The Use of Visual Statistical Features in Empirical Aesthetics

Abstract

AbstractEvolution generally demands that the brain take advantage of the probable statistical structure in the natural environment. Much research in recent decades has confirmed that regular statistical features in natural scenes—especially low-level spatial regularities—can help explain processing strategies in the human visual system. Basic statistical features in various classes of human-created images broadly match those found in natural scenes. Such regularities can be seen as evolved constraints on the visual structure of aesthetic images and therefore human visual aesthetics. Some researchers have also attempted to find statistical features whose variation from natural images is associated with variations in preference and other aesthetic variables. There is evidence that variations in statistical features of luminance and color could be exploited by the visual system in certain situations. However, there is much ambiguity and variability in most reported relationships between variations in image statistical features and variations in measures of human aesthetics. In contrast, basic statistical constraints that align with efficient visual system processing are almost never violated in aesthetic images. Put simply, statistical features may constrain but may not explain variability in visual aesthetics. The chapter concludes with an outlook on future directions for research.