Modelling delusions as temporally-evolving beliefs.

Introduction : Delusions demand an explanation in terms of their neural, psychological, and sociological mechanisms. We must bridge these levels of explanation in order to understand and ultimately treat delusions. To this end, debates continue as to the number of contributing factors, how those factors interact, and their underlying computational mechanisms. Methods : One popular family of models suggests that two separate insults are necessary, a problem with perception and an independent problem with belief. In particular, new work proposes that the belief problem entails a bias against disconfirmatory evidence - yielding the characteristic fixity of delusions. Here, we evaluate that claim, as well as explanations of delusions more broadly. Results : We suggest that such a bias may not explain enough of the variance in belief updating in delusional participants, and, more fundamentally, it might rule out specific accounts of delusions, since, such a bias might prevent them from forming in the first place, under particular assumptions about cognitive architectures. Conclusion : We suggest conceptualising delusions as an evolving uncertainty driven negotiation between beliefs and evidence, in which initial formation is fuelled by unexpected uncertainty, but, once formed, the delusion engenders new expectations about uncertainty that tune down updating but also facilitate the elastic assimilation of contradictory evidence.