Employing virtual reality to reveal individual locusts' decision-making

Abstract

Animals are constantly faced with choices where to feed, move next, or search for night shelter. Appropriate choices maximize survival and reproductive success. Research on decision-making in animals has mainly focused on which choice animals make, rather than on how they choose. In collectives, the mechanism of decision-making has been studied extensively in the last twenty years, yet little is known about the underlying mechanism in the individual's brain. Sridhar et al. (in prep) propose a neuronal model that describes the underlying mechanism of an individual's decision-making process. It predicts animals to average the direction of presented targets, moving in the centroid direction. As the angle between targets increases, the animal bifurcates at a critical angle, and eventually turns towards one of the targets. Multi-choice situations are then broken down to binary choices, successively eliminating choices, and resulting in a multi-bifurcation pattern. Sridhar et al. validate the model experimentally in Drosophila melanogaster, yet with kinematic limitations. Here, I tested the model's spatial and kinematic predictions of decision-making in the desert locust Schistocerca gregaria. In a state-of-the-art virtual reality system, I presented freely walking individual locusts with equally attractive targets. In accord with the model's predictions, locusts made decisions in a (multi-)bifurcation pattern. These cross-species results suggest that the model represents a generic, species-unspecific algorithm, robust across scales and therefore applicable to both individual and collective decision-making.