Monday, 31 March 2014

Explaining Delusions (4)

This is a response to Phil Corlett's contribution to the blog, posted on behalf of Max Coltheart.

Max Coltheart
I think it would be helpful if at this stage the distinction between monothematic and polythematic delusions were introduced. A monothematic delusional condition is one where the patient has only a single delusional belief (or a small set of delusional beliefs related to a single theme). A polythematic delusional condition is one where the patient has many different and unrelated delusional beliefs. This distinction between monothematic and polythematic delusion was offered by Davies, Coltheart, Langdon and Breen (2001) and has been discussed by the philosopher Jennifer Radden in her book (Radden, 2010).

As Coltheart (2013) noted “Well-known cases of polythematic delusion include Daniel Schreber, a judge in the German Supreme Court, who believed that he had the plague, that his brain was softening, that divine forces were preparing him for a sexual union with God, and that this would create a new race of humans who would restore the world to a lost state of blessedness (Bell 2003); more details of his case are provided on pp. 50-52 of Radden’s book. Another example was the Nobel laureate John Nash (who was diagnosed with schizophrenia); among the delusional beliefs he held were that he would become Emperor of Antarctica, that he was the left foot of God on Earth, and that his name was really Johann von Nassau (Capps 2004).”
Forms of monothematic delusion include Capgras delusion (“My wife has been replaced by an impostor”), Cotard delusion (“I am dead”), Fregoli delusion (“People I know are following me around, but in disguise, so I can never recognize them”), mirrored-self misidentification ("The person I see when I look in the mirror isn’t me, but some stranger who looks like me”, the delusion of alien control (“Other people can cause my limbs to move without my intending them to move”), anosognosia (“My left arm is not paralysed”, when the patient’s armactually is paralysed) and somatoparaphrenia (“This isn’t my arm, it is my aunt’s”; here the patient is referring to her own arm).

In various papers on our two-factor account of delusional belief, we have raised the question of the scope of our account. We commit ourselves to the position that ouraccount is meant to apply to every form of monothematic delusion. This requires us to propose what Factor 1 is for each of the delusions mentioned above. Since Factor 1 is what determines the content of a monothematic delusion, and since the content of a monothematic delusion differs from delusion to delusion, Factor 1 will differ from delusion to delusion; in contrast, Factor 2 (impaired belief evaluation) is the same for all forms of monothematic delusion. We have taken the position that whether the two-factor account of delusional belief also offers a plausible account of polythematic delusion remains an open question, difficult to resolve at present because the kinds of detailed single-case studies that have been done with patients with monothematic delusions to yield data to which the two-factor theory canbe applied have simply not been done with cases ofpolythematic delusion.

Phil, is your work on delusion intended to apply both topolythematic delusions and also to the monothematic delusions I mention above?

Now to respond to your questions:

Q: “If prediction error is intact in people with delusions . . .”.

A: I did not understand what’s meant here by prediction error being intact. But in any case in my first response I pointed out that in our 2-factor account the occurrence ofprediction errors are crucial for the genesis of delusion. The role of Factor 1 in delusion is, precisely, that it causes prediction error (e.g. in Capgras delusion, the prediction that when I see my wife’s face there will be a response by my autonomic nervous system is erroneous). What remains intact here is the system that detects prediction errors and also the normal response to the detection of such errors, this normal response being what Phil terms “new learning” and what we term “belief system updating”. There is of course, in a person with a monothematic delusion, a second cognitive system that is not intact, the belief evaluation system: impairment of that system is our Factor 2.

Q: According to your answer, prediction error is neither factor 1 nor factor 2 but the interaction between them.

A: No, that isn’t what I said. As explained in my response to the question above, prediction error is caused by Factor 1.That is not the same thing as saying that prediction error is factor 1. If I have a stroke one night that disconnects my face recognition system from my autonomic nervous system, that is factor 1 re the Capgras delusion. But a prediction error will not occur at that time. It won’t occur until I next see the face of someone very close to me: my prediction of an autonomic response will now be erroneous here. In order for future encounters with my wife not to generate prediction error, I have to adopt the belief that the person with this face is not my wife, but some stranger. Once I have that belief, I will (correctly) predict no autonomic response when I see this face, thus achieving the goal that all of us always have: minimization of prediction error (see my comment on Jakob Hohwy’s Imperfect Cognitions post about his book “The Predictive Mind”).

Q: According to this explanation, belief formation is intact in your model – the imposter hypothesis is appropriate given the perceptual experience, and that candidate is adopted and then never updated? Is that correct?

A: Yes, that’s correct – that’s what the 2-factor account says about Capgras delusion. With the qualification that it seems that tactful CBT-style conversation about the delusional belief might sometimes lead to temporary doubts about the delusional belief: see Coltheart (2007, pp. 1053-1054) for a verbatim transcription of one such conversation with a Capgras patient.

Q: Do the patients with perceptual disruption but no delusions (critical to 2-factor theory) have normal genesis of candidate explanations also? So the imposter explanation crosses their mind but they discount it?

A: Yes, that’s the simplest position we can take, so we take it given that there is nothing to rule it out. These patients are able to discount the explanation because of the intactness of their belief evaluation systems (i.e. the absence of Factor 2). So we don’t need the third factor of which you spoke.


Bell, D. 2003: Paranoia. Cambridge, UK: Icon

Capps D. 2004. John Nash’s delusional decade: a case of paranoid schizophrenia. Pastoral Psychology, 52, 193–218

Coltheart, M. (2013). On the distinction between monothematic and polythematic delusions: Some comments on On Delusion (Radden, 2011). Mind & Language, 28, 103-112.

Davies, M., Coltheart, M, Langdon, R. and Breen, N. 2001: Monothematic delusions: Toward a two-factor account. Philosophy, Psychiatry & Psychology, 8, 133-158.

Radden, J. (2011): On Delusion. London: Routledge.


  1. Hi Max and Phil,

    I am really enjoying the conversation! There are certainly some important differences between the two. But, I am more interested in the similarities and potential synthesis between the views.

    Think for example about Capgras delusion. Both Max (Coltheart et al. 2010) and Phil (Corlett et al. 2010) accept Ellis and Young's famous hypothesis about disconnection between autonomic nervous system and face recognition system. The disconnection produces the first factor (Max) or affective prediction-error (Phil). Both think that Capgras delusion is a reasonable response to the abnormal input due to the disconnection (although people like McKay (2012) disagree). Max wrote, "if the subject with Capgras delusion unconsciously reasons in this way, he has up to this point committed not mistake of rationality on the Bayesian model" (Coltheart et al. 2010, 278). Phil wrote; "we argue that phenomenology of the percepts are such that bizarre beliefs are inevitable; surprising experiences demand surprising explanations" (Corlett et al. 2010, 360).

    A difference arises when it comes to the sufficiency of abnormal input for delusional beliefs. Max and other two-factor theorists think that the abnormal input is not sufficient for Capgras delusion because the patients with the damage in ventromedial prefrontal cortex share the same input but do not develop Capgras delusion. Phil isn't convinced by this. He argues, for example, that some patients with damage in ventromedial and lateral prefrontal cortex show delusion-like confabulations.

    This disagreement is certainly a serious one. But, my impression is that that say similar things on this too after all.

    Max thinks that adopting Capgras hypotheses is perfectly Bayesian-rational given the abnormal data. In the post-adoption stage, however, the subjects with Capgras delusion irrationally maintain the hypothesis due to the failure of considering new conflicting evidence. This failure explains the difference between the subjects with Capgras delusion and the patients with ventromedial prefrontal damage. The latter people can successfully remove Capgras hypothesis in light of conflicting evidence, but he former people can't.

    Phil says something similar. He seems to accept after all that, in a sense, the aberrant prediction-error signal is not sufficient for a full account of delusion formation, because it doesn't tell much about why delusions are maintained in post-adoption stage. He admits that the simple prediction-error model “accounts for why delusions emerge but not for why they persist” (Corlett et al. 2013, 2). And, he presents an additional hypothesis about maintenance of delusion according to which delusions are maintained through the process of memory reconsolidation. (For this reason, there is a sense in which Phil's view is another version of "two-factor" account although he has tried to distance himself from two-factor theorists.)

    So, after all, both Max and Phil agree that a full account of Capgras delusion involves something more than first factor or affective prediction error. In particular, it needs to say something about the way in which delusions are maintained.

    Do I understand your views correctly?


    Coltheart, M., Menzies, P., & Sutton, J. (2010). Abductive inference and delusional belief. Cognitive Neuropsychiatry, 15(1-3), 261-287.

    Corlett, P., Taylor, J., Wang, X., Fletcher, P., & Krystal, J. (2010). Toward a neurobiology of delusions. Progress in Neurobiology, 92(3), 345-369.

    Corlett, P. R., Cambridge, V., Gardner, J. M., Piggot, J. S., Turner, D. C., Everitt, J. C., et al. (2013). Ketamine Effects on Memory Reconsolidation Favor a Learning Model of Delusions. PloS one, 8(6).

    McKay, R. (2012). Delusional inference. Mind & Language, 27(3), 330-355.

  2. Thanks Kengo - I think you are right - I believe our theories have much in common and that perhaps Max and I are addressing different levels of explanation. I have written another post on this that should come out tomorrow on the blog.

    I do have to disagree with you in regards our explanation of delusion maintenance.

    We showed in the paper that you cite that aberrant prediction errors also engage memory reconsolidation. So this is a prediction error driven explanation of delusion fixity - maintenance is not driven by a second separate factor, but the same factor that caused delusion formation.

    Happy to elaborate on this.

    Best wishes and thanks for the comment,



Comments are moderated.