Accurate quantification of bumblebee foraging

Bumblebees have been shown to learn simple forms of tool use and to transmit these skills to other members of the colony. They are thus an excellent animal model system to study learning and social interactions. However, such studies are complicated by the difficulty of precisely quantifying bumblebee behavior and interactions with the environment even in simple tasks, such as foraging. To address this, we designed a flight chamber with 3D printed artificial flowers to accurately quantify individual and collective bumblebee foraging. A radio frequency identification (RFID) system is employed to identify the individual foragers and detect their presence in flowers. A microfluidic system releases carefully controlled sucrose solution droplets of varying sizes in each flower in response to the presence of specific bees, and we then detect when the droplets are consumed. We quantify individual foraging behavior of naive bees exposed to two flowers with different reward probabilities in the course of multiple days. This mimics the setting of the well studied two-armed bandit problem in behavioral economics. We analyze the operant learning of the relation between a flower and a reward probability by the bees, and we compare the observations to predictions of various theoretical models.