A dissociation between the use of implicit and explicit priors in perceptual inference

Abstract

AbstractThe brain constantly uses prior knowledge of the statistics of its environment to shape perception. These statistics are often implicit (not directly observable) and gradually learned from observation; but they can also be explicitly communicated to the observer, especially in humans. In value-based decision-making, these priors are treated differently depending on their implicit or explicit origin creating the “experience-description gap”. Here, we show that the same distinction also applies to perception. We created a pair of categorization tasks with implicit and explicit priors respectively, and manipulated the strength of priors and sensory likelihood within the same human subjects. Perceptual decisions were influenced by priors in both tasks, and subjects updated their priors in the implicit task as the true statistics changed. Using Bayesian models of learning and perception, we found that the weight of the sensory likelihood in perceptual decisions was highly correlated across subjects between tasks, and slightly stronger in the implicit task. By contrast, the weight of priors was much less correlated across tasks, and it increased markedly from the explicit task to the implicit task. The same conclusion holds when using the subjects’ reported priors. Model comparison also showed that different computations underpinned perceptual decisions depending on the origin of the priors. Taken together, those results support a dissociation in perceptual inference between the use of implicit and explicit priors. This conclusion could resolve conflicting results generated by the indiscriminate use of implicit and explicit priors when studying perception in healthy subjects and patients.Significance StatementThe use of prior knowledge is ubiquitous in brain processes. However, this use is not always appropriate, and the weights assigned to priors in perceptual decisions are quite heterogeneous across studies. Here, we tested whether the origin of priors can partially explain this heterogeneity. Priors that are explicitly communicated (e.g., a road sign indicating the likely presence of wildlife) appear to be used differently from priors that remain implicit and are learned from experience (e.g., previous encounters with wildlife). We show that the use of implicit and explicit priors is unrelated across subjects. This dissociation may explain why previous findings on the use of explicit and implicit priors (e.g. abnormally strong or weak priors) are often inconsistent.