Home | Suchen | Kontakt | Impressum

Sie befinden sich hier: Aktuelle BeiträgePatrice Soom and Gottfried Vosgerau


A Functionalist Approach to the Concept of ‘Delusion’

Patrice Soom and Gottfried Vosgerau

[Journal für Philosophie & Psychiatrie, März 2015, Original paper]



Despite the considerable advances in neuropsychological research, still no adequate theoretical definition of the concept of delusion has been provided. This paper aims to offer a functional definition of delusions, which is useful for a systematic classification of sub-types of delusions. 


Based on a critical evaluation of the DSM-5 definition of delusions, a definition of delusions in terms of their asymmetrical inferential profile is proposed. This account is in line with the neuropsychological Two-Factor Account of delusions and it resolves the debate between doxastic and non-doxastic accounts of delusions. Moreover, it is shown how the specific etiology of different types of delusions contributes to the individuation of psychological functional sub-types of delusions, which constitute an empirically informed classification of delusions. 


Keywords: neuropsychology, delusions, belief revision, two-factor account, psychiatric classification

The Heterogeneity of the Clinical Phenomenon of Delusions 

Delusions are the hallmarks of psychoses. Their fascinating character stems from the fact that, at least prima facie, patients endorse extraordinary beliefs in the absence of any apparent perceptual malfunctioning, suggesting that their rationality is deeply affected. In spite of their general ‘strangeness’, delusions notably are heterogeneous, both with regard to their content and to the psychiatric conditions in which they occur. This renders the task of providing a theoretically adequate definition difficult. 


With regard to their content, delusions come in several variants. In a famous case report, Passer and Warnock described ‘Mrs D., a 74-year-old married housewife [… who] had received the diagnosis of atypical psychosis because of her belief that her husband had been replaced by another unrelated man […]. She easily recognized other family members and would misidentify her husband only’ (1991). Mrs D's case is a prototypical example of Capgras delusion, in which the subject believes that a close relative has been replaced by an impostor (Young, 2000). Delusions concerning relatives can also be directed in the reverse direction: In the Fregoli delusion, patients tend to think that their social environment is populated by people they do know, but who nevertheless remain unrecognizable, since they are in disguise (Ellis, Whitley & Luauté, 1994). Misidentification delusions may also concern the ability to recognize oneself in mirrors (Breen, Caine, Coltheart, Hendy & Roberts, 2000). 


The range of possible delusional content also includes cases in which people claim that they are not in control of their own actions (delusion of control; Blakemore, Oakley & Frith, 2003) or even that their thoughts are not their own (delusion of thought insertion; Graham & Stephens, 1994). Even more strangely, some individuals claim that they are actually dead, a phenomenon which is described as the Cotard delusion (Gardner-Thorpe & Pearn, 2004). These are striking examples of abnormal beliefs. More common delusional themes concern persecution (McKay, Langdon & Coltheart, 2005), grandiosity (Knowles, McCarthy-Jones & Rowse, 2011), jealousy (Soyka, Naber & Völker, 1991) or even erotomania (Berrios, 2002). Providing an exhaustive list of all possible types of delusional contents extends beyond the scope of this paper, since the delusional character of a thought does not stem from its content, as will be argued for below. 


Delusions are symptoms of several mental disorders. They constitute a cardinal symptom of psychotic disorders such as schizophrenia, schizophreniform disorders, schizoaffective disorders, Alzheimer dementia, mood disorders as well as many other major neuropsychological impairments. Delusions can also occur in cases of Huntigton’s disease, Parkinson’s disease, multiple sclerosis or even traumatic brain injuries, as well as in cases of substance-induced disorders. Delusional beliefs may also appear in isolation, as the only symptom presented by patients. In such a case, delusions occurring over a period of at least one month constitute a sufficient criterion for the diagnosis of a delusional disorder, which might then be categorized into further sub-types, depending on the content of delusions. The heterogeneity of disorders involving delusions as a possible symptom and the diversity of co-occurring symptoms undermine the idea of defining delusions in relation to the psychiatric conditions in which they occur. The heterogeneity of the phenomenon of delusions leads empirical research to consider delusions as a symptom deserving specific investigations (Coltheart, Langdon & McKay, 2011, p. 282). 


As a specific unit of scientific investigation, delusions raise important difficulties. The first of these difficulties is that, in spite of its omnipresence in both the research and the clinical psychiatric literature, no adequate definition of the concept of ‘delusion’ has been provided yet. This situation obviously undermines current and future research aiming to account for the underlying mechanisms responsible for delusions. The aim of providing a cognitive neuropsychological account of delusion requires a theoretically adequate definition of delusions in the first place, on the basis of which a relevant group of individuals might be selected, for the purpose of investigating, in a second step, what the neuropsychological underpinning of delusions is. 


This paper aims to propose a theoretically adequate definition of the concept of ‘delusion’, according to which delusions are mental properties characterized by an abnormal asymmetrical inferential role. It does not aim to discuss the nature of the various psychiatric conditions in which delusions occur as a symptom or the pathological character of these conditions. We start with a critical evaluation of the DSM-5 definition of the concept of ‘delusion’ (section 2) in order to shape our positive account, according to which delusions are mental properties characterized by their asymmetrical inferential role in the sense of being immune to the inferential power of the other mental properties a subject instantiates (section 3). This proposal, which escapes the criticisms addressed to the DSM-5 definition, is critically evaluated and defended against several possible objections (section 4). Finally, we show how such a functional definition might benefit from recent neuropsychological research to individuate higher-level sub-types of delusions, in the interest of building an empirically informed psychological classification of delusions (section 5).

What Delusions are Not

The few definitions available in the current research literature focus on the behavioral manifestations of delusions and, most of the time, characterize them in epistemic terms. As such, they follow the recommendations of the American Psychiatric Association holding since the publication of the DSM-III (APA, 1980, p. 7), which favors an atheoretic and descriptive approach to psychiatric conditions. In addition to the above mentioned heterogeneity, the attempt of defining the concept of ‘delusion’ faces several further difficulties, which shall be illustrated by a critical analysis on the often quoted DSM-IV-TR (APA, 1994, p. 821) definition, which is also to be found with a minor modification in the recently published DSM-5 According to the latter, a delusion is characterized as ‘a false belief based on incorrect inference about external reality that is firmly held despite what almost everyone else believes and despite what constitute incontrovertible and obvious proof or evidence to the contrary. The belief is not one ordinarily accepted by other member of the person’s culture or subculture’ (APA, 2013, p. 819). The DSM-5 is a diagnostic manual, which is subject to multiple and heterogeneous, including practical and clinical, constraints and it would be unfair to consider the above mentioned characterization as a genuine theoretical definition. Keeping this in mind, we shall nonetheless proceed to a critical review, which shall frame our positive proposal of definition (section 3). 


A first set of controversies stems from whether delusions are necessarily beliefs. A number of authors have casted doubts on such a doxastic conception of delusion, emphasizing the experiential and phenomenological character over the doxastic one (Gold & Hohwy, 2000) or interpreting delusions as attitudes towards representations in the sense of imaginings (Currie, 2000). Non-doxastic accounts of delusions are also motivated by the claim that delusions are not beliefs because they are not correctly integrated with other beliefs and patients might fail to act in accordance with their delusions. This failure of epistemic and agential rationalities suggests that the conditions for belief ascription are not met in the case of delusions. Moreover, it is a disputable question whether beliefs having impossible content, e.g. ‘I am both in Paris and in Boston’, or self-defeating content, e.g. ‘I am dead’, are receivable beliefs. We shall come back to debate about the doxastic status of delusions in sect. 4.4. 


Second, the content of delusions might contingently turn out to be true (Peters, 2001). Cases were reported in which patients had delusional jealousy, in the sense that they had strictly no indication justifying their beliefs, which turned out to be, contingently, true (Enoch, 1991). Thus, falsehood is not necessary for delusionality. It is obviously not sufficient either, with the result that the opposition truth/falsehood is irrelevant to the delusional status of beliefs. Thus, it should not figure in an adequate theoretical definition of delusions. 


Third, the DSM-5 definition explicitly treats the delusional character of beliefs as context-dependent upon the social environment of the deluded subject, with the result that delusions are treated as extrinsic properties. Admittedly, the context-sensitive approach of the DSM-5 definition is highly supported by pragmatic considerations, since it prevents the ‘pathologization’ of billions of individuals with regards to their religious convictions, for instance. After all, religious convictions share several features with delusions in the sense that they are often firmly sustained and strongly resistant to counter-evidence, while being weakly supported by evidence. Thus, from a pragmatic point of view, it seems justified to have inserted a clause that is designed to exclude such examples of culture-dependent beliefs from the range of possible delusions. 


However, binding the delusional character of beliefs to the opinions of members of a given community is highly questionable from a theoretical point of view. Admittedly, being isolated in believing that p is not sufficient for delusionality. Had it been sufficient, it would have led to a systematic rejection of minority opinions as delusional, which is in turn not acceptable. It is not necessary either, since it would imply that it is not possible for the majority of the individuals of a given referential context to be deluded in believing that p simply because they are a majority. Moreover, to the extent that one accepts a distinction between being deluded and simply being mistaken toward the same propositional content p, the same argument would lead us to consider that it is not possible to have a delusion that p if the majority of the surrounding collective believes, mistakenly but not delusionally, that p. For these reasons, the delusionality of a given belief having content p should not be evaluated relatively to the opinions of the surrounding social community towards p. 


Fourth, the content of delusions might concern highly subjective domains where it is not obvious how to discern truth and falsehood, for instance in the case of aesthetic judgments (Blaney, 2009). This is obviously a domain in which truth and falsehood are difficult to evaluate, and thus where this distinction is of little help in determining the delusional character of a belief. 


Fifth, the DSM-5 requirement that delusions are ‘firmly held’ appears to be empirically inadequate. Myin-Germeys et al. (2005) have shown that the confidence of patients in the veracity of their delusions may vary over the course of one single day, suggesting that delusions are not always firmly sustained. Garety et al. (2005) report that some delusional patients accept that they might be mistaken and that their beliefs, which are believed nonetheless, might be false. Thus, although firm sustainment might be indicative of delusionality, it cannot be considered as a necessary condition. It is obviously not a sufficient condition either. 


Sixth, delusional contents are not necessarily about the ‘external world’. Delusions might concern internal parts of the patient’s body, like in somatic delusions, or delusions about internal states of the patient’s body, like in the Cotard delusion. Delusions might also refer to states or properties of the patient’s mind, as in the case of thought insertions. As a matter of fact, the proximal cause of inserted thoughts is the deluded patient herself who has the delusional belief that these thoughts were inserted into her mind. 


Moreover, assuming that delusions are either bizarre or non-bizarre1, the following general argument shows that delusionality cannot be equated with specific contents. First, specific non-bizarre delusional contents are neither necessary nor sufficient for delusionality. The content that ‘the FBI is tracking me’ is obviously not necessary for delusionality since there are other delusional contents. It is neither sufficient, since it can turn out that the subject having that belief is actually being tracked by the FBI. Consider now cases of bizarre content. Although bizarreness might be sufficient for delusionality, like in the Cotard delusions for instance, it cannot be necessary for the reason that there are non-bizarre delusions; consequently, content bizarreness cannot be equated with delusionality. Thus, although specific contents might constitute useful clinical indicative clues, the delusionality of beliefs in general cannot be assessed on the basis of content alone. 


At this point, it is clear that the DSM-5 definition of delusion is highly inadequate from a theoretical point of view. The short evaluation of the DSM-5 definition revealed that i) the doxastic status of delusions is disputed and that defining delusions either ii) in relation to the world, as being false beliefs, or iii) in relation to beliefs shared in the social environment of the subject is inadequate. This is especially vivid when iv) the delusion concerns highly subjective judgments. Moreover, v) delusions are not necessarily firmly sustained and vi) their content does not necessarily concern the external world either. In fact, delusionality cannot be equated with any specific content. The descriptive character of the DSM-5 definition undermines its theoretical adequacy since it focuses on ‘surface’ features of delusions, which are individually neither sufficient nor necessary for delusions, and to this extent inadequate to define delusions. In what follows, we aim to propose an adequate definition of delusions using the descriptive tools of functionalism, which is in line with, and substantiates, the idea introduced in the recently published DSM-5 that delusions are “not amenable to change in light of conflicting evidence” (APA, 2013, p. 87). 

Towards a Functional Definition of Delusions

Broadly construed, functionalism is the thesis that mental properties are individuated by the causal roles they play in the production of behavior and their relations to further mental properties. As any major position in philosophy, functionalism comes in several flavors, each of which has its own virtues and shortcomings. In what follows, we rely on Lewis’s approach of the individuation of functional properties (1966, 1972). The core idea is that mental predicates are theoretical terms defined by their relations to other terms, both theoretical and non-theoretical. Thereby, each mental predicate grasps a network of causal relations obtaining between the corresponding mental property and the environment, the behavior and other mental properties instantiated of the individual under consideration. In a nutshell, the definition of mental terms is derived from a whole theory T(M1, …, Mn) defining necessarily a collection M1, …, Mn of such terms. Skipping well discussed technical details (for more details relative to ramsifications, see Braddon-Mitchell & Jackson, 1996; Levin, 2010; Lewis, 1970), it is possible to derive applicability conditions for mental predicates, as illustrated here, where x1, …, xn designate other mental property types. 


(F) y has mental property Φ if and only if (∃x1)(∃x2)(∃x3) … (∃xn) [T(y has x1) ∧ (x1 is caused by environmental conditions c1 ˅ c2 ˅ … ˅ cn) ∧ (x1 causes behavior b1 ˅ b2 ˅… ˅ bn) ∧ (x1 stands in relation to x2, x3xn such that x1R1x2 ∧ x1R2x3, ∧ … ∧ x1Rn-1xn)]


To give an over-simplified example, the belief that there is a vanilla ice cream in the fridge is i) typically caused either by the fact that there is a fridge containing a vanilla ice cream in the immediate environment of the subject or by someone saying that it is the case; ii) it typically causes the behavior of walking to the fridge if iii) the subject knows where the fridge is and if she desire to eat the vanilla ice cream.


As definition (F) shows, the functional definition of any mental property is a conjunction of three distinct and themselves disjunctive specifications, describing i) possible environmental causes, ii) possible behavioral effects and iii) relations they necessarily have to other types of mental properties. The holistic feature grasped by (iii) stems from the fact that functionalism defines mental types starting from a whole psychological theory, which specifies the relations between these types. 


Although it is debatable whether delusions are beliefs, it is uncontroversial that delusions are mental properties with the result that they are, to the extent that the functionalist picture is correct, functionally definable. In what follows, we shall investigate whether functionalism enables to grasp what is distinctive about delusions. We shall therefore examine each of these clauses relative to delusions. 


Is there anything special to the causes of delusions? Note that a distinction has to be drawn between causes of mental property instances as past events and the neurological underpinnings of these mental properties. The former diachronically cause instantiations of delusional mental properties. The latter concern the underlying neurological properties, constituting the neurological supervenience basis of these mental properties, in virtue of which, synchronically, these ones occur. The underlying neurological properties are targeted by neuropsychological reductive accounts of delusions. However, with regards to the functional individuation of delusions, we are interested in first category of causes. For these will contribute to the individuation of a typical causal role, which in turn might itself deserve a reductive neuropsychological explanation. 


Regarding higher-level diachronic causes of delusions, let us make two observations. First, there is not any available evidence of any encompassing higher-level non-mental cause of delusions. It is not the case that delusions are always caused by prototypical and regular events that might be subsumed under a general category, which then would abstractly grasp a generic cause of delusions. Second, the very idea of isolating a generic higher-level cause is undermined by current neuropsychological research, which provides good evidence that the crucial variable explaining certain delusions, such as the Capgras delusion, are brain abnormalities (for two partially converging but nonetheless competitive accounts, see Coltheart, et al., 2011; Corlett, Taylor, Wang, Fletcher & Krystal, 2010). Provided that these abnormalities escape the descriptive resources of psychology, it is highly unlikely that a general account of delusions relying exclusively on higher-level causes could be offered. 


Let us turn to the behavioral effects of delusions. To begin with, the existence of behavioral effects of delusions is indisputable. If delusions had no effects, then we would not be able to identify them and diagnose patients. The most prominent behavioral effects are the verbal reports, on the basis of which delusional beliefs are recognized. However, as explained earlier, the content of these verbal reports cannot per se constitute a functional marker because any verbally reported content may or may not be delusional, as argued for above. Aside these verbal reports, delusions might contribute to steer behavior, sometimes with dramatic consequences (see for instance Ames, 1984). However, as pointed out in anti-doxastic accounts, delusions might fail to have expected behavioral implications (Sass, 2001). For instance, Capgras patients often fail to take any initiative against the believed stranger living in their home (Lucchelli & Spinnler, 2007). This lack of systematicity in behavioral effects prevents a functional definition of delusions to focus on that point. Furthermore, such an attempt would furthermore raise the problem that other mental properties might also fail to have the expected behavioral effects for various reasons. Finally, besides the fact that delusions impact on verbal behavior and might impact more broadly on behavior, we are not aware of any study showing that deluded patients have systematic and distinctive behavioral abilities or impairments, which could be isolated in a specific experimental behavioral protocol. 


Provided that delusions fail to have any encompassing causes or specific behavioral implications, a functional definition of delusions shall focus on the causal and inferential relations between delusions and the other mental properties instantiated by the delusional subject, i.e. the third clause of the functional definition. The resistance of delusions to counter-evidence suggests that they might be adequately described as mental properties presenting an abnormal inferential profile, which would be responsible for the apparent disconnection of the delusional beliefs from the rest of the patient’s system of mental properties. We focus on the inferential role as a special case of the broader category of causal roles. The reason for this is that it is obviously possible to influence delusional beliefs causally without this process qualifying as "inferential", e.g. by means of medication. Inferences are here taken to be mental events, i.e. the transition from one mental state to another, which is describable in terms of specific relations between the contents of the states. In other words: Inferences are mental events that can be adequately described by arguments. This general characterization is meant to include different forms of inferences, notably monotonic (deductive) and non-monotonic (inductive) forms.


Let us elaborate further on this proposal, in the interest of making it more specific. One might first consider that delusions are in total inferential isolation relatively to other mental properties a subject has. Such a view entails that i) neither delusions are sensitive to implications of other mental properties ii) nor that they inferentially affect them. This double feature might be grasped in the last bracketed conjunct of definition F by breaking the set of inferential relations R1, …, Rn-1, in which the delusion stands towards other mental properties, into two categories. The first category describes how delusions forwardly affect other mental properties. For instance the delusion that one's wife is an impostor, together with the belief that this has not always been the case, implies that she has at some point in time been replaced. The second category of inferential connections specifies how the mental property is backwardly affected by other mental properties. For instance, the belief that it is currently raining should be revised if one comes to form the belief that the lawn is dry. 


Requirement i) captures the clinical reality of delusions, which are characterized by their absolute resistance to counter-evidence and counter-argumentation. However, requirement ii) is definitively too strong for several important reasons. First, if delusions had no forward-looking inferential power, they might not give rise to desires and intentions of actions, with the result that patients would simply not manifest their delusion at all. As a consequence, it would be impossible to ascribe delusions to anyone. Second, it does not match the clinical reality of delusions. Patients tend to re-arrange their system of mental properties in order to make as much sense of the delusion without, however, revising it (Stephens & Graham, 2004). This clearly shows that delusions do inferentially impact on the patient’s mental life. Finally, as we shall see in more detail below, current neuropsychological accounts assume that delusions explain the content of antecedent abnormal experiences. Such an explanatory function presupposes that delusions have certain inferential power. The same holds for psychodynamic accounts of delusions, according to which delusions serve a psychological defensive function. For instance, delusions of persecution aim to protect one from low self-esteem and depression, by attributing the responsibility of negative events to some malevolent third party rather than to oneself (Bentall & Kaney, 1996; Kinderman & Bentall, 1997). Thus, these accounts also confer a certain degree of inferential potency to delusional thoughts.2 


Hence, our central claim is that rather than standing in complete inferential isolation, delusions are adequately characterized as thoughts presenting an abnormal asymmetrical inferential profile, in the sense that they might inferentially impact on other mental properties instantiated by a patient, while themselves being inferentially insensitive to these other mental properties. With regard to the formal definition of mental properties provided earlier, the proposed characterization of delusions is to be understood as follows. The last conjunct of the functional definition of delusions, ‘x1 stands in relation to x2, x3xn such that x1R1x∧ x1R2x3, ∧ … ∧ x1Rn-1xn’ describing the relations R1, …, Rn-1 among mental property type x2, x3…xn, includes only forward-looking inferential relations describing the inferential impact of the delusion on other mental properties. However, this expression, in comparison to the description of non-delusional beliefs having the same propositional content, discards backward-looking specifications, grasping thereby the fact that delusions are not inferentially affected by other mental properties. Our proposal is thus to define delusions as mental properties fulfilling a psychological functional role, which has the specificity of presenting an asymmetrical inferential profile. This abnormality accounts for the insensitivity of delusional beliefs to counter-evidence and their impact on other mental properties. The next section aims to discuss the theoretical adequacy of this definition. 

Virtues and difficulties 

Immunity to Inferential Revision and DSM-5 Definition Criticisms

Let us begin by making clear that the suggested definition is immune to the criticism addressed to the DSM-IV and DSM-5 definition in the previous section. First, our account of delusions does not require delusions to be belief-type properties, although we are sympathetic to doxastic accounts of delusions (see section 4.4). The proposed definition focuses only on the inferential role of delusions. Thus, our account defines delusional thoughts independently of their epistemic relations to the world or to the predominant judgments of the surrounding social environment. We thereby avoid objections targeting attempts to define delusions, second, as false beliefs, or, third, in relation to beliefs of other individuals. This truth-independency, fourth, makes our account immune to the eventual subjective character of delusional judgments. Finally, nothing prevents the delusional content to refer to internal states of the patient’s body or mind. In fact, it does not presuppose anything relative to content. 

Immunity to Revision and the Two-Factor Account 

A second major advantage of the proposed definition is that it matches the Coltheart and colleagues’ analysis of delusion that led to the Two-Factor Account of delusions (for a recent overview, see Coltheart, et al., 2011; hereafter TFA; first introduced to our knowledge by Davis, Coltheart, Langdon & Breen, 2001). Before going into the detail, note that while we consider this match as a welcome result, we are not committed to this specific account. It might turn out that there is empirical evidence against the TFA, which nevertheless does not affect our account. 


The TFA core idea is to explain why delusions occur and why they persist by appealing to two distinct cognitive factors. In order to clarify both the TFA and why we favor this model over competing neuropsychological accounts, such as the one proposed by Corlett and colleagues (2010), let us focus on the example of the Capgras delusion. Ellis and Young (1990) proposed that the Capgras delusion occurs as the result of a breakdown at the level of the autonomic processing of familiar face recognition. In a nutshell, patients with Capgras are able to consciously recognize and identify familiar faces without undergoing the affective response associated with perception of familiar faces with whom the subject is emotionally related (Ellis, Young, Quayle & de Pauw, 1997; Hirstein & Ramachandran, 1997). According to Coltheart et al., patients confronted with such an unexpected and abnormal experience explain the absence of emotional arousal by forming, on the basis of abductive reasoning, the explanatory hypothesis, i.e. the Capgras delusion, that the relative they are visually recognizing has been replaced by an imposter. In this way, patients are able to explain the absence of emotional arousal (Coltheart, 2007, p. 1048). 


However, this factor is not sufficient to explain the occurrence of the Capgras syndrome. Tranel et al. (1995) reported patients presenting an abnormally low autonomic answer to familiar faces without developing any form of delusional thinking. Assuming that such dissociations can also be observed in other forms of delusions, Coltheart and colleagues introduced a second explanatory factor. More specifically, this second factor consists in a defective system of belief evaluation, whose explanatory function is to account for the non-rejection of the delusional explanatory hypothesis. This is fully consistent with numerous reports of right hemisphere dysfunctions in several types of delusional thinking. Given space limitations, we refer the readers to Coltheart’s papers for the application of the TFA to several cases of delusions (especially Coltheart, 2007, 2010; Coltheart, et al., 2011). 


Evidence reported in Tranel et al. raises difficulties for alternative neuropsychological accounts of delusions that attempt to subsume both factors under a unique deficit in Bayesian inference. According to these accounts, abnormalities in predictive learning mechanisms engender inappropriate percepts which update future priors, leading to the formation and maintenance of delusions (Corlett, Frith & Fletcher, 2009; Fletcher & Frith, 2009). Corlett et al's account is motivated by evidence that delusions might be induced in healthy volunteers by the administration of ketamine (2007). The yet unmet challenge for one-factor accounts of delusions is either to account for the specific content of the delusion developed by patients if the ketamine action is understood as impacting on the belief evaluation system or, if the ketamine action is interpreted as generating the abnormal experience initiating the delusional thinking, to account for the dissociation reported by Tranel et al. 


In brief, the general explanatory scheme of the TFA is that delusions occur as a result of the interplay of two distinct factors. The first, labeled ‘A-factor’, is responsible for the occurrence of an unexpected abnormal experience. This abnormal experience leads to a belief that explains the occurrence of the abnormal experience in an abductive style. As such, the A-factor aims to explain the specific content of a delusional belief. However, if the A-factor explains why strange beliefs might be instantiated, it does not per se explain why such beliefs are not rejected as inconsistent with the rest of the system of mental properties characterizing a given individual. TFA precisely explains this phenomenon by reference to a second factor, labeled ‘B-factor’, which consists in a defective belief evaluation process, which prevents the explanatory thought from being rejected. Let us add two remarks: First, abductive hypothesis generation and belief evaluation processes are two distinct cognitive processes (Coltheart, 2010, p. 24). Second, it is important to note that according to TFA, the B-factor is the same in any case of delusion, whereas the A-factor varies with the content of the delusion. For instance, patients with the Fregoli delusion undergo an abnormally high autonomic response to unknown faces, while Cotard patients do not undergo any autonomic response to familiar faces at all. Accordingly, the TFA account explains the immunity to revision by the B-factor and the specific content by the A-factor. 


Two broader comments are worth making here. First, the TFA and our account are mutually complementary in the sense that the B-factor reductively explains the immunity of delusions to revision. Thus, the B-factor might be seen as what gives the delusional character to the explanatory hypothesis developed by the patient. However, and this is the second remark, it could be argued at this point that delusions are indeed sensitive to the inferential impact of other mental properties, precisely because delusions occur as a result of an abnormal experience. This would suggest that delusions are indeed inferred from the content of that experience, which obviously conflicts with the proposed view. However, this objection is based on a misinterpretation of the modus operandi of the delusional explanatory hypothesis. The latter, instead of being inferred from the abnormal experience, aims to account for its content in the sense that if the hypothesis were true, then the abnormal content would be expectable. In other words, the explanatory value of the delusional hypothesis stems from its explanatory success towards the problematic content (Coltheart et al., 2011, p. 284). Hence, the explanatory value of the delusional hypothesis can be fully understood on the basis of its own (forward-looking) inferential power without assuming any (backward-looking) inferential sensitivity. 

Overinclusive character

The process of redefining a phenomenon always raises difficulties and our proposal does not escape this rule. In spite of its theoretical inadequacy, the conjunctive form of the DSM-5 definition imposes several types of constraints, which, taken together, have the effect of limiting the extension of the concept of ‘delusion’. The suggested view may seem overinclusive at first glance, since it focuses only on the inferential profile of delusions. However, immunity to revision is a stronger criterion than resistance to counterevidence. Despite the fact that focusing on the inferential profile might enlarge the extension of the concept of ‘delusion’, the chosen criterion is stronger, in the sense that it rules out beliefs that are resistant but not immune to counter-evidence. Thus, although the balance between these constraints might be debatable, the proposed definition does not lead straightforwardly to an enlargement of the extension of the concept of ‘delusion’. 

The Doxastic Status of Delusions 

A major objection to the proposed account of delusions consists in saying that immunity to revision might well be necessary for delusionality, while denying that it is a sufficient condition, given that there are beliefs with an asymmetrical inferential profile that should not be considered delusional. Logical truths, Wittgensteinian framework propositions or religious beliefs might be presented as counter-examples. In order to shed light on this objection, let us first clarify our view relative to the doxastic status of delusions. 


Focusing on their inferential profile, the proposed account does not strictly require delusions to be belief-type mental properties. This is welcome since both the aforementioned existence of non-firmly sustained delusions and the fact that delusions might strongly conflict with the norms of both epistemic and agential rationalities suggest that at least some delusions do not meet the conditions for beliefs ascription. However, Bortolotti (2010, especially chapt. 3 and 4) convincingly argued that these arguments rely on too demanding standards of rationality. Indeed, perfectly normal individuals fail to meet these constraints, which stem mainly from normative standards of academic scientificity  rather than from descriptive psychology. 


Furthermore, the theoretical function of delusions in most neuropsychological accounts suggests that delusions are indeed beliefs. According to these accounts, delusions are abductive explanatory hypotheses that are supposed to restore order on a disordered perceptual and cognitive world: patients form and maintain delusions because delusions contribute to explain abnormal experiences. As explained, delusions are explanatory in the sense that if they were true, then the abnormal perceptual or cognitive phenomena affecting deluded patients would indeed be expectable (Coltheart et al., 2011; Corlett et al., 2010; Fletcher & Frith, 2009). This explanatory function of delusions supports the doxastic account of delusions because delusions can play this explanatory role only if the deluded patient takes the world to conform to the delusional content at least in some sense. 


At this point, however, we propose that delusions might be beliefs without taking this proposal to express a necessary requirement for delusions. Let us put forward the following view. Omitting technical details, Bayesian approaches of learning assume that any propositional content might be more or less ‘believed’ by a subject, in the sense of having a certain degree of credence (for an overview, see Joyce, 2008). The credence in a given propositional content is reinforced by positive evidence and weakened in light of counter-evidence. The Jeffrey conditioning (Jeffrey, 1987) describes the ideal way of updating credence values in a system of beliefs in the light of new evidence. The explanatory success of abductive delusional hypotheses in accounting for the abnormal experiences contributes to the reinforcement of their conviction in their delusional hypotheses. In normal cases, such reinforcement would be counter-balanced by the inconsistency of the delusional content with other beliefs of the subject, which should have the consequence of lowering down the credence to the delusional content. However, deluded patients are also impaired at that level. Coltheart et al. (2011) precisely assume that deluded patients suffer from a double impairment. The first one is responsible for abnormal experiences. The second consists in a dysfunctional Bayesian mechanism of belief evaluation, which prevents the correct evaluation of the delusional thought. Our account might be regarded as grasping this insensitivity to counter-evidence by appealing to the descriptive resources of functionalism, i.e. in terms of an asymmetrical inferential profile. In other words, it is because of their abnormal inferential profile that the Bayesian reinforcement of credence in the abductive explanatory hypothesis is not counter-balanced by negatively relevant evidence. This is precisely why the explanatory hypothesis is insensitive to counter-evidence. 


This interpretation nicely fits the fact that different degrees of conviction might be empirically observed. On the one hand, some delusions are indeed not absurd or in obvious contradiction with available evidence. As said above, they can even be contingently true. Delusions might indeed receive a very high degree of credence either because the abnormal experience that they aim to explain occurs frequently, or because they might contribute to explain other phenomena. For instance, the belief that an impostor has replaced my wife might explain why my fourteen years old daughter is that difficult with and rude to my wife. On the other hand, other delusions might explain only very rare abnormal perceptual experiences without offering any other explanatory benefits, with the result that they benefit from very little support and have a considerably low degree of credence. 


How weak can the credence of a delusion be? Neuropsychological accounts require delusions to explain an abnormal experiential content. This bottom line establishes the minimal degree of credence of a delusion. The question is now as to whether such a minimal degree of conviction is sufficient for a delusion to qualify as a belief. Since our focus here is not on extensively discussing the conditions for belief ascription (we are rather aiming at providing support for our claim that delusions are mental properties characterized by an asymmetrical inferential profile), let us be relatively liberal. On the one hand, one might follow Bortolotti in saying that delusions are beliefs although patients sometimes fail to be consistent and act on their delusions (Bortolotti, 2010). This is perfectly in line with our approach. On the other hand, if one relies on very high standards of rationality, one might always consider that even though delusions are not beliefs, they are nonetheless thoughts in the sense of amodal, abstract, conceptual mental representations (Vosgerau & Synofzik, 2010), which are in this case characterized by an asymmetrical inferential profile. Accordingly, weakly sustained delusions, qua being successful explanatory hypotheses, have a degree of credence placing them on a continuum between purely subjunctive thoughts and beliefs, at a location which might be labeled ‘successful abductive hypothesis without any further support’. Whether or not this has to be counted as a belief is a question we let to the reader; the important point is that both options, doxasticism and moderate anti-doxasticism, are compatible with our account. Let us thus continue assuming that delusions are, at least, thoughts in the above sense and that they might be beliefs. 

Immunity to Revision is Not Sufficient for Delusionality 

As already mentioned, one might argue that there are beliefs that are immune to revision that should not be considered delusional. Logical truths, Wittgensteinian framework propositions or religious beliefs could count as counter-examples to the suggested view. Considering such cases of beliefs, we now return to the objection according to which immunity to revision is not sufficient for delusionality. 


Arguably, it is highly unlikely that circumstances requiring revision of statements expressing logical truths would occur. This, however, does not mean that logical truths might not be questioned and that they are in principle not revisable. Famously, Quine argued that any proposition is in principle subject to change (1951). History of science exhibits several examples in which axiomatic principles of logic were in fact modified. For instance, the law of the excluded middle has been shown questionable with respect to certain interpretations of quantum mechanics (Esfeld, 2000) and the axioms of Euclidian geometry have been dismissed by Einstein’s General Theory of Relativity (Misner, Thorne & Wheeler, 1973, p. 191). 


Campbell (2001) and Eilan (2000) proposed an understanding of delusions as Wittgensteinian framework propositions. Framework propositions are anderstood as a ‘heterogeneous class of propositions that we do not question and which globally constrain our inferences and our interpretation of our experience’ (Eilan, 2000, p. 108), without any restrictions to foundational logical and philosophical principles. Equipped with this idea of framework propositions, Campbell and Eilan attempted to shed light on the incorrigibility of delusions. However, Thornton (2008) showed that delusions and framework propositions are dissimilar in several aspects, pushing the analogy in question to the status of a metaphor, and thus allowing it only little explanatory interest. But the crucial point here is that the fact that we actually do not question framework propositions does not entail that they lie beyond the range of questionability and revisability; and if they do, then we do not see why the corresponding beliefs should not be considered delusional.


Finally, the DSM-5 excludes beliefs spread out in the community in order to prevent the pathologization of billions of individuals because of their religious beliefs, for instance. The fact that our proposal does not have such ad hoc restrictions raises the question as to whether it precisely implies the pathologization of religious beliefs. First, we do not see why such beliefs should in general be considered immune to revision. It is not because believers do not revise their convictions that those are immune to revision if certain possible circumstances were actualized. Our account implies that delusions are immune to revision in the sense that there is no possible evidence in light of which delusions may be inferentially revised. This is not necessarily the case with religious beliefs.3 From that point of view, it seems that such beliefs should not necessarily be considered delusional, although they might be delusional depending on their revisability status. Second, considering that any non-revisable belief is delusional does not imply that any delusional belief is pathological, as will be argued below. 

Pathological Status of Delusions 

Admittedly, the daily lives and social interactions of most patients suffering from delusions are considerably impaired, with the result that their condition is called pathological. However, these impairments are remote and indirect effects of delusions and are not constitutive of delusions. Delusions consist in abnormal thoughts or beliefs, which directly cause patients to formulate (eventually socially inadequate) verbal reports and, in some cases, to engage in inadequate behaviors, which in turn negatively impact on their lives. It is, however, due to this class of indirect effects that the condition of patients is termed ‘pathological’. One way to appreciate the difference between ‘delusionality’ and ‘pathologicality’ is to observe that delusionality should be defined intrinsically for the reasons given above, whereas pathologicality is an extrinsic property. Surely, in a certain sense, patients with delusions are often in a pathological psychiatric condition because of their delusions. But the pathological character of their condition does not merely depend on their delusion: it is also partly because the patient’s surrounding social environment is not sharing their conviction, that patients are driven into important difficulties in several dimensions of their lives. 


To make this point clearer, imagine a possible world w in which someone, call him John, has the delusion that there are a certain number of synthetic non-human individuals that look exactly like humans in his environment. Assume also that in this world the other human individuals with whom John discusses about these synthetic agents share John’s opinions for the reason that it is true that there are such individuals in w. All of this is fully compatible with the proposed definition since the latter does neither rely on the absence of justification of the delusion nor on its epistemic relations to the world. John is deluded because, contrary to the other inhabitants of this world, his belief is immune to revision. However, it is unclear why John's condition would have to be considered as pathological, because nothing prevents John from living a normal life. The crucial element is that it is not the delusion itself that makes a psychiatric condition pathological. Rather, it is the problems for one’s integration into social formations, i.e. the abilities to work, organize one’s life, etc.4 Indeed, the American Psychiatric Association explicitly recognizes that ‘the symptomatic presentation by itself […] is not inherently pathological and may be encountered in individuals for whom a diagnosis of ‘mental disorder’ would be inappropriate’ (APA, 1994, p. 8). In a nutshell, our claim is that whereas the pathological status of a psychiatric condition is an extrinsic property, the delusional status of the properties contributing to this medical condition is an intrinsic property. Thus, delusions are not necessarily pathological. As a result, the intuition according to which there are beliefs that are immune to revision but not delusional because they are not pathological should be defused. Indeed, there are no reasons to think that a definition of delusion should ipso facto explain why delusions are pathological. 

Waxing and Waning Delusions

A fourth source of difficulties is raised by reported cases of waxing and waning delusions, which suggest that delusions might be revisable, and that, consequently, our proposal is too strong. Two categories should be distinguished here. The first concerns cases discussed in Garety’s (2005) and Mying-Germeysa’s (2005) studies in which the strength with which a delusion is held was shown to vary. These cases are unproblematic for our proposal since the patients involved did not revise their delusional beliefs on inferential grounds. Patients are rather waxing and waning relative to the degree of conviction they manifested toward their delusional thoughts. 


The second category is slightly subtler since it concerns patients that are not always deluded. Bisiach, Rusconi and Vallar (1991) reported waxing and waning delusions in somatoparaphrenia in response to vestibular stimulation. The authors demonstrated that the delusion of patient AR that her paralyzed left arm was her mother's disappeared for a short period of time after the reception of a cold caloric left vestibular stimulation. However, the case of patient AR is not a counterexample to our account of delusions, since lower-level intervention on AR’s nervous system perfectly accounts for observed psychological changes, and there is therefore no reason to think that AR’s delusional convictions were modified by means of an inferential process subsequent to an intervention on the A-factor, since the limp was still paralyzed. The best available explanation of these findings is that the lower-level intervention restored AR’s belief revision system, with the result that AR was subsequently able to revise her beliefs and was therefore, subsequently, no longer delusional. The crucial point here is that AR’s temporary remission was not due to an inferential process of revision simpliciter, but to the lower-level intervention, which temporarily fixed revisability, and led to a subsequent normal process of revision. 

Delusions are Revisable under Medication

The issue of the impact of lower-level interventions on delusions raises the general question of the role of medication in the evaluation of the proposed account of delusions. The efficacy of anti-psychotic drugs is difficult to estimate because of the heterogeneity of psychiatric conditions in which delusions occur, but also due to the evolution of medication itself and the wide range of complementary therapeutic measures possibly received by patients. Focusing on delusional disorders, Manschreck and Khan (2006) where able to confirm previous results by Munro and Mok (1995), showing that medication of patients led in 49.3% of studied cases to complete recovery, to a significant reduction of symptoms in 40.3%, whereas 10.4% of patients have been found resistant to medication. In patients with schizophrenia, conventional antipsychotic drugs appear to produce complete remission of positive symptoms, including delusions and hallucinations in 70-80% of cases (Kane, 1996; Robinson, Woerner & Alvir, 1999). Stronger evidence for this crucial role of medication toward belief revision is even provided by the above mentioned results by Corlett et al., which shows that delusions might be induced and extinguished in healthy volunteers by administration of ketamine (Corlett, Honey & Fletcher, 2007). Given that one-factor based accounts of delusions encounter difficulties in explaining the specific contents of delusions, which are accounted for by Coltheart's A-factor, we are inclined to interpret these results as suggesting that anti-psychotic medication impacts on the B-factor. 


The impact of medication on the revisability of delusions should be interpreted in parallel to AR’s case. The administration of anti-psychotic drug has to be considered as a lower-level intervention on the patient’s cognitive system, which has first the effect of restoring the (at least partially impaired) ability to evaluate and revise beliefs normally. From there onwards, patients are no longer deluded. In a second step, they might consequently revise on purely inferential grounds thoughts that were delusional. In other words, the medication restores the patient’s capacity for belief evaluation and revision, such that they revise thoughts or beliefs that are, at that time, no longer delusional but simply incorrect. Accordingly, it is not the case that medicated deluded patients inferentially revise their convictions simpliciter; the remission process is crucially a two-step process. This interpretation of the remission process is supported by recent evidence concerning the efficiency of cognitive behavioral therapies (hereafter CBT) in the treatment of delusions. 

Cognitive Behavioral Therapy is Efficient on Delusions

A fifth objection is raised by studies having shown the efficiency of CBT in the treatment of delusions (Durham et al., 2003; Garety, Fowler & Kuipers, 2000; Landa, Silverstein, Schwartz & Savitz, 2006; O’Connor et al., 2007). The modus operandi of these therapies suggests that delusions might be revisable on inferential grounds. However, it has to be observed, first, that this form of therapy does not aim to confront deluded patients directly with counter-evidence. Indeed, there is evidence showing that confrontation with counter-evidence indeed strengthens delusions (Milton, Patwa & Hafner, 1978) and that patients, in such circumstances, engage in a maladaptative processes of rationalization, which enables them to interpret these new facts in accordance to their delusion (Joseph, 1986). CBT rather aims at providing patients with alternative interpretations of delusional contents and teaches them to only gradually engage in reality testing (Turkington, Kingdon & Weiden, 2006, p. 368). 


Second, patients involved in these studies were medicated and we are not aware of any study having shown that CBT enables the inferential revision of delusions in the absence of medication, which arguably has the central role here. As explained above, medication constitutes a low-level chemical intervention on the neurobiological supervenience basis of delusions that has to be interpreted as having the effect of restoring the ability to revise delusional beliefs. The conjunct action of medication and CBT has therefore to be interpreted alongside the case described in the previous sub-sections. Namely, medication in the first instance restores the ability to evaluate and revise beliefs normally. From this point onwards, the patient is no longer deluded and CBT might help patients to engage in a critical evaluation of beliefs that, to the extent that the patient is responsive to medication, are no longer delusional. Against this background, the fact that CBT in combination with medication has been shown to have some efficiency, does not challenge our account of delusions, especially since these studies provide evidence for a reduction in psychotic symptoms in medication-resistant patients, but no complete remission. This supports the claim that medication is the primary path to remission, although patients might highly benefit from an additional CBT.

The Continuum Account of Delusionality 

The view that delusions are categorically distinct from other mental properties (going back to Jasper, 1963) has been regularly challenged since Strauss’ work (1969), which showed that full conviction was not required for delusionality. This observation, combined with the high prevalence of unusual beliefs in non-clinical population, led to the idea that delusions might lie on a continuum with normal beliefs and that delusional ideation is widespread among non-clinical population (Blackwood, Howard, Bentall & Murray, 2001; Chadwick, Birchwood & Trower, 1996; Freeman, Pugh, Vorontsova, Antley & Slater, 2010; Johns & Os, 2001; Peters, Joseph & Garety, 1999). Not only the continuum approach fits the dimensional approaches of psychiatric disorders (Brown & Barlow, 2005), it also enables taking into account differences in the courses of disorders, and in turn facilitates the study of early stages of psychoses, as it contributes to reduced stigmatization of patients by highlighting similarities with the non-clinical population (Mullen, 2003). 


Let us first highlight that our functionalist account of delusions might be adapted to the continuum view, by amending the proposed definition as follows. Instead of considering that the functional description of delusions is characterized by a complete absence of backward-looking inferential specification, this absence might be partial, i.e. only some backward-looking inferential connections might be missing with the result that the delusions would be insensitive to certain patterns of cognitive probing only. Accordingly, just as this absence of backward-looking inferential connections would be a matter of degree, delusionality would also come in degrees. At the extremes of the continuum in question would be, on the delusional side, thoughts that are characterized by a complete lack of backward-looking inferential links, whereas normal beliefs would be on the other side. For any intermediate case, only certain inferential patterns would lead to the rejection of the delusion.


However, the continuum view raises important difficulties, the first of which being to know when delusionality starts. On the one hand, thoughts that are only very weakly inferentially disconnected will not even be identified as defective in any respect. It is only if beliefs present a minimal resistance to counterevidence that one might suspect being confronted with a case of delusion. On the other hand, very high resistance to counter-evidence is not sufficient for delusionality. Scientists, for instance, are extremely resistant at giving up their theories, even when confronted with negative evidence. However, this does not justify considering them as deluded. The problem here is that we don’t see any plausible way how to set up a clear demarcating line distinguishing ‘strong convictions’ from ‘delusions’ elsewhere than between any degree of revisability and complete immunity.


Let us sum up. We started from the uncontroversial claim that delusions are mental properties. Accordingly, to the extent the picture of functionalism is correct, delusions are functionally describable. We argued that the specificity of delusions is their immunity to revision and that this specific feature can be functionally described. The causal role of delusions is characterized by their asymmetric inferential profile: the functional definition of delusions does not include any backward-looking inferential connections. This absence makes these thoughts insensitive to counterevidence. 

Individuating Delusional Sub-Types 

Our claim that delusions are characterized by an abnormal asymmetrical inferential role, which constitutes a necessary but also sufficient condition for delusionality, should not occlude the fact that delusions vary considerably with respect to their etiology, content and behavioral effects. Here, we aim to show how taking this diversity into account might constitute the starting point for developing a causal classification of delusions against the background of the proposed functional definition. 


Let us start with some theoretical considerations. We argued that delusions are mental properties that are functionally characterized by the absence of backward-looking inferential relations, which is specified in the last part of the definition. In order to keep our notation easy to read, let us adopt the following notation: 

(1) DDef: ∀x (Dx  Mx ∧ Sx)


Definition (1) should be read as the claim that anything which is a delusion D necessarily qualifies as fulfilling a mental role M, whatever it is, and as fulfilling an asymmetrical inferential role S. Classifying delusions amounts to construct a collection of functional sub-types Ds1, …, Dsn fulfilling the functional definition DDef but differing in some other respect C1,…, Cn such that any delusion falls under exactly one of these sub-types. Note that parameters C1,…, Cn might themselves be conjunctive. They are here designed to grasp functional specificities that distinguish sub-types of delusions to the extent that they are conjoined with parameter M and S, i.e. Ci is necessary for Ds and the conjunction of Ci, M and S is jointly sufficient for Ds. Thus, the following adequacy conditions holds for the set Ω of sub-types Ds1, …, Dsn of the superordinate type D. 


(2) ∀Dsi∈Ω ∀x (Dsix  Dx

(3) ∀Dsi∈Ω ∃Ci∈Φ ∀x (Dsix  Mx ∧ Sx ∧ Cix


Condition (2) grasps the idea that any sub-type Dsi is a sub-type of D. Condition (3) secures that any delusion falls under exactly one delusional sub-type by requiring that each sub-type of D has to be distinct in some respect Ci in Φ from the other sub-types of D, i.e. Ci is necessary for Di and the conjunction of features M, S and Ci is jointly sufficient for Di.


The permitted values C1, …, Cn of parameter Ci in Φ are constrained by the functionalist character of our approach, i.e. sub-types of D have to be functional (sub-)types in the sense of functionalism. Thus, Ci’s specifications have i) to be descriptive resources available to psychology and ii) to concern the elements figuring in functional descriptions of mental properties, namely environmental causes, behavioral effects or inferential relations to other mental properties, to the obvious exception of backward-looking inferential relations. Thus, the set Φ of specificities C1, …, Cn contains any description which is receivable from a functional point of view minus backward-looking inferential relations. Note that specifications Ci do not have to be specific to delusions (i.e. it is perfectly possible that ∃x (Cix ∧ ¬Dx)). 


In order to illustrate how the proposed functionalist approach might benefit from neuropsychological results when it comes to a taxonomy of delusions, let us return to the TFA. As explained earlier, the TFA completes the suggested view in the sense that it provides lower-level neuropsychological explanations of the given higher-level characteristics of delusions, namely immunity to revision and content. First, qua referring to an abnormal process of belief evaluation, the TFA provides a lower-level reductive explanation of the immunity to counter-evidence. Thus, the B-factor explains why patients satisfy condition D in definition (1). 


Second, the A-factor explains reductively other characteristic of delusions, which shall be grasped in our account by means of parameter C. Since the delusional explanatory hypothesis has the function of accounting for the content of the experience in question, delusions must have certain implications towards the content in question. It is the only way delusions can explain and thus legitimize such experiential contents. Call this inferential specificity I. Furthermore, delusions typically cause verbal reports capturing the delusional content at least partly. Call this functional specificity V. With these precisions at hand, we can claim that parameter Ci in expression (3) should be constructed, for any delusion Di, as a conjunction of specifications Ii and Vi


Consider, for instance, the Capgras delusion. The A-factor consists in an abnormally weak level of autonomic emotional response to familiar faces. This abnormal experience needs to be explained by the patient in order for them to deal with it. The Capgras delusion (Dscap) has then the role of accounting for the content of this experience (Icap) and typically contributes to the production of verbal reports expressing the idea that a stranger has replaced a relative (Vcap). Provided that the B-factor ensures that the Capgras delusion is immune to revision and that the delusion is a mental property, we can, on that basis, construct the Capgras delusion sub-type as follows: 


∀x (Dscapx  Mx ∧ Sx ∧ Ccapx) where Ccapx  (Icapx ∧ Vcapx )


The Fregoli delusion sub-type is also constructed in the same way. The A-factor consists in an enhanced level of autonomic emotional response to unknown faces, which leads patients to form the belief that they are surrounded by familiar people in disguise. Similarly to the Capgras case, the Fregoli delusion has the role (Ifreg) of explaining the content of this abnormal experience and typically contributes to the production of verbal reports describing the content of the delusion (Vfreg). Thus we have:


x (Dsfregx  Mx ∧ Sx ∧ Cfregx) where Cfregx  (Ifregx ∧ Vfregx)


It is not necessary to multiply the examples here. The general idea is to illustrate how a functional definition of delusions might constitute the starting point for the construction of an empirically informed taxonomy of delusions, in the sense that our approach proposes to take into account neuropsychological evidence in order to individuate sub-types of delusions on the basis of conceptual resources that are nonetheless purely functional. Adopting the TFA, a general procedure to classify delusions might be described as follows (closely related to neuropsychological investigation methodology described in Coltheart et al., 2011, pp. 285-286): 


First, for any new mental property token satisfying the proposed definition, examine if the delusion falls under a known sub-type. Second, if the delusion does not fit any existing taxon, then investigate the existence of a neuropsychological impairment, which would be responsible for a perceptual or affective abnormal experience. Third, explain how the content of the delusion under consideration explains the content of the abnormal experience, i.e. show that if the delusion was true, then the abnormal content would be expectable. Fourth, investigate right frontal deficits, which might be held responsible for the irrevisability of the delusion. This step aims to secure that the mental property under consideration is genuinely immune to revision, as opposed to simply being resistant to the actually presented challenging evidence. Finally, on the basis of the verbal reports of the patients and of the content to be explained by the delusions, build a new functional sub-type of delusions, which focuses on i) the absence of backward-looking inferential relation to other mental properties, ii) the forward-looking inferential connections towards the content of the abnormal experience and iii) the verbal reports typically produced by the delusion. 


The resulting taxonomy presents us with two advantages. First, the fact that the functional characterization of delusions is built only out of psychological descriptive resources secures that individuals falling under the same category share genuine and objective psychological similarities. All of them instantiate mental properties presenting an abnormal inferential profile and those falling under the same sub-type of delusions share more specific functional similarities. Second, the possibility of constructing higher-level functional psychological sub-types while taking into account neuropsychological findings enables the taxonomy to be implicitly informative relative to the underlying neurological impairments characterizing the affected patients. 


Furthermore, our account perfectly illustrates how neuroscientific inquiries, broadly conceived, might lead to mutual influences between higher- and lower-level approaches. Indeed, philosophers of neuroscience recently emphasized that it is inappropriate to describe contemporary advances of neuroscience as only a one-way movement toward reductive explanations of higher-level psychological and cognitive phenomena in terms of lower-level neurobiological processes. Arguably, such a simplistic picture ignores important and frequently occurring episodes of scientific progress, during which higher-level research strongly impacts on explanatory choices made by scientists focusing on lower-level investigation, and how these choices might in turn lead to changes at the higher-level (for an historical analysis focusing on spatial memory, see Craver, 2005). Our account might be conceived as being the product of mutual influences between higher and lower-level research, since it combines a higher-level analysis of the concept of delusion with neuropsychological evidence in order to provide an empirically adequate classification of delusions. 

Conclusive Remarks 

Our critical review of the DSM-5 definition showed that delusions ought to be defined independently of their truth-value, if any, and independently of widespread opinions in the surrounding social community. We have also shown that delusions cannot be defined by reference to their content, although bizarre content might actually be sufficient for delusionality. The underlying reason is that, for any non-bizarre content, we can conceive of a scenario in which the belief is perfectly true of the described situation and yet delusional. 


In order to escape these difficulties, we put forth a functionalist definition of delusions focusing on their inferential role. Delusions are mental properties characterized by an abnormal asymmetrical inferential role, which is responsible for their immunity to the inferential impact of other mental properties. This functional characterization of delusions might constitute the starting point for the construction of an empirically informed taxonomy of sub-types of delusions. Taking systematically neuropsychological research on delusions into account, this approach illustrates how higher- and lower-level inquiries might successfully interact within cognitive sciences.5


Ames, D. (1984). Self shooting of a phantom head. British Journal of Psychiatry, 145, 193-194.

APA (Ed.). (1980). Diagnostic and Statistical Manual of Mental Disorders (DSM-III). Washington DC.

APA (Ed.). (1994). Diagnostic and Statistical Manual of Mental Disorders (DSM-IV). Washington DC.

APA (Ed.). (2013). Diagnostic and Statistic Manual of Mental Disorders, Fifth Edition (DSM-V). Arlington: Amercian Psychiatric Association.

Bentall, R. P. & Kaney, S. (1996). Abnormalities of self-representation and persecutory delusions: a test of a cognitive model of paranoia. Psychological Medicine, 26, 1231–1237.

Berrios, G. E. (2002). Erotomania: a conceptual history. History of Psychiatry, 13(52), 381-400.

Bisiach, E., Rusconi, M. L. & Vallar, G. (1991). Remission of somatoparaphrenic delusion through vestibular stimulation. Neuropsychologia, 10, 1029-1031.

Blackwood, N. J., Howard, R. J., Bentall, R. P. & Murray, R. M. (2001). Cognitive neuropsychiatric models of persecutory delusions. American Journal of Psychiatry, 158, 527-539.

Blakemore, S. J., Oakley, D. A. & Frith, C. D. (2003). Delusion of alien control in the normal brain. Neuropsychologia, 41(8), 1058-1067.

Blaney, P. (2009). Paranoid and delusional disorders. In P. Blaney & T. Millon (Eds.), Oxford Textbook of Psychopathology (pp. 361-396). New York: Oxford University Press.

Bortolotti, L. (2010). Delusions and other irrational beliefs. Oxford: Oxford University Press.

Braddon-Mitchell, D. & Jackson, F. (1996). Philosophy of mind and cognition. Oxford: Blackwell.

Breen, N., Caine, D., Coltheart, M., Hendy, J. & Roberts, C. (2000). Towards an Understanding of Delusions of Misidentification: Four Case Studies. Mind & Langage, 15, 74-110.

Brown, T. A. & Barlow, D. H. (2005). Dimensional versus categorical classification of mental disorders in the fifth edition of the Diagnostic and statistical manual of mental disorders and beyond: Comment on the special section. Journal of Abnormal Psychology, 114(4), 551-556.

Campbell, J. (2001). Rationality, meaning, and the analysis of delusion. Philosophy, Psychiatry & Psychology, 8, 89-100.

Chadwick, P., Birchwood, M. & Trower, P. (1996). A cognitive view of delusions and voices Cognitive therapy for delusions, voices and paranoia (pp. 1-24). Chichester: Wiley.

Coltheart, M. (2007). The 33rd Sir Frederick Bartlett Lecture: Cognitive neuropsychiatry and delusional belief. The Quarterly Journal of Experimental Psychology, 60(8), 1041-1062.

Coltheart, M. (2010). The neuropsychology of delusions. Annals of the New York Academy of Sciences 1191, 16-26.

Coltheart, M., Langdon, R. & McKay, R. (2011). Delusional Belief. Annual Review of Psychology, 62, 271-298.

Corlett, P. R., Frith, C. D. & Fletcher, P. C. (2009). From drugs to deprivation: a Bayesian framework for understanding models of psychosis. Psychopharmacology, 206, 515-530.

Corlett, P. R., Honey, G. D. & Fletcher, P. C. (2007). From prediction error to psychosis: ketamine as a phramacological model of delusions. Journal of Psychopharmacology, 21(3), 238-252.

Corlett, P. R., Taylor, J. R., Wang, X.-J., Fletcher, P. C. & Krystal, J. H. (2010). Towards a neurobiology of delusions. Progress in Neurobiology, 92, 345-369.

Craver, C. F. (2005). Beyond reduction: mechanisms, multifield integration and the unity of neuroscience. Stud. Hist. Phil. Biol. & Biomed. Sci, 36, 24.

Currie, G. (2000). Imagination, delusion and hallucinations. In M. Coltheart & M. Davies (Eds.), Pathologies of Belief: Blackwell.

Davis, M., Coltheart, M., Langdon, R. & Breen, N. (2001). Monothematic Delusions: Towards a Two-Factor Account. Philosophy, Psychiatry & Psychology, 8(2/3), 133-158.

Durham, R. C., Guthrie, M., Morton, R. V., Reid, D. A., Treveling, L. R., Fowler, D., et al. (2003). Tayside - Fife clinical trial of cognitive-behavioural theoraphy for medication-resistant psychotic symptoms. British Journal of Psychiatry, 182, 303-311.

Eilan, N. (2000). On understanding of schizophrenia. In D. Zahavi (Ed.), Exploring the self (pp. 97-113). Amsterdam: John Benjamins.

Ellis, H. D., Whitley, J. & Luauté, J. P. (1994). Delusional misidentification: The three original papers on the Capgras, Fregoli and intermetamorphosis delusions. History of Psychiatry, 5, 117-146.

Ellis, H. D. & Young, A. W. (1990). Accounting for delusional misidentifications. British Journal of Psychiatry, 157, 239-248.

Ellis, H. D., Young, A. W., Quayle, A. H. & de Pauw, K. W. (1997). Reduced autonomic responses to faces in Capgras delusion. Proceedings of the Royal Society of London: Biological Sciences, 264, 1085-1092.

Enoch, D. (1991). Delusional Jealousy and Awareness of Reality. British Journal of Psychiatry, 159(14), 52-56.

Esfeld, M. (2000). Quine's Holism and Quantum Holism. Epistemologia, 23, 51-75.

Fletcher, P. C. & Frith, C. D. (2009). Perceiving is believing: a Bayesian approach to explaining the positive symptoms of schizophrenia. Nature Review Neuroscience, 10, 48-58.

Freeman, D., Pugh, K., Vorontsova, N., Antley, A. & Slater, M. (2010). Testing the Continuum of Delusional Beliefs. Journal of Abnormal Psychology, 119(1), 83-92.

Gardner-Thorpe, C. & Pearn, J. (2004). The Cotard syndrome. Report of two patients: with a review of the extended spectrum of ‘délire des négations. European Journal of Neurology, 11(8), 563-566.

Garety, P. A., Fowler, D. & Kuipers, E. (2000). Cognitive-Behavioral Therapy for Medication-Resistant Symptoms. Schizophrenia Bulletin, 26(1), 73-86.

Garety, P. A., Freeman, D., Jolley, S., Dunn, G., Bebbington, P. E., Fowler, D. G., et al. (2005). Reasoning, emotions, and delusional conviction in psychosis. Journal of Abnormal Psychology 114, 373–384.

Gold, I. & Hohwy, J. (2000). Rationality and schizophrenic delusion. Mind and Language, 15, 146-167.

Graham, G. & Stephens, L. (1994). Philosophical Psychopathology: MIT Press.

Hirstein, W. S. & Ramachandran, V. S. (1997). Capgras syndrome: A novel probe for understanding the neural representation of the identity and familiarity of persons. Proceedings of the Royal Society of London B, 164, 437-444.

Jasper, K. (1963). General psychopathology. Manchester: Manchester University Press.

Jeffrey, R. C. (1987). Alias Smith and Jones: The Testimony of the Senses. Erkenntnis, 26, 391-399.

Johns, L. C. & Os, J. (2001). The continuity of psychotic experience in the general population. Clinical Psychology Review, 21, 11251141.

Joseph, R. (1986). Confabulation and delusional denial: frontal lobe and lateralized influence. Journal of Clinical Psychology, 42, 507-520.

Joyce, J. (Ed.) (2008) The Stanford Encyclopedia of Philosophy (Fall ed.).

Kane, J. M. (1996). Treatment-resistant schizophrenic patients. Journal of Clinical Psychiatry, 57.

Kinderman, P. & Bentall, R. P. (1997). Causal attributions in paranoia and depression: internal, personal, and situational attributions for negative events. Journal of Abnormal Psychology, 106(2), 341-345.

Knowles, R., McCarthy-Jones, S. & Rowse, G. (2011). Grandiose delusions: A review and theoretical integration of cognitive and affective perspectives. Clinical Psychology Review, 31(4), 684-696.

Landa, Y., Silverstein, S. M., Schwartz, F. & Savitz, A. (2006). Group Cognitive Behavioral Therapy for Delusions: Helping Patients Improve Reality Testing. Journal of Contemporary Psychotherapy, 36(1), 9-17.

Levin, J. (Ed.) (2010) The Stanford Encyclopedia of Philosophy (Winter 2009 Edition ed.).

Lewis, D. (1966). An argument for the identity theory. The Journal of Philosophy, 63, 17-25.

Lewis, D. (1970). How to Define Theoretical Terms. The Journal of Philosophy, 67(13), 427-446.

Lewis, D. (1972). Psychophysical and theoretical identifications. Australasian Journal of Philosophy, 50(3), 249-258.

Lucchelli, F. & Spinnler, H. (2007). The case of lost Wilma: a clinical report of Capgras delusion. Neurological Science, 28(4), 188-195.

Manschreck, T. C. & Khan, N. L. (2006). Recent Advances in the Treatment of Delusional Disorders. Canadian Journal of Psychiatry, 51, 114-119.

McKay, R., Langdon, R. & Coltheart, M. (2005). Paranoia, persecutory delusion and attributional biases. Psychiatry Research, 136, 233-245.

Milton, F., Patwa, V. K. & Hafner, R. J. (1978). Confrontation vs. belief modification in persistently deluded patients. British Journal of Medical Psychology, 57, 127-130.

Misner, C. W., Thorne, K. S. & Wheeler, J. A. (1973). Gravitation. San Francisco: W. H. Freeman.

Mullen, R. (2003). Delusions: the continuum versus category debate. Australian and New Zeeland Journal of Psychiatry, 37, 505-511.

Munro, A. & Mok, H. (1995). An overview of treatment in paranoia/delusional disorder. Canadian Journal of Psychiatry, 40, 616-622.

Myin-Germeysa, I., Marcelisa, M., Krabbendama, L., Delespaula, P. & Osa, J. v. (2005). Subtle Fluctuations in Psychotic Phenomena as Functional States of Abnormal Dopamine Reactivity in Individuals at Risk. Biological Psychiatry, 58(2), 105–110.

O’Connor, K., Stip, E., Pélissier, M.-C., Aardema, F., Guay, S., Gaudette, G., et al. (2007). Treating delusional disorder: A comparison of cognitive-behavioural therapy and attention placebo control. The Canadian Journal of Psychiatry, 52(3), 9.

Passer, K. M. & Warnock, J. K. (1991). Pimozide in the Treatment of Capgras's Syndrome. A Case Report. Psychosomatics, 32, 446-448.

Peters, E. (2001). Are delusions on a continuum? The case of religious and delusional beliefs. In I. Clarke (Ed.), Psychosis and Spirituality: Exploring the New Frontier. London: Whurr Publ.

Peters, E. R., Joseph, S. A. & Garety, P. A. (1999). Measurement of delusional ideation in the normal population: introducting the PDI. Schizophrenia Bulletin, 25(553-576).

Quine, W. V. (1951). Two Dogmas of Empiricism. The Philosophical Review, 60(1), 20-43.

Robinson, D. G., Woerner, M. G. & Alvir, J. M. (1999). Predictors of treatment response from a first episode of schizophrenia or schizoaffective disorder. American Journal of Psychiatry, 156(4), 544-549.

Sass, L. (2001). Self and world in schizophrenia: Three classic approaches. Philosophy, Psychiatry & Psychology, 8(4), 251-270.

Soyka, M., Naber, G. & Völker, A. (1991). Prevalence of delusional jealousy in different psychiatric disorders. An analysis of 93 cases. The British Journal of Psychiatry, 158, 549-553.

Stephens, L. & Graham, G. (2004). Reconceiving delusions. International Review of Psychiatry, 16(3), 236-241.

Strauss, J. (1969). Hallucination and delusions as points on continua fonction: rating scale evidence. Archive of General Psychiatry, 21(5), 581-586.

Thornton, T. (2008). Why the idea of framework propositions cannot contribute to an understanding of delusions. Phenomenology and the Cognitive Sciences, 7, 159-175.

Tranel, D., Damasio, H. & Damasio, A. R. (1995). Double dissociation between overt and covert face recognition. Journal of Cognitive Neuroscience, 7, 425-432.

Turkington, D., Kingdon, D. & Weiden, P. (2006). Cognitive Behavioral Therapy for Schizophrenia. American Journal of Psychiatry, 163, 365-373.

Vosgerau, G. & Synofzik, M. (2010). A Cognitive Theory of Thoughts. American Philosophical Quarterly, 47(3), 205-222.

Young, A. W. (2000). Wondrous Strange: The Neuropsychology of Abnormal Beliefs. Mind & Langage, 15, 47-73.


1 Although bizarreness of delusion no longer plays a role in the new diagnostic criteria for schizophrenia, the distinction between bizarre and non-bizarre delusion is still mentioned in the DSM-5. Accordingly, the content of a delusion is deemed bizarre if it is "clearly implausible, not understandable, and not derived from ordinary life experiences’" (APA 2013). 

2 Mental states that are inferentially encapsulated (i.e. fulfill conditions i and ii), are rather illusions than delusions (at least if the illusion is recognized as such, inferences from it are typically blocked). One famous example for such a mental state is the perceptual state a subject is in when looking at the Müller-Lyer-Illusion.

3 Indeed, a lot of stories about Christian Saints, for example, involve periods of deep religious doubts. 

4 There are also examples from other cultures and past times, where people who putatively had hallucinations were treated as oracles or prophets and thus important members of the society. 

5 Acknowledgements: The authors are thankful to Jürgen Zielasek, Alex Tillas, Michael Sollberger, Jean-Marie Droz, Max Seeger. The authors gratefully acknowledge support from the German Research Foundation (DFG, CRC 991.

Authors’ information: 


Dr. Patrice Soom studied philosophy at the University of Lausanne, where he received his doctor degree in 2010. He was Post-Doc at the University of Düsseldorf until October 2014 in the DFG-funded project B06 “Frames in psychiatric classification” within the CRC 991. Since November 2014 he is working for the Swiss State Secretary for Education, Research and Innovation.


Prof. Dr. Gottfried Vosgerau studied philosophy, cognitive science and general linguistics at Freiburg and Bonn. He worked as a doctoral student at the Universities of Tübingen and Bochum, where he received his doctoral degree in 2007 and continued to work as a Post-Doc until he was appointed Junior-Professor at the University Düsseldorf. He became full professor for Philosophy of Mind and Cognition at Düsseldorf in 2014. His main research areas include psychiatric classification, positive symptoms of schizophrenia, mental representation, grounding cognition, and linguistic meaning.



Institut für Philosophie

Heinrich-Heine-Universität Düsseldorf

Universitätsstr. 1

40225 Düsseldorf

E-mail: vosgerau@phil.uni-duesseldorf.de


alttext   Seite drucken


Juckel, G., Hoffmann, K. (Hrsg.)
Ethische Entscheidungs-situationen in Psychiatrie und Psychotherapie

Fahrenberg, J.
Theoretische Psychologie – Eine Systematik der Kontroversen

Schmitt, C.T.
Was ist Klugheit? Wertebezogenes Handeln im Führungskontext

Fahrenberg, J.
Zur Kategorienlehre der Psychologie

von Thienen, J.

W. Blankenburg
Der Verlust der natürlichen Selbst-verständlichkeit. Ein Beitrag zur Psychopathologie symptomarer Schizophrenien

P. Spät
Der Mensch lebt nicht vom Hirn allein. Wie der Geist in den Körper kommt

M. Musalek, M. Poltrum (Hrsg.)
Glut und Asche - Burnout. Neue Aspekte der Diagnostik und Behandlung

E. Baumgartner
Liquid Structures. Auf dem Weg in die narzisstische Gesellschaft

Martin Poltrum
Klinische Philosophie

Kupke, C.; Brückner, B. (Hrsg.)
Das Verschwiden des Sozialen

Stumpf, C.

Heinze, M.; Loch-Falge, J., Offe, S.. (Hrsg.)

Lamiell, J.T.
William Stern (1871-1938): A Brief Introduction to His Life and Works

Musalek, M., Poltrum, M. (Hrsg.)
Ars Medica. Zu einer neuen Ästhetik in der Medizin

Vogeley, K., Fuchs, T., Heinze, M. (Hrsg.)
Psyche zwischen Natur und Kultur

Kupke, Ch.
Der Begriff Zeit in der Psychopathologie

Fuchs, T., Vogeley, K., Heinze, M. (Hrsg.)
Subjektivität und Gehirn

Heinze, M., Fuchs, T., Reischies, F. (Hrsg.)
Willensfreiheit – eine Illusion? Naturalismus und Psychiatrie