We often assume that cognitive traits, such as social learning abilities or spatial memory, are evolutionarily advantageous. Yet this assumption may be seriously flawed, because investment in cognitive abilities can come at a significant cost, trading off against other fitness-determining traits. It has proved surprisingly difficult to test whether “clever” animals outperform their conspecifics in the wild, because it is hard to control for previous experience and other confounding factors that might influence performance in cognitive tests. In this project, we will focus on the bumblebee Bombus terrestris- a model for which an impressive range of ecologically relevant cognitive tests now exist. Bees will be tested in the laboratory, and then released into the wild to assay foraging performance in the real world. By capitalizing upon the extraordinary tractability of the bee system, we will seek to understand the link between cognitive traits and fitness in the wild.