Adrian Horridge

In the 1920s, Mathilde Hertz found that trained bees discriminated between shapes or patterns of similar size by something related to total length of contrasting contours. This input is now interpreted as modulation in green and blue receptor channels as flying bees scan in the horizontal plane. Modulation is defined as total contrast irrespective of sign multiplied by length of edge displaying that contrast, projected to vertical, therefore, combining structure and contrast in a single input. Contrast is outside the eye; modulation is a phasic response in receptor pathways inside. In recent experiments, bees trained to distinguish color detected, located, and measured three independent inputs and the angles between them. They are the tonic response of the blue receptor pathway and modulation of small-field green or (less preferred) blue receptor pathways. Green and blue channels interacted intimately at a peripheral level. This study explores in more detail how various patterns are discriminated by these cues. The direction of contrast at a boundary was not detected. Instead, bees located and measured total modulation generated by horizontal scanning of contrasts, irrespective of pattern. They also located the positions of isolated vertical edges relative to other landmarks and distinguished the angular widths between vertical edges by green or blue modulation alone. The preferred inputs were the strongest green modulation signal and angular width between outside edges, irrespective of color. In the absence of green modulation, the remaining cue was a measure and location of blue modulation at edges. In the presence of green modulation, blue modulation was inhibited. Black/white patterns were distinguished by the same inputs in blue and green receptor channels. Left-right polarity and mirror images could be discriminated by retinotopic green modulation alone. Colors in areas bounded by strong green contrast were distinguished as more or less blue than the background. The blue content could also be summed over the whole target. There were no achromatic patterns for bees and no evidence that they detected black, white, or gray levels apart from the differences in blue content or modulation at edges. Most of these cues would be sensitive to background color but some were influenced by changes in illumination. The bees usually learned only to avoid the unrewarded target. Exactly the same preferences of the same inputs were used in the detection of single targets as in discrimination between two targets.