Bayes for Schizophrenics: Reasoning in Delusional Disorders — LessWrong