Year: 2026
Type: preprint
Abstract: Visual illusions reveal that perception is constructive rather than a faithful copy of the retinal image. This paper proposes that the brain supplements ambiguous or incomplete sensory input with virtual edges, filled-in surfaces, brightness and size cues, and illusory contours. Using gray-level Shannon entropy, we compared original and perceived images across five illusion classes: edge detection, surface filling-in, brightness perception, size perception, and the Hermann grid. In all cases, perceived-image entropy exceeded real-image entropy. These findings support a unified account of visual illusions as constructive inference: the brain introduces additional information to achieve global coherence, reflecting a trade-off between low-entropy processing and perceptual interpretation.
Authors: Avijit Paul, Amrita Mukherjee
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7234458573
