In this article [in Dutch], we provide an accessible overview of newly developed ideas about how people with autism process information and deal with uncertainty based on the predictive brain theory. Like Sherlock builds the plot of a murder to explain the traces observed at the scene of the crime and to predict new traces, the brain also builds a model of our observations and how they are caused. That mental model is constantly adjusted using prediction errors: the comparisons of the predictions that the model generates and the actual incoming signals of our senses. However, not every prediction error needs to be taken at face value, given that some errors are caused by noise or by variability in sensory signals that does not matter (in a particular situation or task).Prediction errors must therefore be given a correct weight based on an estimate of the uncertainties in the sensory information for a given context. We explain the importance of this for learning and understanding our (social) environment and how difficulties in estimating the weight of prediction errors can lead to the symptoms and cognitive profile associated with autism.