You seem to be conflating the original schizophrenic state with the residual after the patients get antipsychotic medication: the latter may be readily amenable to reason; the former, the therapist would breach rapport with the patient, by challenging full delusions.

Medication is part of the standard treatment for schizophrenia--usually, the major part. Drawing conclusions about delusions from the residuals following treatment seems to shield you from what would be obvious had you observed unmedicated patients. Delusions aren't failures of Bayesian rationality: they involve, typically, accepting a few self-evident priors, and these are driven by intense affect.