Psychotic experiences (PEs) occur in the general population, beyond psychotic disorders. PEs are a risk factor for mental ill health in young people but can occur benignly in selected samples of adults. Environmental factors predispose to PEs but their underlying mechanisms are not well-understood. Progress in understanding PEs may be limited by diverse conceptualisations, imprecise measurement and a lack of explanatory frameworks that can bridge the gaps between aetiological factors, their effects on the brain and their behavioural manifestations. In this thesis, I undertook a comprehensive investigation of the measurement, health implications, aetiology and computational mechanisms of PEs in adolescents and young adults using data from two large cohort samples, supplemented with smaller-scale behavioural studies. I first investigated the measurement of PEs. I assessed and optimised the measurement of PEs in young people by two self-report instruments. I then used latent variable modelling to show that a self-report and interview instrument measured the same underlying psychotic phenomena. Both instruments were able to measure severe PEs, while the self-report questionnaire also measured more mild psychotic phenomena. I then investigated the health implications of PEs. Using cluster analysis in both cohorts, I found replicable patterns of PEs at similar levels of intensity and persistence but with and without depressive symptoms and with varying risk of mental disorder. Paranoid ideation was more associated with depressive symptoms than non-paranoid unusual perceptions and beliefs. Childhood adversity was associated with both PE-prone groups, but later social support from family and friends was far higher in those with PEs and low depressive symptoms than those with PEs and high depressive symptoms. Subsequently, I investigated the role of the social environment in the development of PEs and psychopathology using longitudinal structural equation modelling. I found that asocial dispositions increased or preceded increase in PEs over one year, mediated by detriment to social support. Conversely, PEs did not precede or increase asociality. I then showed that dimensions of PEs and depressive symptoms were promoted by childhood adversity but differentially affected by later social support, with paranoid ideation being more influenced by support than non-paranoid unusual perceptions/beliefs. Finally, I investigated specific mechanisms of PEs in two behavioural studies. In the seventh study, I used computational modelling of reward learning to link PEs to reduced ability to modulate learning by confidence, replicating computational effects of a pharmacological model of psychosis. I also used a novel visual task to show that the manifestation of PEs as anomalous perceptions versus anomalous beliefs might be explained by over-reliance on different types of prior knowledge in perceptual inference. These results suggest that different conceptual approaches to PEs might be synthesised despite issues with their measurement. PEs in young people, while not entirely benign, are heterogeneously associated with psychopathology. Importantly, they characterise a minority of young people who are at very high transdiagnostic risk of mental illness but also occur without distress in young people, often in the context of a supportive social environment. Health outcomes in young people with PEs are predicted and potentially modified by social functioning and social relationships. PEs might arise from atypicalities in how the influences of information sources on perception and belief-updating are modulated according to their reliabilities.