Abstract
Twilight presents a challenging visual environment for the perception of surface color. Color signals are highly variable during this semidiurnal transition between photopic and scotopic vision. This variability stems from rapidly changing light level, shifts in spectral illumination, and changes in sensory processing due to rod involvement. Research on surface color perception at low light levels is typically conducted in a laboratory setting that simulates twilight illumination but does not duplicate it (e.g., Middleton & Mayo, 1952; Porkorny, Lutze, Cao, & Zele, 2006). The present study investigates surface color perception under natural illumination in the absence of anthropogenic light sources. Observers used eleven basic color terms (Boynton & Olson, 1987) to categorize the color of spectrally calibrated Munsell color chips that were representative of the Munsell color space. The color chips, selected from three constant lightness planes (V4, V6, V8), were presented in randomly ordered arrays. Judgments were made in the spring and summer months under clear skies at specified time intervals during civil and early nautical twilight. For the majority of the chips, surface color was perceived over the duration of civil twilight for all lightness planes and light levels. Two general shifts in color categorization were observed as light level decreased: 1) the number of chips categorized as red, brown, and gray increased, and 2) the number of chips categorized as blue, green, yellow, purple, pink, and orange decreased. These shifts occurred for all lightness planes and are not accounted for by changes in spectral illumination. Rapid loss of surface color began at nautical twilight making reliable color judgments impossible. Although color categorization shifted, most surface colors could still be categorized into one of the eleven basic colors during civil twilight. This finding suggests that surface color perception is maintained at low light levels in a natural environment.