(Text Version)

  1. introduction: the problem of defining delusions
  2. delusions as psychotic beliefs; the term "delusion"
  3. overview of issues


As surprising and embarrassing as it is, we can't define delusions, one of the most important symptoms of psychopathology. The traditional definition -- that delusions are false beliefs, unshared by one's culture, arising from a faulty process, held with extraordinary certainty and unshakeable even in the face of contradictory evidence -- cannot distinguish delusions from many other abnormal or unusual beliefs, such as 'overvalued ideas' (for example, the anorexic's belief that she is overweight), 'confabulations' (such as the belief of a stroke patient that his paralyzed limb belongs to someone else), or idiosyncratic religious, political or scientific beliefs (for example, the belief, when originally conceived, that the Holocaust never happened). As Cutting (1997) wrote of "the extraordinary problems involved" in the definition of delusions, "all extant definitions are unsatisfactory in some respect " (p. 194).

On the other hand, most clinicians believe they can recognize a delusion when they see one. For example, most clinicians, if told that John Nash believed extraterrestrials were communicating with him through encrypted messages in The New York Times (Nasar, 1998), would immediately think delusional psychosis. As McKenna (1984) wrote, delusions have an "indefinable, but easily recognizable alien quality" (p. 583). Delusions are like pornography, you can't define them, but you know them when you see them.

What prevents us from being able to articulate our tacit knowledge about delusions? As Cutting (1997) went on to write about defining delusions, "the very intractability of the problem, as it has been tackled over the last 100 years, should make one wonder whether the wrong questions are being asked" (p. 194).

What questions would clarify the problem of defining delusions?

For starters, we could make explicit the subtext of the question "what is a delusion?" Arguably, the search for the defining characteristics of delusions is, de facto, the search for expressions or markers of psychosis (an equally elusive concept).



Delusions are delusions by virtue of their psychotic nature. The characterization of delusions as psychotic beliefs is implicit in nosology, terminology and research agendas.

Delusions are considered evidence of psychosis in even the most restrictive definitions of psychosis (DSM-IV, see DEFINITIONS). Although not necessary (required) for a diagnosis of psychosis, they are sufficient (evidence enough) for a diagnosis of psychosis, and are the sole positive criterion for one psychotic disorder (Delusional Disorder). Delusion-like phenomena in nonpsychotic diagnoses are given other names, like "overvalued ideas", obsessions, or "confabulations," in order to reserve "delusions" for psychotic disorders. Research aims at drawing distinctions between delusions and these other phenomena on the basis of degree of conviction with which the false belief is held (cf. Kozak & Foa, 1994), or on other formal aspects (Jones & Watson, 1997). The question "how should we define delusions" translates operationally as "what characteristics, unique in patients we diagnose psychotic, set apart beliefs we can call delusions?"

Making this implicit question explicit helps clarify the question of defining delusions, but also highlights a problem with the concept "psychosis." If delusions are considered sufficient for a definition of psychosis, and we can't define delusions, then we haven't adequately defined psychosis. At this point, we are in the untenable position of using psychotic diagnosis as a criterion for terming beliefs delusions (or worse, using psychotic diagnosis to diagnose beliefs as delusions), and delusions to diagnose psychosis, a circularity in which what gets obscured is the fact that neither term is precisely defined.

So we need to articulate the crux of what's psychotic about a delusion. But in clarifying what makes a delusion a psychotic belief, we may also specify something about the vague "gross impairment in reality testing" that characterizes psychosis and may be associated as well with other symptoms of psychosis (e.g. hallucinations and thought disorder). The crux of what is psychotic about a delusion could be a more general manifestation of psychosis, as opposed to a characteristic specific to delusions. For example, one property of many delusions -- the loss of perspective on how others view the delusion -- could be but one example of a psychotic loss of perspective on how others would view one's unusual verbal productions, be they references to delusions or hallucinations, or formally thought disordered communications (Harrow & Miller, 1980). Thus it's possible that the psychotic aspect of a delusion is a more general manifestation of psychosis that is brought into relief in reference to the delusion.

Another possibility is that delusions have no uniquely distinguishing properties, that they are indefinable in the way that Wittgenstein found the word "games" to be indefinable (M. N. Eagle, personal communication, January, 2000), and that they are being diagnosed in practice by their concurrence with gross or subtle indicators of disorders consistent with psychosis, so that the same belief would be deemed eccentric in one person and delusional in another. In that case, the term delusion would be meaningless, and the task would then be to articulate the extrinsic psychotic cognition or other trait to which clinicians are responding when they call a belief a delusion.

Before reviewing the efforts to define delusions, there's the problem of how to use the term "delusion" in this discussion. Because something awfully like delusions is found in 'nonpsychotic' disorders, one possibility would be to use the more specific term "psychotic delusions," which is less ambiguous, acknowledges the definitional problems and leaves open the possibility of a more generic use for the word "delusion." However, because I'm arguing that the subtext of the attempt to define delusions is to articulate their psychotic quality, the word "delusion" will be used to indicate beliefs that have some property specific to psychosis. The point is to compare (psychotic) delusions with similar phenomena, without worrying for now what we call these things, in the hope that a uniquely psychotic property will emerge.



[Considerations of space prohibit a comprehensive review. See BACKGROUND READING LIST.]

The attempt to define delusions has been a search for differences from other phenomena, repeatedly foiled by counterexamples. The traditional criteria for defining delusions (see DEFINITIONS) have been challenged not only by examples of delusions that don't meet the criteria, but by evidence that many normal beliefs are, for example, equally false, unshared by others, unshakeable, etc.

For example, the criterion of fixity (unshakeability) has been challenged not only by evidence that some delusions are amenable to change through cognitive therapy (review, Bouchard, Vallieres, Roy & Maziade, 1996), but also by the fixity of, say, religious beliefs. Indeed, fixity may be a characteristic of belief per se, by definition -- a necessary but not sufficient defining characteristic of delusions -- and other properties of delusions characterize many farfetched beliefs (eg. implausible, preoccupying). There is currently no universally accepted classical definition --necessary and sufficient characteristics-- of delusions.

("Necessary" properties are those that would have to be present for a belief to be regarded as a delusion, but that might not be evidence enough to regard a belief as a delusion. "Sufficient" properties are those that by themselves are evidence enough to regard a belief as a delusion. A particular defining property could be necessary, sufficient, or necessary and sufficient.)

Because of the problem of coming up with a classical definition, an alternative approach has been to enumerate characteristics, none of which is necessary or sufficient, but each of which enhances the suggestion of delusionality -- a kind of prototype approach. Oltmanns (1988) lists 7 such frequently mentioned characteristics. To paraphrase him, the belief is: totally incredible to others, unshared, unshakeable, preoccupying, personally referent (rather than religious, scientific or political), distressing or interfering with functioning, unresisted (in contrast with an obsession). However, the value of a prototype approach is mitigated by the number of characteristics delusions share with other beliefs, unless one were to weigh certain traits more heavily than others (in which case we would be closer to a classical definition, see Wakefield, 1999).

Adding to the difficulty of defining delusions is the finding that properties of delusions like conviction or preoccupation may exist along continua rather than being present in an all-or-none manner (Kendler, Glazer & Morgenstern, 1983; Sacks, Carpenter & Strauss, 1974; Strauss, 1969). Further, the various dimensions tend not to be correlated in intensity (Kendler et. al., 1983), suggesting that delusionality is not a unified entity but has several independent components. However, neither of these qualities necessarily rules out a classical definition. The lack of precise boundaries does not preclude the usefulness of a classical criterion -- for example, in defining "bachelor" as "an unmarried man," the fuzziness of the boundary between boy and man does not negate the usefulness of the requirement that a bachelor be a man (Wakefield, 1999). And the lack of correlation among dimensions could represent their differential expression in the differently wired brains of individuals.

Because it's been so hard to pinpoint the descriptive characteristics of a delusion, it has sometimes been suggested that a delusion be defined by the etiological process giving rise to the belief above and beyond its descriptive characteristics. Jaspers thought that even a true belief could be considered a delusion if it arose from a psychotic process. An example would be an accidentally true belief that one's partner was unfaithful, so that one is right for the wrong reasons (Jaspers 1963, p. 106). While most definitions today would require falsity or implausibility, the etiological criterion is still alive in the current requirement that the belief develop by way of an inferential error. This, along with the unaccepted by one's culture standard, eliminates second hand, culturally transmitted implausible beliefs (e.g. religious beliefs such as the resurrection of Christ), but inferential error is still too broad and nonspecific to rule out many other normal beliefs. (Moreover, the inferential error standard may be inconsistently applied in DSM in the case of Shared Psychotic Disorder, in which a delusion is acquired by hearsay.)

There have been many attempts to pin down the nature of a faulty or pathological inferential process specific to delusions. The term inferential error is generally understood to mean errors of logic or of conscious rational thought, but there has been little research support for the existence of a characteristic rational or logical error (e.g. syllogistic error) that produces delusions (Maher & Spitzer, 1993). Moreover, in the nowadays of cognitive science, when it is understood that much (if not most) information is processed unconsciously, it would seem archaically restrictive to look for an inferential error when it is at least as likely that unconscious or intuitive error could produce the distortions of delusions. Indeed, predating the current Zeitgeist, Jaspers wrote that real delusions were "unmediated by thought" (Walker, 1991) And some information processing errors attributed to delusional patients, such as a "jump to conclusions" style (see Butler & Braff, 1991; George & Neufeld, 1985), which has been characterized as a rational error, could as easily be understood as an unconscious induction error (Garety, 1992, calls it a "preconscious abnormality"). Either way, we're dealing with "cognitive error," that is, an error that now falls within the confines of a broadly defined concept of "cognition" (i.e., "pertaining to the encoding, transformation, storage, and use of information for the purpose of regulating behavior."- George & Neufeld, 1985, p.264).

In any case, it is not clear that "jumping to conclusions" would not characterize the process by which some beliefs of normals develop -- such as a certainty before O.J. Simpson's trial that he was innocent (or guilty). Indeed, some have suggested that delusions are developed by many of the same processes by which normal false beliefs develop, e.g. judgmental shortcuts such as heuristics and response biases, on which psychotics overly rely because of other, albeit abnormal, information processing disorders (Chapman & Chapman, 1988; Kihlstrom & Hoyt, 1988). The most influential current theory has it that delusions are developed like every other belief that explains a distressingly puzzling experience, the only difference being that psychotics may have more to explain, such as false experiences of the special "significance" of events (Maher, 1974, 1988; Maher & Ross, 1984; Maher & Spitzer, 1993). This suggests that such prodromal "delusional mood" experiences may be the only identifiable aspects of the development of a delusion that are uniquely psychotic (cf. Roberts, 1992) (and, incidentally, more consistent with unconscious induction error than with rational "inferential" error).

While most attempts at defining delusions have at least implicitly sought to articulate expressions of a psychotic process, Spitzer (1990), in a twist, suggested that some beliefs that express a psychotic process should be excluded from being called delusions. He suggested that statements about subjective experiences, such as "thought insertion" (e.g. "My thoughts are not thought by me") should be considered not delusions but rather "disorders of experience." Such common "delusions" can hardly be considered idiosyncratic, he argued, nor, as declarations of internal experience, can they be considered "wrong." He would limit delusions to statements about the external world (e.g. "The CIA controls my thoughts").

However, despite the epistemological difference between "my thoughts are not thought by me" and "the CIA controls my thoughts," both statements would share arguably psychotic characteristics if uttered with loss of perspective on how the statement would be received by others or with lack of "insight into the pathological nature" of the experience (which distinguishes psychosis from nonpsychosis in the case of hallucinations).

For purposes of the present discussion, which seeks to define delusions by articulating their psychotic quality, neither of Spitzer's examples need be ruled out as psychotic delusions, because they both derive from psychotic disorders of experience. Indeed the purpose of this Web site could be said to be to define one expression (i.e. delusion) of what we decide should be called psychotic disorders of experience.

In pursuing this purpose, there is no logical limit to the kinds of DELUSIONAL PROPERTIES that potentially could express psychotic disorders of experience -- be they descriptive of a delusion (e.g. implausible), etiological (e.g. belief is preceded by experiences of "significance"), descriptive of the relationship to the delusion (e.g. preoccupying), or epistemological (e.g. experienced as intuition).

Of course, there is no external criterion --- e.g. pathophysiology, course, prognosis, treatment response -- that can be used to validate a particular definition of delusions as a psychotic expression (how could there be, when there are no precise defining criteria for psychosis with which to correlate a validating criterion?). As for the practical strategy of validating delusional psychosis by treatment response, the question would be "which treatment?," since delusions, as currently defined, appear to be a final common pathway for diverse disorders (McAllister, 1992) that respond to different treatments depending on their etiology (Opler, Klahr, & Ramirez, 1995). Nonetheless, there are some data about relationships between delusional properties and treatment response that, along with the intuitions of clinical practice, may offer clues to what are essential or nonessential distinctions among delusion-like beliefs.

In particular, the degree of conviction with which a belief is held may not be as important as it's been thought to be, as a basis for differentiating beliefs. Degree of conviction is the sole basis in DSM-IV for distinguishing overvalued ideas and obsessions from their delusional versions, and thus the basis for diagnosing the psychotic (Delusional Disorder) versions of non-psychotic diagnoses such as Body Dysmorphic Disorder and Obsessive-Compulsive Disorder (DSM IV, pp. 300-301). Yet Phillips, Dwight & McElroy (1998) found that delusion-level conviction does not rule out a good response to serotonin reuptake inhibitors (SRI's) for delusional Body Dysmorphic beliefs, and preliminary results suggest that delusion-level conviction may not rule out a good response to SRI's in other conditions, such as Obsessive-Compulsive Disorder (Eisen, Rasmussen, Phillips, Lydiard, & Pigott, 1995).

Although some have interpreted these data as suggesting that SRI's have neuroleptic effects (Pies, 1997), it's also possible that there is some meaningful pathophysiological distinction between "majorly psychotic" delusions and certain content-specific delusions, especially as this finding echoes earlier reports that Delusional Disorders, unlike Schizophrenia, do not respond well to neuroleptics (Opler et al., 1995). It also echoes research indicating that Delusional Disorder does not run in families with Schizophrenia or Affective Disorders (Kendler, 1984). In other words, there may be some basis for the longstanding shared implicit judgment that there is some important distinction between unrealistic beliefs we've called overvalued ideas (for example, that one is ugly or overweight) and those we've viewed as major psychotic delusions (for example, that one's thoughts are broadcasting or that one is Jesus Christ), a difference marked not by degree of conviction but by thematic content, and possibly by other properties as well.

If we could specify the properties that distinguish overvalued ideas or Delusional Disorder delusions from "majorly psychotic" delusions, we would still be left with the dilemma of how to express that distinction in terminology. Should only "majorly psychotic" delusions be considered delusions, and others be called overvalued ideas or obsessions? Or are the similarities between "delusions," overvalued ideas, confabulations, and even some "normal" beliefs compelling enough to justify making the concept "delusion more inclusive and generic, so that we speak of psychotic vs. nonpsychotic delusions, or some other subgroups? Or is "delusion" indefinable, a phenomenon that will always be diagnosed by some other aspect of the clinical picture, so that we should stop trying to define it, and focus instead on trying to articulate that other, psychotic thing to which people are responding when they call someone's belief a delusion? Or what?

There have been efforts to systematize intuitions about disorder per se, as well as proposals for how disorder should be defined (Wakefield, 1999). In looking to define delusions, this Web site seeks to systematize lay and clinical intuitions about delusions in order to help answer the question "what should be deemed psychotic?"


© 2000 Alice Krakauer, Ph.D.
Posted 10/6/00