Fig. 5. Results of four control tests to check for the possible use of olfactory cues. Bees trained in the category discrimination experiments of Fig. 2 were presented with tests in which the sample as well as the four test stimuli were identical (in this case, they were grey-level versions of F1). The feeder was placed behind one of the test stimuli, as shown by the circular symbol. The bees chose randomly among the four stimuli, indicating that they were not using olfactory cues to make their discriminations in our experiments.