International Journal of Comparative Psychology

Bumblebees were trained to discriminate between two patterns, one rewarding (S+) and another unrewarding (S-) consisting of four orthogonal bars. Training and testing conditions were manipulated in a 2 X 2 between groups design. The training patterns differed only in the positioning of the bars in the inferior or the superior portion. The same was true of the testing patterns, both of which were unrewarding. A significant interaction between training and testing conditions was obtained on preference for one pattern, the diamond, which was present at testing for all four conditions. For the groups that were trained with the patterns that differed only in the inferior portion, when tested with patterns that differed in (1) the inferior portion: the diamond, for which the inferior portion matched that of the S+, was chosen at a level significantly above chance (2) the superior portion: the preference for the diamond disappeared-- no discriminationwas found, even though the alternative to the diamond was the same as the S+. For the groups that were trained with the patterns that differed only in the superior portion, the opposite effect of testing conditions was found: when tested with patterns that differed in (1) the inferior portion: the bees avoided the very same diamond that was preferred by the bees trained differently, and favoured the alternative, which was the same as the S+ (2) the superior portion: no pattern discrimination was found (i.e. the avoidance of the diamond disappeared). Two predictions were disconfirmed: that during testing bees would only (1) approach the pattern that was the same as the S+, or (2) discriminate between patterns that differed in the same area (inferior or superior) as did the training patterns. The data were in line with the interpretation that during differential conditioning the visual field used in future pattern discriminations is expanded to include not only the inferior portion of the pattern but more of the superior portion as well.