Seeing the light: illumination as a contextual cue to color choice behavior in bumblebees

Abstract

The principal challenge faced by any color vision system is to contend with the inherent ambiguity of stimulus information, which represents the interaction between multiple attributes of the world (e.g., object reflectance and illumination). How natural systems deal with this problem is not known, although traditional hypotheses are predicated on the idea that vision represents object reflectance accurately by discounting early in processing the conflating effects of illumination. Here, we test the merits of this general supposition by confronting bumblebees (Bombus terrestris) with a color discrimination task that can be solved only if information about the illuminant is not discounted but maintained in processing and thus available to higher-order learned behavior. We show that bees correctly use the intensity and chromaticity of illumination as a contextual cue to guide them to different target colors. In fact, we trained bees to choose opposite, rather than most similar, target colors after an illumination change. This performance cannot be explained with a simple color-constancy mechanism that discounts illumination. Further tests show that bees do not use a simple assessment of the overhead illumination, but that they assess the spectral relationships between a floral target and its background. These results demonstrate that bees can be color-constant without discounting the illuminant; that, in fact, they can use information about the illuminant itself as a salient source of information.

Fig. 1.

Relative spectral reflectances of artificial flowers and background (A) and relative spectral distribution of illumination (B).

Fig. 2.

Experimental setup, color loci, and behavioral results with two chromatic illuminants. (A) Schematic illustration of the two testing conditions: blue illumination on the left and green illumination on the right. (B) Location of spectral stimuli arising from the flowers within bee color space (blue and yellow, inner dots) under blue and green illumination (outer rings). Note that the space presented is the central portion of the color hexagon. The x axis represents UV vs. green antagonism, and the y axis represents blue vs. UV-green antagonism. (C) Results of testing the ability of bees to use changes in chromatic illumination to guide visual behavior. “Correct” indicates choosing a target color consistent with the test illuminant. Error bars indicate the standard deviation of each population, which combines bees from each of the two paradigms, because their χ² distributions did not vary significantly from each other. Significance of difference from random behavior (noted in the plot) was determined by using the χ² test.

Fig. 3.

Experimental setup, color loci, and behavioral results with white light but two differently colored arena floors. (A) Schematic illustration of the two testing conditions, which are described in the text. (B and C) Location of spectral stimuli arising from the flowers within bee color space under blue or white light. Blue-filtered blue flowers are represented as darker blue circles. Outer rings represent the color of the illumination. Surrounding squares in C represent the blue-filtered green background. The x axis represents UV vs. green antagonism, and the y axis represents blue vs. UV-green antagonism of the color hexagon. (D) Results of testing the ability of bees to use changes in chromatic illumination to guide visual behavior.

Fig. 4.

Experimental setup, color loci, and behavioral results with two different illumination light intensities. (A) Schematic illustration of the two testing conditions, which are described in the text. (B) Location of spectral stimuli arising from blue and yellow flowers (dots) under bright and dim light (rings) within bee color space. Color loci of the x axis represent UV vs. green antagonism, and color loci of the y axis represent blue vs. UV-green antagonism. (C) Results of testing the ability of bees to use changes in illumination intensity. “Correct” indicates choosing a target color consistent with the test illuminant.