Tuesday, 1 December 2015

Bayesian Accounts and Black Swans

In this post Ryan McKay (picture above), summarises his recent paper 'Bayesian Accounts and Black Swans: Questioning the Erotetic Theory of Delusional Thinking'.

Matthew Parrott and Philipp Koralus (hereafter P&K) offer a fresh take on 'imperfect cognitions'. In their recent post they outline how their 'erotetic theory' can account for certain instances of fallible human reasoning. They illustrate this with an example about a fridge containing either beer or wine and cheese (I confess that I fell for the fallacy here; I presume my critical faculties were disarmed by my stomach).

My purpose in this brief post is not to contest their analysis of such examples, but to summarise my evaluation of their erotetic approach to delusional thinking, raising my own questions about their theory in the process.

The Core Claim

P&K’s core claim is that deluded individuals are less inquisitive than healthy individuals; in particular, deluded individuals are selectively deficient in raising endogenous questions, while having no problem raising or answering exogenous questions (which include 'default questions in response to external stimuli' as well as questions posed by others). However, without any rigorous way of distinguishing endogenous questions from exogenous questions, the hypothesis that deluded individuals are impaired in raising the former seems hard to falsify – any question that a deluded individual shows themselves capable of asking could be rationalised as 'externally stimulated', and thereby exogenous, after the fact. Meanwhile, the claim that deluded individuals 'would have no problem taking on board and answering questions that are put to [them] by someone else' (Parrott and Koralus 2015: 400) is already contradicted by available evidence, as some deluded individuals are completely impervious to external questioning (e.g., see Breen, Caine, and Coltheart 2002).
  • Q1) Given that P&K suggest that some questions a person asks are 'externally stimulated', and thereby exogenous, how can we reliably distinguish endogenous questions from exogenous questions?
  • Q2) If deluded individuals are selectively deficient in raising their own questions, why are they unable to fully utilize and retain questions that others raise?

Capgras Delusion

According to P&K, we represent other people we know as bundles of features. P&K claim that Capgras delusion arises when we encounter someone emotionally close to us but fail to represent them as being emotionally close to us (in P&K’s terminology, we fail to represent them as having the 'C-feature'). It stands to reason that if we represent a particular person we know as having the “C-feature”, and we encounter someone who we represent as not having this feature, we may not classify the person we have encountered as the person we know. But P&K’s analysis gives no reason to think we would classify them as a stranger posing as the person we know.
  • Q3) If I encounter a person lacking the C-feature, what could lead me to speculate that the person is an impostor?
  • Q4) If the absence of the C-feature prompts impostor speculations when I encounter someone I know, why would the absence of other relevant features (e.g., a moustache that has been shaved off) not do the same?

Conditional Reasoning in Schizophrenia

Mellet et al. (2006) found that patients with schizophrenia outperformed healthy control participants on a particular reasoning task, which involved falsifying conditional rules with a negation in the antecedent (e.g., If there is not a red square on the left, then there is a yellow circle on the right). P&K suggest that healthy participants adopt a 'suppositional reasoning strategy' to solve such problems. It is true that this strategy, which involves finding a state of affairs that confirms the affirmative clause and contradicts the negative clause of the rule in question, yields incorrect answers when applied to conditional rules with a negation in the antecedent. But P&K do not explain why this strategy should require a capacity for endogenous-question-raising. Indeed, what P&K say elsewhere in their paper suggests just the opposite. So it is unclear how the erotetic theory offers any explanation of these results. Whether this is a significant problem for them depends on the scope they intend their theory to have. If their theory is intended as an account of delusional cognition there is no major problem, because delusions (a symptom) and schizophrenia (the diagnostic syndrome investigated by Mellet et al., 2006) are not the same thing.
  • Q5) Why would the suppositional reasoning strategy require a capacity for endogenous-question-raising, whereas the correct logical strategy wouldn’t?
  • Q6) Is P&K’s theory a theory of delusions, of schizophrenia, or of both delusions and schizophrenia?

Jumping to Conclusions

That deluded individuals 'jump to conclusions' on the well known 'beads task' is one of the most important and influential findings in the literature on delusions. The task is to determine from which of two jars a sequence of coloured beads is being drawn (e.g., Jar A, containing 85 red beads and 15 blue, or Jar B containing 85 blue beads and 15 red). A robust finding is that deluded individuals require fewer bead draws before making this decision. According to P&K, this effect occurs because deluded individuals (who have a reduced tendency to endogenously raise questions) ask themselves questions that explicitly represent fewer bead sequences, the minimal such question being 'do I have Jar A, which has red on its first draw, or do I have Jar B, which has blue on its first draw?'
  • Q7) What is the equivalence between raising fewer questions and raising “lesser” questions (questions that represent fewer alternatives)? 
  • Q8) Does the suggestion that deluded individuals are less inquisitive than healthy individuals add much to what we already know (which is that deluded individuals seek less evidence than healthy individuals when forming beliefs and making decisions)?
To conclude, I think the erotetic theory is intriguing, and I like the suggestion that deluded individuals are less inquisitive than healthy individuals. But I’m not sure this suggestion does much more than re-describe existing findings with fresh terminology. I’d like to know what novel, falsifiable predictions the theory makes about delusions. I think answers to my above questions would enable a fuller assessment of this interesting theory’s original contribution to understanding delusional cognition.

No comments:

Post a Comment

Comments are moderated.