Psychiatry Classics #3: A theory of abberant predictability estimates in schizophrenia
This classic 2009 review paper by Fletcher and Frith, currently cited 456 times, attempts to explain the two major positive symptoms of schizophrenia, hallucinations and delusions, as due to a common high-level cognitive mechanism.
But first, they consider one of the simplest hypotheses: might people with schizophrenia have disordered reasoning in general? The authors reject this hypothesis because patients with schizophrenia do not have problems with logic in general; at least in the n = 32 study they cited, control as opposed to deluded people were actually slightly more likely to fall for fallacies in logical questions.
Instead,Fletcher and Frith's hypothesis relates to a failure to correctly perform conditional reasoning.
As they point out, stimuli that do not challenge a belief are usually ignored, which is often necessary in order to deal with the large number of stimuli in one's environment.
Intriguingly, in animals, this prediction error-dependent learning is highly dependent on the dopaminergic system. And there is a wealth of evidence showing that the dopamine system is implicated in schizophrenia, including the ventral striatum.
Berns et al., in a 2001 fMRI study, showed that the bilateral nucleus accumbens (a part of the ventral striatum) is involved in responses to predictable stimuli
Their hypothesis, then, is that there is a quantitative divergence in the prediction error-dependent learning for every day stimuli in schizophrenia.
This leads normal stimuli to feel unduly important and thus makes properly attending to one's environment challenging. This can explain delusions, because people must attempt to explain why those stimuli feel so surprising.
This could also make internal thoughts (which are perceived stimuli just like any other) appear more likely to be under external rather than internal control, because they are imbued with a particular sense of surprisingness. This, of course, can explain hallucinations.
Of course, this is just a model and is probably flawed in various ways, but it's a pretty thorough one and worthy of consideration.
Reference
Fletcher PC, Frith CD. Perceiving is believing: a Bayesian approach to explaining the positive symptoms of schizophrenia. Nat Rev Neurosci. 2009;10(1):48-58.