Long term recordings of Drosophila melanogaster behavior at high temporal resolution

Behaviors evolve over many different timescales, from the short, seconds-long individual articulations required for vocalizations to the sometimes decades-long process of aging. Initial forays into uncovering the long timescale structure of behavior have shown that behavior evolves in a non-Markovian fashion, and is hierarchically organized in some systems¹. However, these studies have been limited to the approximately hour-long timescales of the previously available behavioral datasets. We have developed a method to take continuous, high-dimensional, high-resolution behavioral data of the freely moving fruit fly Drosophila melanogaster spanning days and weeks. We apply modern techniques of pose-estimation (SLEAP)² and computational ethology (MotionMapper)³ to quantify the behavior of recorded flies, and use these data to study circadian features of behavior, and to characterize how D. melanogaster's habituation to novelty varies in different behavioral modes across days. We are also studying how behavior changes with age, and how early life behaviors may predict later activity.

1. Berman, G.J., et al. Proc Natl Acad Sci USA, 113.42 (2016): 11943-11948.

2. Pereira, T.D., et al. Nat Methods, 19.4 (2022): 486-495.

3. Berman, G.J., et al. J R Soc Interface, 11.99 (2014): 20140672.