Regularity of colour statistics in explaining colour composition preferences in art paintings

This study explores the role of colour statistics in painting preferences and tests the ‘matching-to-nature’ hypothesis which posits that the preference for the colour composition of paintings depends on the extent to which the paintings resemble the colour statistics of natural scenes. A preference judgement experiment was conducted with 31,353 participants using original and hue-rotated versions of 1,200 paintings. Multiple regression analyses were performed between the measured preferences and paintings’ colour statistics to reveal which colour statistics explained the preference data and to what extent. The colour statistics of art paintings that explained the preference data were compared to the colour statistics of natural scenes. The results identified the colour statistics that significantly contributed to explaining painting preferences, and the distributions of the paintings’ colour statistics systematically differed from those of natural scenes. These findings suggest that the human visual system encodes colour statistics to make aesthetic judgements based on the artistic merit of colour compositions, and not on their similarity to natural scenes.