Using virtual reality to explore differences in memory biases and cognitive insight in people with psychosis and healthy controls

•A scenic virtual reality paradigm was used to elicit false memories.•Feedback on errors was used to correct overconfidence in false memories.•Accuracy and confidence of psychosis patients were compared to healthy controls.•Patients with psychosis showed impaired social cognition.•Patients with psychosis showed evidence for impaired metacognition. Memory biases (e.g. overconfidence in false memories) are implicated in the pathogenesis of delusions. Virtual reality (VR) may provide an opportunity to observe such biases and improve cognitive insight in patients with psychosis via corrective feedback. Thirty-nine patients with psychosis and 20 healthy controls explored VR environments designed to elicit false memories and subsequently had to recollect items and faces. We used a randomised-controlled design where half of the sample received performance feedback on the recollection task in order to correct overconfidence. Changes in cognitive insight were measured using the Beck Cognitive Insight Scale. Regarding accuracy, patients performed worse on the social task (recollection of faces) only. Patients displayed overconfidence in false memories for emotions and gave more high-confident responses compared to healthy controls on the social task. Feedback did not improve cognitive insight. Patients rated their cognitive insight higher than healthy controls. Future research should address problems with subjective measurements for cognitive insight. To conclude, patients with psychosis showed impaired social cognition and there was evidence for impaired metacognition, as patients reported higher cognitive insight despite comparable or worse performance as well as overconfidence relative to controls.