Search Papers | Poster Sessions | All Posters
Poster C109 in Poster Session C - Friday, August 9, 2024, 11:15 am – 1:15 pm, Johnson Ice Rink
Probing behavioral compositionality in a reconfigurable 3D environment
Tzuhsuan Ma1 (), Ann Hermundstad1, Jakob Voigts1; 1HHMI Janelia Research Campus
In natural settings, animals must navigate richly-structured sensory surroundings and rapidly adapt to changes in these surroundings. While many studies have explored navigation in mazes and open arenas, relatively little is known about how animals navigate in naturalistic terrain without clearly defined routes. Here, we probe the structure of mouse behavior in a complex, reconfigurable 3D arena in darkness and without explicit reinforcement. Mice quickly converge on a set of running and jumping paths through the arena. We developed new algorithms to partition this behavior into a set of motifs based on re-occurring path segments; this enabled us to describe long paths as hierarchical compositions of shorter sub-paths. Among the paths taken by the mice, highly compositional paths are lower in entropy, emerge later in time, and traverse more sensory diverse regions of the arena. We then introduced a local perturbation to the arena that interacted with existing composite paths. This led to a rapid emergence of new motifs that elicited non-local changes in behavior. Together, these results provide a lens for studying complex, long-timescale behavior by quantifying its compositional structure.
Keywords: compositionality naturalistic behavior complex behavior navigation