Some problems related to prediction in psychiatry are examined. The available data show, in a similar way to the social sciences, wide limitations in predictive capacity, especially with regard to suicide, violence and other behaviors. The difficulties arising from the difficulty of deriving the future from aspects of the person’s history, the lack of coherence between personality dimensions and behavior, and the privilege given to psychopathological instruments focused on the individual case, compared to actuarial instruments with testistic and statistical evaluations, are examined. The numerous cognitive biases that distort predictions are also highlighted, in particular the fundamental attribution error, which prioritizes personality variables over situational variables. But other biases have an important distortive action, from representativeness to availability, from statistical errors to framing or priming. It emerges a psychiatry still very tied in practices to common sense and folk psychology, with the richness but also the many errors that characterize it. Finally, a small ability for prediction, as recognized by popular psychology and psychiatry, exists, but linked more to situational constraints than to models of personality and psychopathology, and, in any case, poorly reliable for clinical prediction in psychiatry.
Keywords: Prediction in psychiatry; Suicide; Fundamental attribution error; Personality and behavior; Cognitive biases