Task complexity and social learning in bumblebees

A buff-tailed bumblebee, Bombus terrestris. Photo by Kathleen Purser. 

This week's paper is in Behavioral Ecology, and is lead-authored by David Baracchi, a post-doctoral researcher at the University of Toulouse in France. Baracchi and his colleagues have examined how the difficulty of a foraging discrimination task (being able to tell which flowers are rewarded) affects the use of social information in bumblebees. In bumblebees, one source of social information is the presence of another bee on a flower. The presence of that bee could indicate to an observer that a flower has available resources. Social information comes with costs and benefits (see my previous post about social learning here). For example, the presence of that bee on the flower also could indicate that the observer bee would have to compete with that bee for those floral resources. Researchers have therefore proposed that individuals should only use social information selectively, and should rely on information they have acquired through their own experience the rest of the time. This has lead to the development of "social learning strategies" which are predictions about when animals might be expected to use social information. For example, you might expect an individual to be more likely to use social information when they are inexperienced, or uncertain about the environment. Baracchi and his colleagues have examined how difficulty in discriminating rewarding from unrewarding flowers affects social information use. 

The difficulty in the floral discrimination task was determined by the difference in the number of bars displayed by rewarded and unrewarded flowers (see figure above from Baracchi's paper), such that one bar versus four bars is an easy discrimination task, whereas five bars versus four bars is a hard discrimination task. Bees were first trained to two different stimuli. They were trained to associate the presence of four bars with sugar solution, and to associate the presence of another bee with sugar solution. Then bees were given a range of tests and Baracchi et al. examined the percentage of errors (landings on unrewarded flowers) made by bees. Bees received the flowers alone, the flowers with social information consistent with the floral information (the bee was present on the four bar flower they had been trained was rewarding), or a conflicting information scenario in which the bee (which they had been taught was rewarding) was now present on the other, non-four bar, flower (either one or five, which they had been taught was not rewarding). 

Baracchi et al. show that when the task is easy, bees perform similarly regardless of whether the social information is available, consistent, or in conflict. In contrast, when the task is hard, errors are substantially reduced when the social information is consistent. These results highlight that while social information may not be important to bees when faced with easy discrimination tasks, in the cases of hard discrimination tasks, bees may be able to use additional sources of information to increase their foraging success.