is that the subject, apparently without intending to deceive, repeatedly asserts p; the evidence against such an ascription is that they act on the putative belief in rather circumscribed ways at best, and seem not to hold it on the basis of reasons. In self-deception, in contrast, there is a question in a different way about whether the subject believes that p, since here, the opposed evidence indicates not that they have no belief on the matter, but that they have the contrary belief. The situation is quite different in the case of wishful thinking, where the subject does indeed believe that p; likewise, arguably, in the case of polythematic delusion.
 Thus, to reiterate, my thesis is not quite that in many cases delusions are not beliefs; rather, it is that there is no fact of the matter concerning whether S believes that p. This position is subtly different from that of a number of recent writers who have argued that delusions should not be regarded as beliefs.4 It also offers a less radical alternative to the thesis that the deluded subject has lost understanding of the terms they are using. This thesis, defended recently by John Campbell, is one that I will be concerned to undermine; but first some scene-setting of the contemporary terms of debate is required.
Delusions, reasons and causes
Henceforth, while noting the existence of non-psychotic cases, I will focus on psychotic delusion; ‘delusion’ will generally function as a contraction of ‘psychotic delusion’. Accounts of psychotic delusion have tended to follow two opposed tendencies—rationalist and irrationalist. The irrationalist tendency is exemplified by Jaspers’ pioneering treatment in General psychopathology (1963, first published 1923), which argued that schizophrenic delusions are not comprehensible—that is, they do not have a rational explanation in terms of common-sense psychology, only a causal explanation in terms of brain disease. In the rationalist camp, in contrast, belongs the extensive recent psychological research that regards deluded subjects as having a general tendency to faulty inference and biased reasoning. An example is the work of Davies and Coltheart, although their commitment to a reasoning bias explanation is tempered by adducing a tendency towards self-serving bias of the kind often regarded as a mechanism for maintaining self-esteem. They allow that such biases will not by themselves explain delusional beliefs, and claim that some delusional beliefs may be arrived at by taking an illusory experience to be veridical (Davies and Coltheart 2000, p. 13). But the general tendency of rationalist views is to assimilate delusions to mistaken beliefs based on faulty reasoning; and conversely, perhaps, to exaggerate the extent of irrationality in everyday thinking.
 Both rationalist and irrationalist accounts should be rejected, I believe. Psychotic delusions are not beliefs explained by reasoning biases within the scope of common-sense psychology, or by rational reactions to bizarre experience; but nor should one concede that they involve mental attitudes subject only to causal explanation. My main target is the rationalist account, however, and here I simply emphasize that in rejecting it, one is not committed to the position of Jaspers and the irrationalists. (Here I make no particular comments about schizophrenic delusion, thought-insertion and control.) The latter are mistaken in dismissing the possibility that delusions express the individual’s psychology or personality—though it should be conceded that the more the delusion does express this, the less it will be regarded as psychotic. There is an important distinction between explaining why the subject comes to have delusions, and why they come to have the kind of delusions that they do. The latter explanation might be individual, in terms of depth psychology, but it will have a social and cultural dimension also—for instance only in industrialized societies, which have television and the internet, could subjects believe that their thoughts are controlled through these media. A depth-psychological explanation would however be incompatible with causal explanation in terms of brain disease.
 The problem with rationalism is that it makes psychotic delusions appear too comprehensible. Some cases cited by rationalists seem too ‘rational’ to be psychotic delusions at all, while genuine psychotic delusions are interpreted as a kind of error in thought. In the former category is Sedler’s example of a woman who ‘in the wake of an automobile accident in which [her] husband and child are killed…develops the conviction that she caused their deaths; later, she comes to believe that she is evil and must be destroyed, and, therefore, attempts suicide; finally, she is improved after a course of electroconvulsive therapy’. Sedler comments that at the level of manifest content the delusion seems neither random nor empty of significance. There may be a pre-existing dynamic—unconscious guilt, a forgotten wish to kill her father, or a more proximate wish to divorce her husband—or the psychosis may create a delusion that incorporates in distorted form the affects naturally arising out of the trauma; individual cases may differ, Sedler concludes (Sedler 1995, p. 258).
 I would comment that as the case is presented, the woman’s belief that she is evil is a comprehensible result of a sequence of events, and seems to lie close to the sphere of reasons. More detail is required—clearly it makes a difference whether the woman was driving the car and was therefore causally responsible for her husband’s and child’s deaths. If the woman believes that she is responsible morally and not just causally, this seems like a non-psychotic delusion; while even the delusion that she was evil might not count as psychotic. There are parallels with a real-life case reported some years ago in the British press, in which a woman ran over and killed a child who had suddenly run into the road. The woman had a strong Christian belief, and developed overwhelming feelings of guilt, including guilt that her own child, who was not involved in the accident, was still alive—that another parent and not herself had been bereaved. She came to believe that she must be evil and was suffering divine punishment. Such feelings and beliefs would not be incomprehensible even within the religiously moderate Anglican subculture. The women in both examples might benefit from psychotherapy, and it is Sedler’s tendentious introduction of ECT treatment that helps to skew his example towards the psychotic.
 The preceding cases are, I have argued, too comprehensible to count as psychotic delusions. The converse error made by rationalists is to attempt to explain genuinely psychotic delusions in a way that renders them too readily comprehensible. Proponents of reasoning bias fail to recognize that the ‘conclusions’ characteristic of bizarre psychotic delusion are so wayward that one cannot speak of a process of reasoning at all. As Wittgenstein would have put it, the alleged error of the psychotically deluded is ‘too big for a mistake’. In On certainty, he comments on the distinction between mistakes and mental disturbances (Wittgenstein 1969, paras 71, 74):
If my friend were to imagine one day that he had been living for a long time past in such and such a place, etc., etc., I would not call this a mistake, but rather a mental disturbance, perhaps a transient one. …Can’t we say: a mistake doesn’t only have a cause, it also has a ground? I.e. roughly: when someone makes a mistake, this can be fitted into what he knows aright.
 I take it that Wittgenstein means that where the subject is mistaken, one can for instance describe their faulty steps of reasoning, or their inattentive observation, which takes them from ‘what they know aright’ to an erroneous conclusion; the mistakes are, at least to some extent, understandable ones to make, if they are pointed out with enough patience and clarity, the subject will acknowledge them, and so on. This would not be possible in the case of psychotically deluded subjects. Delusion may be described as a failure of reason, but to say that someone has lost their reason is not to say that they reason badly; indeed, psychotic patients seem to fare no different to others in standard tests of reasoning.
 Wittgenstein’s remarks are cited in the recent discussion by John Campbell (2001a) and Naomi Eilan (2001). Campbell classifies recent psychological accounts of delusion under the headings of ‘empiricist’ and ‘rationalist’, although in the terms that I have been using, both accounts are rationalist in assimilating delusions with ordinary error. Empiricists, for Campbell, are those who regard the patient as making a broadly rational response to some very unusual (presumably hallucinatory) experiences.5 According to empiricist accounts of Cothard’s or Capgras’s delusion, then, it is an abnormal lack of perceptual affect that causes the patient to think that they are dead or that their partner has been replaced by a double. On rationalist accounts, in contrast, there is a ‘top-down disturbance in the subject’s beliefs’, which explains the change in affect. On this view, the Capgras’ delusion ‘My partner has been replaced by a double’ becomes a local ‘framework proposition’ in the sense of Wittgenstein’s On certainty—a proposition that is groundless and foundational to a practice.
 Most important for present purposes is Campbell’s characterization of the basic philosophical problem raised by delusion: In general we have to ascribe meaning to utterances in a way that makes the subject rational, yet in the case of the psychotic patient, we seem unable to formulate the content of their delusion; it seems that they do not retain a stable grasp of meaning through the mental disturbance (Campbell 2002, p. 91). I will call this ‘Campbell’s Problem’, and clearly it does not apply to non-psychotic delusions. Eilan believes that Campbell’s Problem yields a criterion for distinguishing (psychotic) delusions from mere mistakes, viz., ‘We are in the realm of delusion rather than mistake when the failure of reason is such as to put into doubt the subject’s understanding of the terms used to express his purported beliefs’ (Eilan 2001, p. 123). (One should note that this criterion would make non-psychotic delusions into mere mistakes, and so for that reason alone would require qualification.)
 Campbell’s analysis is that the Capgras’ patient no longer grasps what he terms the memory demonstrative ‘that [remembered] woman’.6 Thus when the patient says ‘That [currently perceived] woman is not my wife’, the underlying delusion is ‘That [currently perceived] woman is not that [remembered] woman’, while the patient also has the presumably correct belief that ‘That [remembered] woman is my wife’. And yet, Campbell continues, the patient does not in any way try to verify the negated identity statement—notably by checking whether the woman he currently perceives shares memories of events in which he and his wife took part. Campbell asks how the patient can be said to retain a grasp of the meaning of their remarks, in particular the memory-demonstrative, when they use words in such a deviant way? He has a similar view of the schizophrenic patient who looks at a row of empty marble tables in a cafe and apparently becomes convinced that the world is coming to an end, commenting that it is problematic how any experience at all, let’alone one of marble tables, could be relevant to verifying the proposition ‘The world is ending’. Campbell concludes that there must be ‘top-down loading’ of the experience by the patient; the consequence is that in expressing their delusion, the patient has lost an understanding of the words they use.
 My position, in contrast, is that the claim of loss of understanding is too radical, and I will argue instead that there should be no clear attribution of belief to the subject. However, features that may be cited in support of Campbell’s thesis—principally that the belief is groundless and behaviourally inert—also support the view that delusions are not genuine beliefs. (Indeed behavioural inertia is cited by Campbell.) I will present these features and then show why one should not conclude the loss of understanding thesis from them; finally I will try to defuse some objections to a non-belief model of delusion.
Non-belief-like features of delusion
(a) Groundlessness
It is a notable feature of psychotic delusion that either the patient offers no grounds for their delusion, or that if they do, they seem not to take them seriously. More mundane delusions, which share their propositional content with ordinary empirical beliefs based on grounds, are distinguished by being ungrounded. The psychotic patient seems convinced that they are the Virgin Mary or whatever, but the justifications that they offer carry the flavour of confabulation; they seem to be thought up after the question, and appear arbitrary, and psychiatrists comment that ‘They don’t mean anything to the patient’.7 The reasons that the subject gives when questioned seem not to be operative. The self-deceiver recognizes that evidence is applicable, but has a curiously selective way of interpreting it; the psychotic patient, in contrast, seems not to recognize that evidence is applicable at all. This appearance contributes to what Jaspers called the ‘axiom of the abyss’, a therapist’s feeling of encountering in the patient an absolutely enigmatic way of life; for him this was the central criterion for diagnosing schizophrenia.
 It might be said against these arguments that at least in making past-tense claims about events apparently witnessed or experienced, the psychotic patient does not have to offer justifications, but simply has to report them in the way that people normally do. Memory is direct knowledge of the past, and so the subject need offer no justification beyond the claim ‘I remember…’. Thus, the objection continues, the schizophrenic patient who recalls having an abortion in Buckingham Palace, says what anyone would say if they remembered such an event clearly. But though I would concur that memory is direct knowledge, and that memory reports do not require the justifications postulated by inferential accounts of personal memory, it is not obvious that the psychotic patient does say what others do when making personal memory-reports. Memory-judgements have a distinctive expression involving the continuous-verb form, which implies the possibility of a spontaneous manifestation or willed rehearsal of the remembered events in the form of memory images or memory-experience.8 For the objection to be convincing, therefore, the schizophrenic patient would have to say, for instance, ‘I distinctly remember being shown into the reception area, and taken by medical staff to the West Wing…’. It is doubtful that such patients would present their putative memory-delusions using the continuous verb form in this way. More generally, it is sometimes argued by proponents of reasoning bias explanation that many people believe things on little or no evidence. Clearly this is an issue that requires lengthy consideration, but I would suggest that such people are at least prepared to accept the appropriateness of reasons; the behaviour of psychotic patients implies a more radical estrangement.
(b) Behavioural inertia
Many delusions exhibit relative behavioural inertia, and are not fully acted on. As Bleuler commented (1950, p. 129), although many schizophrenic subjects have delusions that they are great leaders, ‘None of our generals has ever attempted to act in accordance with his imaginary rank and station’. Similarly, sufferers from Capgras’ delusion, which is relatively monothematic, often fail to express curiosity or form beliefs about where their spouse has gone, whether they are alive or dead, and so on. The delusion is avowed, and anger or irritation may be expressed towards the double; but often no attempt is made to get rid of them, or to search for the genuine spouse, and the subject may even be actively friendly. It is true that in some cases the delusion is acted on with tragic consequences, and here it is closer to a belief. But as Young notes (2000, p. 53), even when acted on ‘there are often inconsistencies in accompanying affect and a curiously circumscribed quality to the delusion itself’. Sass comments on schizophrenic patients’ tendency to ‘double book-keeping’; their delusory utterances contrast with other statements, which suggest an accurate grasp of their situation. For instance, the patient who denies that she has children will give their names when asked (Sass 1994, p. 21). Jaspers contrasts the ‘specific schizophrenic incorrigibility’ with the normal dogmatism of fanatics or of manic-depressives. The schizophrenic’s delusion is quite unshakeable, he explains; yet in contrast to dogmatics, their attitude to the delusion is ‘peculiarly inconsequent at times’, and such delusions often do not lead to action (Jaspers 1963, p. 4). It is definitive of schizophrenia that the delusions are often not accompanied by an emotional state appropriate to their content; for instance, a schizophrenic patient may report that others are trying to kill them, while apparently remaining completely indifferent to this prospect.
 The distinction between mundane and bizarre delusions is relevant to the question of behavioural manifestation. There are limits to acting on bizarre delusions, which arise from their very unfeasibility; indeed ‘The belief is not feasible’ may just amount to ‘It cannot be acted on’. It may be that nothing could count as acting from a bizarre delusion, since the attempt to do so would conflict with other everyday beliefs and behaviour. How could the belief that I am a tree or a pumpkin be consistent with walking downstairs, for instance? The English patient who claimed to be working for President Bush sen.—a relatively mundane delusion in the present context—did to some extent act on his belief; his flat contained material on US government and politics, and so on. A patient with the bizarre delusion that he was the Virgin Mary acted in accordance with it perhaps as far as one could—he dressed with a veil, spoke softly, and so on. But how could he express concern about Joseph, for instance, unless the delusion was broadened so that other individuals assumed Biblical identities? (This is to assume that the patient has the delusion that he really is the Virgin; perhaps if pressed he may admit that he is merely acting like her.)9
(c) The ‘web of belief’
The final reason for questioning the comprehension or belief-status of delusions is that attribution of belief is governed by a constraint of rationality or reasonableness. Davies and Coltheart concede that if one cannot make any sense of how someone could reasonably have arrived at a particular belief on the basis of experience and inference, then this counts at least provisionally against the attribution of that belief to them (Davies and Coltheart 2000, p. 2). They comment that we expect to find an intelligible link between belief and experience, and that the subject’s beliefs fit together tolerably well. Yet as Young (2000) notes, monothematic delusions are not readily compatible with holistic theories, which emphasize the importance of an integrated ‘web of belief’. This fact is, I would argue, a further reason for denying belief-status to such delusions.

Against the "loss of understanding" thesis
To reiterate, these data—groundlessness, behavioural inertia, and failure to fit the web of belief—yield support both for a loss of understanding thesis and for a non-belief account. So it is necessary to show why the latter approach should be preferred. As noted earlier, Eilan claims that error becomes delusion ‘when the failure of reason is such as to put into doubt the subject’s understanding of the terms used to express his purported beliefs’. Her reference to a ‘failure of reason’ shows that she recognizes that it is not just any doubt about someone’s understanding that implies the presence of delusion; they may attempt to express a belief using words which it is clear they do not understand, without this counting as a delusion. Eilan develops her position by suggesting that we are in the realm of delusion when we encounter doubt or denial of a framework proposition; she argues that this can never count as a mere mistake because such propositions are not accepted on the basis of reasons or evidence in the first place. Eilan glosses what Campbell terms ‘top-down loading’ of the experience as a change of framework belief rather than loss of understanding.10
It is evident that Eilan and Campbell at different times propose two distinct analyses:
(1) that the subject has no understanding of the words they use; and
(2) that they have a deviant understanding (‘deviant’ in the sociologist’s non-evaluative, statistical use of the term).
Neither the loss of understanding nor the deviant understanding thesis is plausible, I believe, and each will be criticized in turn. The claim of loss of understanding is not tenable because the loss is not a general one; a local loss of understanding is implausible, and indeed may make no sense. The Capgras’ patient has not lost a general understanding of the memory demonstrative or any other term, since in other contexts they grasp what ‘my wife’ and ‘that woman’ refer to, and what ‘replaced by a double’ means. Similarly, a schizophrenic patient has been using the words ‘the world’ and ‘ends’ correctly most of their life. In the latter case a loss of understanding is even less plausible since it would have to be intermittent, occurring only during florid phases of the illness. Conceptual role semantics, which argues that an understanding of a sentence consists principally in a grasp of its role in inference, provides no support for Campbell’s position. The schizophrenic patient may well be able to infer from, for instance, ‘The marble tables tell me that the end of the world is coming’ to ‘Something tells me that the end of the world is coming’. Many bizarre delusions—perhaps by definition all—seem to exemplify what Ryle (1949) termed a category-mistake. ‘My internal organs have been removed’, ‘I am made of glass’, and ‘I am a pumpkin’ look like category-mistakes; ‘My wife has been replaced by an imposter’ or ‘I am working for President Bush’, in contrast, do not. Now the person who regards the University as a mysterious entity separate from its component faculties, or the mind as a mysterious entity separate from the body—to take Ryle’s own examples—is meant to have a pervasive misunderstanding of ‘university’ or ‘mind’ (Ryle 1949, pp. 17–25 and passim). On Campbell’s thesis, in contrast, the psychotic subject’s misunderstanding is purely local; they understand ‘that woman’ in most contexts except those featuring in the delusion itself.
 The idea of a local category-mistake, and—it would follow—Campbell’s thesis itself, may in fact be encouraged by a too-literal interpretation of Wittgenstein’s remarks in On certainty concerning G.E. Moore’s attempt to refute the sceptic. Eilan treats these remarks as supporting Campbell’s thesis. Moore based his refutation on apparently commonsensical claims such as ‘I know that I have a hand’, but Wittgenstein wishes to undercut the debate between the ‘common sense’ philosopher and the sceptic: ‘“I don’t know if this is a hand”. But do you know what the word “hand” means?’ (Wittgenstein 1969, para 306). That is, he suggests, if someone attempts to debate whether this is a hand, whether they are a sceptic or an opponent who believes that they are defending common sense, they put into question whether they really understand the word ‘hand’. Note that this is a sceptical doubt concerning a framework proposition, a category into which many bizarre psychotic delusions fall, for instance the delusion that the inside of my skull is empty. G.E. Moore assumes that in entertaining the sceptical doubt he is saying something intelligible; Wittgenstein’s response, ‘But do you know what the word "hand" means?’, questions this assumption.
 However—and this is where Eilan goes wrong, I think—Wittgenstein is not claiming that it really is likely that a reasonably sane person such as Moore, in the course of philosophical discussion, has suffered a local loss of understanding of the word ‘hand’. ‘You have not given your words a clear meaning’ does not imply ‘You do not know what the word "hand" means in this context’; the most that one could say is that it is as if the philosopher does not understand what ‘hand’ means. Hence, although a temporary loss of understanding may perhaps be possible, there is no such thing as a ‘local’ loss of understanding. One cannot make a local category-mistake. Indeed Wittgenstein would be more likely to favour the Humean view that the sceptic does not really doubt that the best way to leave the room is by the door, rather than claiming that they do not understand what they are saying. This Humean view has obvious affinities with the position which I am defending, that there should be no clear ascription of belief in the case of many delusions.
 The alternative thesis of a deviant understanding is even less plausible than that of a loss of understanding. It implies what may be termed the lost tribe Romantic or anthropological view concerning mental illness and psychotic delusion.11 This position is exemplified by writers such as Foucault and R.D. Laing, and constitutes a branch of the anti-psychiatry movement that contrasts with Thomas Szasz’s libertarian wing. Laing and Foucault think of psychotic delusion as expressing a genuine, deviant vision; the deviant subject does not speak gibberish, but uses a linguistic code which non-deviants have not grasped, and has an alternative rationality. For proponents of the anthropological view, there is a continuity between psychotic and non-psychotic delusion; the only real difference between allegedly psychotic delusions concerning the end of the world and the widespread belief in some parts of the United States that 9/11 was a portent of Armageddon is that the latter is more prevalent. Indeed Romantics must say that there is no such thing as psychotic delusion, for by his or her own lights the psychotic subject is rational; so it looks as if the deviant understanding thesis could not be a response to Campbell’s original problem. Certainly Romantics are as mistaken as rational psychologists in assimilating psychotic delusion with the propositional attitudes of common-sense psychology. But a proper assessment of the anti-psychiatric position is beyond the remit of this article.
In fact, Eilan’s suggestion of a deviant understanding seems to conflate two claims:
(1) the Capgras patient has deviant beliefs, so that ‘This woman is a double of my wife’ becomes a framework principle; and
(2) they have a deviant understanding, such that ‘This woman is a double’ means in non-deviant terms, say, ‘God is evil’.
The second claim does not imply that the patient speaks Capgras-ese rather than English, but that they have a different understanding of certain concepts, just as individuals may have different understandings of ‘hoi polloi’ or ‘the moral sciences’. However, this deviance in understanding would have to be local, and locally deviant understanding is no more plausible than local loss of understanding. (In fact the well-defined brain pathology of Capgras’ syndrome means that such patients are not ideal material for the anti-psychiatry case.) There is the further objection that delusions cannot be regarded as expressing alternative framework principles in Wittgenstein’s sense, as Eilan proposes, since not all of them are candidates for framework status. Even those that are candidates exhibit groundlessness in a different way to ordinary framework principles. Such principles—which are better described as framework assumptions—constitute a basis for a wide range of the subject’s beliefs; they are not behaviourally inert, though their behavioural manifestation requires careful presentation. Many delusions, in contrast, seem to ground a very restricted range of attitudes, if indeed they ground these at all. Campbell’s thesis of loss or change of understanding is therefore implausible.
Delusions distinguished from beliefs
In an earlier defence of the authority of avowals of belief, I questioned the alleged holism of belief and desire in the explanation of action (Hamilton 2000). It is a by-product of the present article that another Davidsonian holism, that of belief and understanding, is also questioned. In this case, however, the authority of avowals—often called first-person authority—is not implicated; self-ascriptions of understanding are not authoritative, for clearly the subject may be mistaken in thinking that they understand a word correctly. Campbell’s position, in contrast, may be consistent with the holism of belief and understanding—since if the subject does not understand what they are saying, presumably there is no belief which can be ascribed to them either.12
 In support of the holistic claim is the fact that it may be difficult in some cases to decide whether the subject has lost understanding of the words they use, or whether no belief should be attributed. The general problem of interpretation involves an unclarity about what consequences to draw concerning how the subject will behave, what claims they will infer, and so on. But if my position is correct—that in bizarre cases a subject may be said to understand what they are saying without it being clear that they believe it—belief and understanding may be attributed separately, and so there is no holism.
 The position I wish to defend, which questions whether the psychotic patient really believes what they say, is less radical than Campbell’s. Furthermore, to reiterate, my argument is not that psychotic delusions are never beliefs. Monothematic, behaviourally inert delusions should be denied belief-status, while polythematic, behaviourally active delusions are more belief-like. Where nothing could count as acting from a bizarre delusion, it is correspondingly less plausible to classify such delusions as beliefs. (Insofar as the grounds for a loss of understanding thesis coincide with those for rejecting the belief-status of delusions, perhaps Campbell would allow that polythematic, behaviourally active delusions involve no loss of understanding.) As noted earlier, the distinction between belief-like and non-belief-like delusions is present in non-psychotic cases also; in the case of self-deception, for instance, there is no clear belief. A further, essential feature of the position which I am defending may also be overlooked. It is the following. My conclusion is not quite that the patient does not believe that their partner has been replaced by a double. Rather, it is that there is evidence that the subject believes that p, and evidence that they do not believe that p, and no way of resolving the conflict. In everyday cases, it is possible that even if X believes something firmly without it being any kind of delusion, there still could be evidence against the fact that they believe it. But we usually think that the question concerning what X believes can be settled, and that any undecidability reflects X‘s own indecision. In the case of many delusions, in contrast, the question cannot be resolved. ‘What do they believe, then?’ is the wrong question. There is a proposition concerning which there is evidence that the subject believes it, and evidence that they do not, and that is the best that can be said—a puzzling phenomenon.
 Thus if one asks ‘What is the problem with this person?’, three kinds of answer may be given:
1. He believes that his wife has been replaced by a double.
2. He claims, without intending to deceive, that his wife has been replaced by a double, but he does not believe it because he does not act in ways consistent with the belief, does not base it on grounds, and so on.
3. He claims, without intending to deceive, that his wife has been replaced by a double, but we cannot say that he believes it because he does not act in ways consistent with the belief, does not base it on grounds, and so on. There is no fact of the matter about whether he believes it or not.
I am proposing the third answer.
It may be felt that the basic level of description in such cases cannot be evidential, since the very use of the term ‘evidence’ points to a more basic level—that which the evidence is evidence for. Clearly this is normally so. But in the case of the puzzling phenomena of self-deception and psychotic delusion, the two sets of criteria for the ascription of belief, involving verbal and non-verbal behaviour respectively, do not cohere. These criteria also diverge when philosophers profess Pyrrhonian scepticism, or claim that they are essentially disembodied, yet continue to act purposefully. In such cases I do not argue that the question is indeterminate, but instead follow Hume in claiming that the subject does not really believe what they say; however, in these cases there is the possibility of some kind of insincerity or bad faith.
 In discussing the behavioural inertia of many delusions, I cited Louis Sass, who offers one of the most developed accounts to distinguish delusions from genuine beliefs. Sass is concerned to reject ‘poor reality-testing’ as the fundamental sign of madness and the basic principle of psychiatry, and in this he agrees with lost tribe Romantics such as R.D. Laing. He argues, for instance, that many schizophrenic patients experience their delusions or hallucinations as having a special quality, which sets these apart from their ‘real’ beliefs, or from reality as experienced by a ‘normal’ person (Sass 1994, p. 3). Although the account that I am defending does not imply lost-tribe Romanticism, I agree with Sass in questioning the medical model of ‘poor reality-testing’; as a result the phenomenon of delusion is deflated. Some delusions may retain belief-status, but nonetheless the DSM-IV definition of delusion—‘A false belief based on incorrect inference about external reality that is firmly sustained despite what almost everyone else believes and despite what constitutes incontrovertible and obvious proof or evidence to the contrary…’ —is mistaken in most aspects, not least in accepting without question the belief-status of delusion (APA 2000, p. 765).13 That assumption should certainly be questioned.
Endnotes
1. Young (2000, pp. 50–51) presents some examples of delusion arising from brain injury.
2. Anthony David comments that ‘delusion is the hallmark of psychosis’ (David 1999, p.17).
3. ‘Imposter’ is often used in the literature on Capgras, but it implies hostility by the subject, which fits only a range of cases; hence the more neutral ‘double’ is preferable as a general description.
4. For instance Currie (2000), Graham and Stephens in this volume.
5. For instance Ellis and Young (1990); a similar approach is endorsed by Radden (1985), who writes that the delusion of the psychotic patient may be ‘based upon idiosyncratic…"evidence" provided by hallucinations’ or ‘a private hallucinated world’ (pp. 68–69).
6. The ‘memory demonstrative’ is discussed more fully in Campbell (2001b).
7. Comment to the author by Anthony David.
8. A claim defended in Hamilton (2003) and Hamilton (forthcoming).
9. These cases are discussed in Chung (1992), pp. 395–429, 362–392.
10. Campbell (2002), p. 95.
11. I take the term from Squires (unpublished).
12. He refers to ‘perfectly sincere assertions made by people who seem to understand what they are saying, who may indeed act on the basis of what they are saying’ (p. 91).
13. In writing this article I am indebted to comments from Matthew Broome, Man Chung, Tony David, George Graham, Lucy O’Brien and Roger Squires, and from an audience at the Australasian Association of Philosophy Conference in Adelaide, 2003. I am grateful for support from the British Academy which enabled me to attend this conference.
References
American Psychiatric Association (APA) (2000) Diagnostic and statistical manual of mental disorders (DSM-IV-TR), (4th edn, revised). American Psychiatric Publishing Inc.
Austin, J.L. (1962). Sense and sensibilia. Oxford: Oxford University Press.
Bleuler, E. (1950). Dementia praecox or the group of schizophrenias (trans. J.Zinkin). New York: International Universities Press.
Campbell, J. (2001a). Rationality, meaning and the analysis of delusion. Philosophy, Psychiatry, and Psychology, 8(2/3): 89–100.
Campbell, J. (2001b). Memory demonstratives. In: Time and memory: issues in philosophy and psychology (ed. C. Hoerl). Oxford: Oxford University Press.
Chung, M. (1992). Social non-complicity of schizophrenia. Unpublished PhD thesis. University of Sheffield.
Currie, G. (2000). Imagination, delusion and hallucination. Mind and Language. 15(1): 168–183.
David, A. (1999). On the impossibility of defining delusions. Philosophy, Psychiatry and Psychology. 6(1): 17–20.
Davies, M. and Coltheart, M. (2000). Pathologies of belief: introduction. Mind and Language. 15(1): 1–46.
Eilan, N. (2001). Meaning, truth and the self: a commentary on Campbell and Parnas and Sass. Philosophy, Psychiatry, & Psychology, 8(2/3): 121–132.
Ellis, A.W. and Young, A.W. (1990). Accounting for delusional misidentifications. British Journal of Psychiatry, 157: 239–248.
Freud, S. (1953). The standard edition of the complete psychological works of Sigmund Freud, vol.21. London: Hogarth Press.
Graham, G. and Stephens, L. (ed.) (1994). Philosophical psychopathology. Cambridge MA: MIT Press.
Hamilton, A. (2000). The authority of avowals and the concept of belief. European Journal of Philosophy, 7(2): 20–39.
Hamilton, A. (2003). ‘Scottish commonsense’ about memory: a defence of Thomas Reid’s direct knowledge account. Australasian Journal of Philosophy, (In press).
Hamilton, A. (forthcoming). Memory and the body: a study of self-consciousness.
Jaspers, K. (1963). General psychopathology (trans. J.Hoenig etal.). Manchester: Manchester University Press.
Kershaw, I. (2000). Hitler 1936–45: nemesis. London: Penguin.
Lacan, J. (1993). The psychoses: the seminar of Jacques Lacan, Book III 1955–1956 (trans. R. Grigg). London: Routledge.
Radden, J. (1985). Madness and reason. London: George Allen & Unwin.
Ryle, G. (1949). The Concept of mind. London: Hutchinson.
Sass, L. (1994). The paradoxes of delusion. Ithaca: Cornell University Press.
Sass, L. and Parnas J.(2001). Self, solipsism and schizophrenic delusions. Philosophy, Psychiatry and Psychology, 8(2/3): 101–20.
Sedler, M. (1995). Understanding delusions. The Psychiatric Clinics of North America, 18(2): 251–262.
Squires, R. (unpublished). Mental disorder.
Stephens, G. (1999). Defining delusion. Philosophy, Psychiatry and Psychology, 6(1): 17–20.
Wittgenstein, L. (1969). On certainty. Oxford: Basil Blackwell.
Young, A. (2000). Wondrous strange: the neuropsychology of abnormal beliefs. Mind and Language, 15(1): 47–73.

Notes

Series: International perspectives in philosophy.