Master of Research Thesis
Assessing Flexible Information Use in the European Honey Bee (Apis mellifera)
Abstract
The European honeybee (Apis mellifera) has been shown to be capable of solving a wide range of cognitive tasks. It is however currently unknown what the limitations of their cognitive abilities are. Given the bee has a relatively tiny brain with less than one million neurons, it is surprising that these limitations have not been well characterised. To explore the nature of limits to bee cognition here we explored the capacity of bees to solve conflicting information in two different delayed conditional discriminations and a sequential conditional discrimination. In addition we explored whether there was any evidence bees learning from their errors by assessing for a post-error slowing effect, a phenomenon found in vertebrates. These general learning abilities underpin a range of cognitive tasks, and have been considered hallmarks of intelligent behaviour.
It was found that bees performed above chance levels in all three discrimination tasks, with their accuracy improving over learning trials, demonstrating that bees are capable of resolving conflicting information using conditional information. The reaction time in post-error trials were found to not display a significant slowing phenomenon. This suggests that animals with limited neurons and small brains can adapt to conflicting information but with some limitations in that they do not adapt their behaviour to their errors.
Delayed Conditional Discrimination Task

Individual bees (n=22) were presented with a delayed conditional discrimination task of three stimuli; a reward of sucrose and a punishment of quinine on either a blue or a yellow stimuli and a third unrewarded option of a green stimulus that always offered water. In order to solve this conflicting information task bees were given a contextual cue, in the form of either a black or a white entrance that would indicate if a stimulus was rewarded or punishment (fig. 1). This was counted as a single trial, with each bee being given a total of 30 trials. Half of the sample size (n=11) were presented with condition 1 and the other half (n=11) with condition 2 (fig. 2).

Sequential Conditional Discrimination Task
Bees (n=22) were offered three choices: a rewarded stimulus, a punished stimulus and an unrewarded water stimulus. To solve the task the bees were given a contextual cue in the form of a fixed sequence: three trials of contingency A before switching to three trials of contingency B and so on until 30 trials had been completed (fig. 10).
