A hierarchical structure for perceptual awareness in the human brain

Accounting for why sensitivity to perceptual input (as assayed by discrimination judgments) is not always accompanied by conscious awareness (as assayed by detection judgments) remains a challenge for theories of perception. Here we test a hypothesis that awareness is supported by higher-order inferences within generative models of perceptual content. We develop a novel visual perception paradigm that probes such inferences by orthogonally manipulating expectations about stimulus content (discrimination) and awareness of content (detection). In line with model simulations we show that both detection and discrimination expectations influence reaction times on a categorisation task. By combining a no-report version of our task with functional neuroimaging we reveal a neural dissociation between prediction errors (PEs) on content (discrimination) and awareness of content (detection): content PEs are tracked in posterior sensory cortex while awareness PEs are tracked in prefrontal cortex. Together, our results reveal a hierarchical structure supporting visual detection and discrimination, consistent with a proposal that awareness reflects a higher-order inference within perceptual generative models.