
lecture
Brain Sciences
Dimensionality bottleneck uncovers simple action selection rules in hunting zebrafish
Dr. Lilach Avitan
March 19, 2024
12:30
-
13:30
Animal movements are complex, high-dimensional, and lead to many different consequences. Thus, efficiently quantifying the behavior and uncovering the underlying representation used by the animal pose a great challenge. Tracking freely behaving zebrafish larvae using a high-speed camera and analyzing their movements, we reveal that zebrafish movements can be described using exactly two parameters. Mapping all possible two-dimensional movement representations, we identified the representation used by the fish. We show that fish do not trivially represent distance and angles as separate parameters, but rather mix them nonlinearly. Moreover, when hunting, this specific nonlinear relation depends on the prey angle and further dictates a particular set of potential movements. These results uncover, for the first time, the underlying action selection principles of hunting behavior, suggesting that behind this seemingly complex behavior there is a simple and low-dimensional process.