Memory-like dynamics in an insect's agile tracking flight
ORAL
Abstract
Spatial orientation involves directed behaviors, some requiring explicit positional references and others relying on moment-to-moment directional information. While both types may produce similar kinematics, they imply different neural mechanisms. A positional reference needs internal comparison to a stored condition. Can we infer such stored references (memory) from kinematics?
Agile hawkmoths, Manduca sexta feed from flowers while hovering and track movements at frequencies up to 14 Hz, maintaining a fixed lateral offset. Using robotic flowers, we investigated the dynamics of this tracking behavior. The long-time dynamics of moth offsets follow a drift-diffusion process with a preferred mean offset. The offset shows short-time superdiffusion due to inertia and long-time mean-reversion at a nonzero mean. It varies between individuals and may differ for the same individual feeding multiple times. This suggests short-term memory formation at the start of a feeding bout but may also result from resetting a velocity integrator. We compared continuous and bang-bang control as explanatory models. Thus, hawkmoth tracking dynamics suggest storage of internal position memory, likely a gated multisensory phenomenon allowing adjustments based on feeding success and corrections to drift.
Agile hawkmoths, Manduca sexta feed from flowers while hovering and track movements at frequencies up to 14 Hz, maintaining a fixed lateral offset. Using robotic flowers, we investigated the dynamics of this tracking behavior. The long-time dynamics of moth offsets follow a drift-diffusion process with a preferred mean offset. The offset shows short-time superdiffusion due to inertia and long-time mean-reversion at a nonzero mean. It varies between individuals and may differ for the same individual feeding multiple times. This suggests short-term memory formation at the start of a feeding bout but may also result from resetting a velocity integrator. We compared continuous and bang-bang control as explanatory models. Thus, hawkmoth tracking dynamics suggest storage of internal position memory, likely a gated multisensory phenomenon allowing adjustments based on feeding success and corrections to drift.
*This work was funded by AFOSR grant FA9550-22-1-0315 (FLAP MURI)
–
Presenters
-
Varun P Sharma
- Georgia Tech