Apathy is a debilitating syndrome that is associated with reduced goal-directed behavior. Although apathy is common and detrimental to prognosis in many neuropsychiatric diseases, its underlying mechanisms remain controversial. We propose a new model of apathy, in the context of Bayesian theories of brain function, whereby actions require predictions of their outcomes to be held with sufficient precision for "explaining away" differences in sensory inputs. In the active inference model, apathy results from reduced precision of prior beliefs about action outcomes. We tested this hypothesis using a visuomotor task in healthy adults (N = 47), with experimental manipulation of physical effort and financial reward. Bayesian modeling of performance and participants' perception of their performance was used to infer the precision of their priors. We confirmed that the perception of performance was biased toward the target, which was accounted for by relatively precise prior beliefs about action outcomes. These priors were consistently more precise than the corresponding performance distribution, and were scaled to effort and reward. Crucially, prior precision was negatively associated with trait apathy, suggesting that apathetic individuals had less precise prior beliefs about action outcomes. The results support a Bayesian account of apathy that could inform future studies of clinical populations. (PsycINFO Database Record (c) 2020 APA, all rights reserved).