BackgroundMetamemory describes the monitoring and knowledge about one's memory capabilities. Patients with schizophrenia have been found to be less able in differentiating between correct and false answers (smaller confidence gap) when asked to provide retrospective confidence ratings in previous studies. Furthermore, higher proportions of very-high-confident but false responses have been found in this patient group (high knowledge corruption). Whether and how these biases contribute to the early pathogenesis of psychosis is yet unclear. This study thus aimed at investigating metamemory function in the early course of psychosis.MethodPatients in an at-risk mental state for psychosis (ARMS, n = 34), patients with a first episode of psychosis (FEP, n = 21) and healthy controls (HCs, n = 38) were compared on a verbal recognition task combined with retrospective confidence-level ratings.ResultsFEP patients showed the smallest confidence gap, followed by ARMS patients, followed by HCs. All groups differed significantly from each other. Regarding knowledge corruption, FEP patients differed significantly from HCs, whereas a statistical trend was revealed in comparison of ARMS and FEP groups. Correlations were revealed between metamemory, measures of positive symptoms and working memory performance.ConclusionsThese data underline the presence of a metamemory bias in ARMS patients which is even more pronounced in FEP patients. The bias might represent an early cognitive marker of the beginning psychotic state. Longitudinal studies are needed to unravel whether metacognitive deficits predict the transition to psychosis and to evaluate therapeutic interventions.