A new system to test how the brain learns novel locomotion dynamics

Johns Hopkins University, Locomotion in Mechanical and Biological Systems Lab, Summer 2017

Neural systems enable animals to obtain and process sensory inputs for locomotion control. In this project, we introduce a method to investigate how the glass knifefish, Eigenmannia virescens, learns novel locomotion dynamics. In our experiments, the fish performs a refuge-tracking task in which our computer controlled motor system enables precise control of the refuge. Hence, we can apply open-loop input signals to the refuge for system identification of the fish response and also use the fish response to apply closed-loop signals. Specifically, our system processes real time measurements of the fish, passes those measurements through a transfer function, and uses the output simulation to modify the refuge trajectory, changing the "locomotion dynamics." We hypothesized that when these novel dynamics were applied, the fish would slowly adjust the way it controls its movements. We also expected to observe a post-adaptation period, where the fish returned to its original controller when novel dynamics were removed. Preliminary results support our hypothesis and indicate that there is a learning response when novel closed-loop dynamics were introduced. Additionally, in the post-adaptation phase, the fish recovered its original controller. The experimental data suggest that we can track learning in this system and that Eigenmannia can learn new locomotor behavior while adapting to novel dynamics.

​

*Best Student Poster Finalist, 2018 Society for Integrative and Comparative Biology

*2nd Place Undergraduate Presentation, 2017 American Indian Science and Engineering Society