Contours information and the perception of various visual illusions

TL;DR

This paper introduces a quantitative framework based on information content and surprisal to explain how the human visual system perceives illusions, reconciling global and local interpretation principles.

Contribution

It proposes a novel information-theoretic model that explains visual illusions and perceptual grouping by evaluating contour information and expectations.

Findings

Successfully explains perception of Kanizsa triangle, Ehrenstein cross, Rubin's vase

Reconciles global minimum and local minimum principles in visual perception

Provides a fundamental explanation for illusory boundaries and bistability

Abstract

The simplicity principle states that the human visual system prefers the simplest interpretation. However, conventional coding models could not resolve the incompatibility between predictions from the global minimum principle and the local minimum principle. By quantitatively evaluating the total information content of all possible visual interpretations, we show that the perceived pattern is always the one with the simplest local completion as well as the least total surprisal globally, thus solving this apparent conundrum. Our proposed framework consist of (1) the information content of visual contours, (2) direction of visual contour, and (3) the von Mises distribution governing human visual expectation. We used it to explain the perception of prominent visual illusions such as Kanizsa triangle, Ehrenstein cross, and Rubin's vase. This provides new insight into the celebrated simplicity principle and could serve as a fundamental explanation of the perception of illusory boundaries and the bi-stability of perceptual grouping.