Constancies and illusions in visual perception

Abstract

This paper presents a systematic exposition of the general structure of visual constancies and illusions, including the introduction of a number of conceptual distinctions, illustrated by many examples. The study of these phenomena involves the distal, the proximal, and the phenomenal domain. The relations of concordance and discordance between pairs of domains are defined, followed by the definitions of four visual modes (concordant, proximal, constancy, illusion) as particular constellations of concordance-discordance relations between all three domains. Constancies and illusions are characterized by proximal-phenomenal discordance. Attributes of entities of visual domains are divided into the geometric (size, shape, location, orientation) and the photometric (reflectance, illumination) class. The phenomenal domain involves two types of attributes, one group distally and the other proximally focused. Research on both constancies and illusions can be described as involving the study of the effects of two independent variables on a dependent variable. The first independent variable (target variable) is a distal attribute, such as size, shape etc. In constancy studies, the second independent variable (confound variable), is a variable such as distance, orientation etc, that, together with the confound variable, affects the corresponding proximal variable (such as proximal size, shape etc). In illusion studies, confound variables do not affect the proximal variables, but do affect the corresponding phenomenal variables. The main part of the paper consists in the descriptions of studies of constancies and illusions of size, shape, location, orientation, and achromatic and chromatic color, all presented in a common format, which facilitates the comparison of their similarities and differences. The importance of presentation conditions (full-cue versus reduced cue) and instruction type (distally versus proximally focused) is stressed. Finally, salient cases are pointed out in which relations between phenomenal variables tend to take a form qualitatively similar to the relation of the corresponding non-phenomenal variables.