The psychotic syndrome of schizophrenia, comprising hallucinations and delusions, remains one of the most devastating and costly medical conditions in the US and worldwide. Despite progress in the understanding of its neurobiology, a unifying mechanism remains elusive. Elucidating such mechanism would be a major advance in medicine, as it would provide a scientific explanation to symptoms that have long been considered synonymous with irrationality. This would contribute to de-stigmatization of mental illness. It would also potentially lead to important public health benefits by identifying novel targets that could enable new treatments for a substantial proportion of patients (~1/3) who do not respond to or tolerate current treatments. Previous work has separately studied the substrates of hallucinations (false percepts without corresponding stimuli) and of delusions (false ideas that are maintained with high conviction despite contradictory evidence). However, recent theories of the brain suggest that perception and ideation are part of a unitary process, whereby the brain generates and updates internal predictive models of the external world, that can explain both our idiosyncratic perceptual experiences and how we reach certain conclusions or ideas based on these experiences. Extensions of these ‘Bayesian’ theories have been proposed to explain pathological abnormalities in perception (e.g., hallucinations) and ideation (e.g., delusions) in psychosis as deriving from a common deficit. We suggest that this common deficit is a dopamine-related deficit in updating of beliefs (defined via computational modeling) given excessive reliance of prior beliefs relative to new sensory evidence. We posit that this deficit can occur at two different levels of a hierarchy: a lower level (in sensory cortex) abnormalities in which would cause hallucinations, and a higher level (higher-order prefrontal and parietal cortices supporting inference) abnormalities in which would cause delusions. We propose a converging approach to directly test this novel model it by leveraging cutting-edge tools in computational neuroscience applied to functional neuroimaging (fMRI) and behavioral tools designed to dissect hierarchical belief updating. Specifically, we hypothesize a behavioral and neural double dissociation whereby deficits in the former will be specifically associated with severity of hallucinations while deficits in the latter will be specifically associated with severity of delusions in unmedicated patients with schizophrenia. By anchoring our model on the well-established dopamine dysfunction in psychosis, we will explain how increased dopamine in the midbrain (specifically, in the nigrostriatal pathway, here measured via a novel, high- resolution imaging technique known as neuromelanin-sensitive MRI) alters common neural computations leading to psychotic symptoms (i.e., how it results in deficient belief updating at low and high...