Tuesday 29 July 2014

Questions and Reasoning in Schizophrenia and Delusion

Matthew Parrott
Schizophrenia and delusion are typically described as involving some kind of impairment in a subject's ability to reason (e.g., Coltheart et. al., 2011; Coltheart, 2007; Davies and Egan 2013; Garety and Freeman, 1999). Yet, although there is evidence indicating that subjects diagnosed with these psychiatric conditions reason in anomalous ways, in many cases the pattern of reasoning they exhibit looks to be more optimal than the one exhibited by non-psychiatric controls. Most famously, we have known for a number of years that both schizophrenic and delusional subjects 'jump to conclusions' on probabilistic reasoning tasks (Fine, et. al., 2007) but their performance on these tasks is very close to a Bayesian model of ideal rationality. Secondly, there is some recent evidence that suggests schizophrenic subjects may be better at reasoning with conditionals. According to one study, they seem to be less susceptible to believability biases (Owen, et. al., 2007) and according to another they seem to be better at falsifying conditionals with negated antecedents (Mellet, et. al., 2006).
To improve our understanding of these intriguing phenomena, we need to understand precisely how the reasoning capacities of psychiatric subjects differ from those of other individuals. Yet existing theoretical models have failed to give us a clear picture of this. Investigations of reasoning in delusional and schizophrenic subjects have largely been conducted from within a Bayesian framework (Davies and Egan 2013Fine, et. al., 2007). Yet, despite the widespread popularity of this approach in the cognitive sciences, I think it faces certain difficulties, which, for reasons of space, I cannot go into here - although I'm happy to discuss in comments (cf. Parrott, forthcoming).

As an alternative to the Bayesian approach, Philipp Koralus and I are currently drawing on the Erotetic Theory of reasoning (Koralus and Macarenhas, 2013) in or to develop an account of the various patterns of reasoning demonstrated by psychiatric subjects. One of the central insights of the Erotetic framework is to view reasoning as a cognitive process that aims to answer questions, which the theory represents as sets of alternative possibilities.

The Erotetic Theory suggests an intriguing hypothesis: delusional and schizophrenic reasoning differs from that of non-psychiatric subjects in that it relies less on endogenously generated questions. This hypothesis actually predicts different patterns of success and failure: it predicts that subjects who generate fewer questions endogenously would require less empirical evidence before drawing conclusions and it also predicts that this very same tendency to generate fewer questions would actually lead psychiatric subjects to perform better on tasks involving reasoning with conditionals.

Moreover, the Erotetic framework also suggests two exciting clinical implications. First, rather than envisioning psychiatric conditions as based on some kind of deficit in thought or reasoning, the approach suggests that psychiatric patients may simply be less inquisitive in a certain technical sense. Second, the framework suggests that effective clinical interventions may consist in simply helping patients learn to raise the right sorts of questions.


  1. Many people with schizophrenia are not delusional, and many delusional people are not schizophrenic: for example. the literature on Capgras delusion reports at least 14 different aetiologies in different cases of Capgras delusion (affective disorder, AIDS, Alzheimer’s disease, cerebrovascular disease, frontal lobe pathology, head injury, epilepsy, Lewy-body dementia, multi-infarct dementia, multiple sclerosis, pituitary tumour, schizoaffective disorder, schizophrenia, and viral encephalitis).

    So is the Erotetic framework meant to apply to cases of delusion regardless of whether the patient is schizophrenic or not? Or is it only meant to apply to cases of delusion in which the deluded person is, in addition to being delusional, also schizophrenic?

  2. Dear Max,

    Thank you for commenting. I think the answer to your question is both.

    Our aim is to draw on the Erotetic framework to construct a formal model that predicts a range of experimental results. As you know, much of the work on jumping-to-conclusions has studies delusional subjects or schizophrenic subjects who are also delusional. But, as far as I can tell, Mellet, et. al., (2006) and Owen, et. al., (2007) study patients who have been diagnosed with schizophrenia. It would be interesting to know whether these people were also delusional. If so, it may be that the patterns of anomalous reasoning being exhibited are characteristic primarily of delusion and of schizophrenic subjects only when they are also delusional. Do you have thoughts about that?

    Philipp and I are also using the Erotetic framework to model the Capgras delusion. We think this would be a plausible alternative to existing Bayesian models. Since, as you point out, there are many cases of Capgras that do not involve schizophrenia, we are at least claiming that the framework is applicable to those cases.


Comments are moderated.