A-Bees See: A Simulation to Assess Social Bee Visual Attention During Complex Search Tasks

Foraging bees often search in complex natural environments for "target" flowers that they have learnt provide nectar rewards. To maximise efficiency, bees must avoid landing on "distractor" flowers that do not offer rewards, as this potentially wastes time and energy. This paper reports on artificial-life inspired agent-based simulations of two contrasting approaches different bee species use to scan for targets in a scene containing many flowers. The two scanning approaches simulated are a parallel scan typical of bumblebees that is not slowed by distractors, and a serial scan typical of honeybees that is faster than parallel scan for single element processing, but is slowed by the presence of distractor flowers. The simulations were conducted over a range of target densities, and over a range of target/distractor ratios, to evaluate the types of environment in which each scan mechanism is most effective. Serial scan was found to be generally more effective in environments populated with a single type of rewarding species of flower, and parallel scan appears to be relatively more effective in environments populated with a mix of rewarding and unrewarding flowers. Our results support the hypothesis that environmental factors led to the evolution of different visual processing mechanisms in honeybees and bumblebees. This establishes a firm basis for psychophysical research exploring how and why the two different processing mechanisms may have evolved in these animals.