Delusions, or false beliefs that are held with high conviction, are signature symptoms of several highly distressing psychotic disorders. Because there are few meaningful biomarkers and limited treatment options for psychotic disorders, understanding this specific symptom may offer a useful transdiagnostic target to move the field forward. As psychosis is commonly understood as a break with reality, it stands that learning more about the neurobiology of the systems that underlie perception of reality, may provide greater insight into the pathology of psychotic delusions. A promising but sparse line of research on delusions is centered on prediction error (PE), neural signals that register the difference between our expectations and the outcomes that actually occur. Problems with this process may account for several cognitive functions implicated in delusion formation and maintenance such as impaired salience detection, increased uncertainty, and reduced precision of belief. While prediction error abnormality is posited as a key mechanism of delusional beliefs, the empirical evidence supporting this is limited. In this project, resting state and task-based fMRI data obtained from patients diagnosed with primary psychotic disorders along with healthy controls were used to investigate the network connectivity of brain regions associated with prediction error and delusions. Though abnormalities in the prediction error activation and functional connectivity were found in psychosis, these changes were not transdiagnostically associated with delusions. However, through use of novel computational modeling approaches, effective connectivity alterations within the intrinsic connectivity of prediction error circuits were shown to be associated with both psychosis and delusions. These results offer new insight into the pathophysiology of delusions and may help to guide more directed investigation of neural circuitry in future biological and clinical studies.