A robot-rodent interaction arena with adjustable spatial complexity for ethologically relevant behavioral studies
Publication date: 27/02/2024
Authors: Lai A.T., Espinosa G., Wink G.E., Angeloni C.F., Dombeck D.A., MacIver M.A.
Journal: Cell Reports
Commentary: The study by Lai et al. introduces the “cellworld system” a modular framework designed to bridge the gap between rigid laboratory control and real-world behavioral relevance. By combining a large, reconfigurable honeycomb lattice arena with a high-speed autonomous robot, the researchers moved beyond the constraints of traditional mazes. The system's primary strength lies in its bidirectional reactivity and plasticity; the robotic “pseudo-predator” responds to the mouse’s position with millisecond latency, and the surrounding environment can be continuously adapted. This creates a dynamic interaction space that closely recalls a real-world predator-prey encounter. Specifically, the authors demonstrate that the complexity of the environment is a major driver for cognitive flexibility. Mice defaulted to basic, stereotypical escape behaviors in simpler environments, but in the cellworld maze, they exhibited more sophisticated behaviors such as “peeking” around obstacles and “baiting” the robot. These actions suggest that mice used internal models and high-level path planning, providing a controlled way to study more sophisticated and ethologically relevant behaviors for coping with the environment. Despite these advances, the study acknowledges some evident limitations. The robot currently operates at approximately one-third the speed of a mouse, and the researchers masked olfactory and auditory cues to simplify the mouse-robot encounter. Furthermore, the robot airpuff lacked the true lethal stakes and complex ambush strategies found in the wild. Nevertheless, the cellworld system creates a seamless synergy with reinforcement learning and computational ethology, representing a vital step toward quantifying natural behaviors while maintaining the high level of repeatability required for modern neuroscience.
Commented by: Filippo La Greca
DOI: https://doi.org/10.1016/j.celrep.2023.113671
