Two mathematical laws of behavior derived from work on vertebrate animals were tested in honey bees in my research. One law concerned the ubiquitous phenomenon of generalization in learning. An animal obtaining a reward for a response to one stimulus will often make that response to similar but discriminably different stimuli. Under a suitably ideal characterization, generalization gradients ought to come out exponential in shape (Shepard RN, Science 237:1317–1323, 1987). In spatial generalization in honey bees, this prediction was upheld in a number of different studies. A second law concerned the weighting of different and conflicting evidence. A piece of evidence is supposed to be weighted by its recency, with more recent evidence given higher weight (Devenport L, Hill T, Wilson M, Ogden E, J Exp Psychol Anim Behav Process 23:450–460, 1997). With the passage of time since the last evidence was obtained, overall profitability of a ‘patch’ rather than recency of profits should dominate. Tests with honey bees failed to uphold this law, instead finding circadian modulation of preferences, with ‘patch’ preference highest at the circadian time at which reward was obtained on the previous (training) day. I attempted a speculative reformulation in terms of modulation of preferences according to different oscillators.
|Title of host publication||Honeybee neurobiology and behavior|
|Subtitle of host publication||a tribute to Randolf Menzel|
|Editors||C. Giovanni Galizia, Dorothea Eisenhardt, Martin Giurfa|
|Place of Publication||Dordrecht ; London|
|Publisher||Springer, Springer Nature|
|Number of pages||14|
|Publication status||Published - 2012|