Assessment of pattern preferences by flower-naive bumblebees

Two methods for the assessment of preferences by flower-naive bumble bees (Bombus impatiens) were compared. Bees with and without prior experience on rewarding patterns were given twenty choices of unrewarding patterns (radial vs. concentric) in a radial arm maze. Either way, a preference for radial patterns was obtained. Prior training on grids of circles, squares or diamonds amplified the preference, whereas training on a ring of circles did not. Prior rewarded experience does not merely draw the bees' attention to the patterns in the maze, or serve as a motivator, but also likely leads to a similarity judgment between training and testing. Given that it was possible to test for the choices of truly flower-naive bumblebees, training is at best unnecessary and is at worst a source of bias.