Delusional Inference


  • I thank Peter Brugger, Phil Corlett, Zoltan Dienes, Charles Efferson, Tom Griffiths, Robyn Langdon, Frédéric Schneider and Mark Wildon for helpful discussions. I am indebted to Martin Davies and an anonymous reviewer for detailed feedback on a draft of this manuscript. Special thanks to Max Coltheart for valuable comments, advice and encouragement.

Department of Psychology, Royal Holloway, University of London, Egham, Surrey TW20 0EX, UK.


Does the formation of delusions involve abnormal reasoning? According to the prominent ‘two-factor’ theory of delusions (e.g. Coltheart, 2007), the answer is yes. The second factor in this theory is supposed to affect a deluded individual's ability to evaluate candidates for belief. However, most published accounts of the two-factor theory have not said much about the nature of this second factor. In an effort to remedy this shortcoming, Coltheart, Menzies and Sutton (2010) recently put forward a Bayesian account of inference in delusions. I outline some criticisms of this important account, and sketch an alternative account of delusional inference that, I argue, avoids these criticisms. Specifically, I argue that the second factor in delusion formation involves a systematic deviation from Bayesian updating, a deviation that may be characterized as a bias towards ‘explanatory adequacy’. I present a numerical model of this idea and show that my alternative account is broadly consistent with prominent prediction error models of delusion formation (e.g. Corlett, Murray et al., 2007).