Bumblebee goal choice
How do bumblebees decide between multiple goal options?
How animals acquire, process, and combine information in an uncertain world to navigate accurately between their home and a behaviourally relevant location, such as a food source is a fundamental question in biology. Bumblebees, for example, face the challenge of finding food sources, i.e. flowers, among numerous unrewarding environmental objects. Inexperienced foragers first need to explore the environment to locate objects that may be candidates based on visual and/or odour cues. If a flower proves to be rewarding, its sensory characteristics and location must be learned before returning back to the nest, so that it can be found again on a later foraging trip.
Goal finding has often been considered from the perspective of locating a unique goal that is not visible even from a relatively short distance, as in the case of the nest holes of many ants, bumblebees, or solitary wasps and bees. Many experiments on locating a food source also used largely inconspicuous feeders in the training situation, which were removed altogether in the subsequent tests. However, it is not enough to determine the location of a goal when targeting food sources such as flowers, because these must additionally be identified and distinguished from the many non-rewarding objects in the environment. Visual features such as colour, shape, and texture, but also odours can be important here.
In the foraging context, there are usually several alternatives for flowers as food sources. These may be conspicuous, differ in their sensory characteristics, and may provide a different food quality, implying decisions and possibly flexible learning for the bee. The different constraints for finding the nest and profitable food sources suggest that the underlying goal-finding mechanisms may differ as well.
The overall objective of the project is to develop a comprehensive concept of goal-finding behaviour along the ecological constraints of foraging with a focus not only on goal localisation, but also on the identification of goal objects in cluttered environments, especially in the case of multiple goal options. When deciding between goal options, different values assigned to them through experience may play an important role.
We pursue this objective by combining behavioural experiments and computer-simulated models of the control mechanisms in the bee's neuronal system. In our experiments, bumblebees walk on a air-suspended treadmill in a computer-controlled virtual environment. This allows us to precisely control the stimuli the bumblebee perceives and to apply targeted manipulations to the action-perception loop the animal experiences. The experiments are tightly coupled to the development of a formalized control model simulating the bumblebees’ behaviour in the same virtual environments.