Perceptual Prediction: Rapidly Making Sense of a Noisy World

The Perceptual Prediction Paradox

“…Our sensory receptors are constantly bombarded with enormous quantities of information that changes rapidly across space and time [1]. This information is noisy, partly due to imperfect signal transmission in the brain but also due to indeterminacy in the signals that reach our receptors.…”

“…Like outlined in Bayesian accounts, perception is first biased towards prior knowledge to aid rapid generation of largely accurate perceptual experiences. This process may be operational from the point at which predictions can be made, through pre-activation of expected units [1,48,49]. However, when an event generates high surprise -as would be the case only for highly 'unexpected' events (see below)…”

Conceptual knowledge affects early stages of visual mental imagery and object perception

“…This theoretical framework is neurally plausible 15-20 and supported by evidence that even early stages of perception are subject to top-down influences 15,[21][22][23][24][25][26][27][28][29][30][31] . This suggests that initial aspects of imagery could be fast enough to generate early top-down effects.…”

“…Indeed, imagery and perception recruit overlapping neural circuits, including primary visual areas 1-8 , and the vividness of imagination correlates with the similarity of brain activities accompanying imagery and perception 9 . Predictive processing accounts posit that perception arises from hierarchical Bayesian predictions-essentially imaginations-that are constrained by bottom-up sensory input 10-14 .This theoretical framework is neurally plausible 15-20 and supported by evidence that even early stages of perception are subject to top-down influences 15,[21][22][23][24][25][26][27][28][29][30][31] . This suggests that initial aspects of imagery could be fast enough to generate early top-down effects.This suggestion contrasts with alternative accounts assuming that perception first runs through a strictly hierarchical succession of increasingly complex visual representations, with early stages mainly driven by bottom-up sensory processes.…”

Enhanced metacognition for unexpected action outcomes

“…1a). These accounts share a family resemblance to Bayesian models of perception 9,[12][13][14] , which assume it is adaptive for sensory representations to be weighted towards predicted outcomesmaking us more likely to see, hear or feel sensory events that we expect to occur [15][16][17][18] . Analogously to Bayesian of models of perception, Bayesian models of metacognition suggest that top-down predictions 'sharpen' internal representations of expected events, leading to more sensitive metacognition about predicted signals 11,19 .…”