Decoding Typical Flight States Based on Neural Signals from the Midbrain Motor Nuclei of Pigeons

Long Yang,Erteng Ma,Lifang Yang,Mengmeng Li,Zhigang Shang,Liaofeng Wang,Zuohao Ma,Jiajia Li

DOI: https://doi.org/10.31083/j.jin2304072

2024-04-03

Abstract:Background: Exploring the neural encoding mechanism and decoding of motion state switching during flight can advance our knowledge of avian behavior control and contribute to the development of avian robots. However, limited acquisition equipment and neural signal quality have posed challenges, thus we understand little about the neural mechanisms of avian flight. Methods: We used chronically implanted micro-electrode arrays to record the local field potentials (LFPs) in the formation reticularis medialis mesencephali (FRM) of pigeons during various motion states in their natural outdoor flight. Subsequently, coherence-based functional connectivity networks under different bands were constructed and the topological features were extracted. Finally, we used a support vector machine model to decode different flight states. Results: Our findings indicate that the gamma band (80-150 Hz) in the FRM exhibits significant power for identifying different states in pigeons. Specifically, the avian brain transmitted flight related information more efficiently during the accelerated take-off or decelerated landing states, compared with the uniform flight and baseline states. Finally, we achieved a best average accuracy of 0.86 using the connectivity features in the 80-150 Hz band and 0.89 using the fused features for state decoding. Conclusions: Our results open up possibilities for further research into the neural mechanism of avian flight and contribute to the understanding of flight behavior control in birds.