PRISMA 2020 checklist.

<div><p>Cognitive biases have been studied in relation to schizophrenia and psychosis for over 50 years. Yet, the quality of the evidence linking cognitive biases and psychosis is not entirely clear. This umbrella-review examines the quality of the evidence and summarizes the effect sizes of the reasoning and interpretation cognitive biases studied in relation to psychotic characteristics (psychotic disorders, psychotic symptoms, psychotic-like experiences or psychosis risk). It also examines the evidence and the effects of psychological interventions for psychosis on cognitive biases. A systematic review of the literature was performed using the PRISMA guidelines and the GRADE system for 128 analyses extracted from 16 meta-analyses. Moderate to high-quality evidence with medium to large effect sizes were found for the following interpretation biases: externalization of cognitive events and self-serving bias, when people with psychotic symptoms were compared to control conditions. Regarding reasoning biases, moderate to high quality evidence with medium to large effect sizes were found for belief inflexibility when linked to delusion conviction and global severity in people with active delusions, although measures from the MADS, overlapping with symptoms, may have inflated effect sizes. Moderate quality evidence with medium to large effect sizes were found for jumping to conclusion biases when clinical samples with psychosis were compared to controls, when using data-gathering tasks. Other cognitive biases are not supported by quality evidence (e.g., personalizing bias, belief about disconfirmatory evidence), and certain measures (i.e., IPSAQ and ASQ) systematically found no effect or small effects. Psychological interventions (e.g., MCT) showed small effect sizes on cognitive biases, with moderate-high-quality evidence. This umbrella review brings a critical regard on the reasoning and interpretation biases and psychotic symptoms literature—although most biases linked to psychotic symptoms are supported by meta-analyses in some way, some have only demonstrated support with a specific population group (e.g., aberrant salience and hostility attribution in healthy individuals with psychotic-like experiences), whereas other biases are currently insufficiently supported by quality evidence. Future quality studies, particularly with clinical populations with psychotic symptoms, are still warranted to ascertain the psychosis-cognitive bias link for specific biases.</p></div>