In partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Quantitative Biosciences
in the School of Biological Sciences

Varun Sharma

Defends his thesis:

Multisensory Integration and Memory in Hover Feeding Hawkmoths

Wednesday, July 23, 2025

9:00am Eastern

Location: Howey Physics Building N201-202

Zoom: https://gatech.zoom.us/j/5708722028?omn=99074316368

Advisor: 

Dr. Simon Sponberg

School of Physics, School of Biological Sciences

Georgia Institute of Technology

Committee: 

Dr. Tim Cope

School of Biological Sciences

Georgia Institute of Technology

Dr. Young-hui Chang

School of Biological Sciences

Georgia Institute of Technology

Dr. Gordon Berman

Department of Physics

Emory University

Dr. Flavio Fenton 

School of Physics

Georgia Institute of Technology

Abstract:

     Animals perform continuous, goal-directed behaviors by combining sensory modalities, internal states and neuromuscular control. Agile flight behaviors in insects are especially exciting avenues for studying sensorimotor control, due to their small nervous systems and relatively fast experimental turnaround times. Hawkmoths are a model system for sensorimotor control, due to their remarkably linear hover-feeding behavior, large size, and agile flight.

Flower-tracking behavior in the hover-feeding hawkmoths, Manduca sexta is remarkably amenable to control theoretical treatment. Beyond system identification, it is composed of object recognition, 3D tracking, active sensing, multi-sensory and multi-frequency integration, short-term memory, attention, airflow response and motor adaptation, and may have links to navigation circuits. In this work, we focus on behavioral identification of their control system, and kinematic identification of short-term reference points (goals) in this control. Further, we investigate the neural basis of these behavioral observations and develop novel techniques for behavioral electrophysiology.

In Aim 1, we study the effects of luminance context on visual and mechanosensory feedback control of flower-tracking. Specifically, we test the effect of light-levels on mechanosensory response and measure the tracking performance as a function of luminance change.  We use robotic flowers to present sensory conflict and test the system’s linearity across luminance contexts.
In Aim 2, we observe memory like kinematics in hover-feeding, indicating the presence of positional memory in this control. We apply time-series approaches to formalize this as a memory-based system and further test the positional and velocity contributions to the control using a screen-feeding free-behavior VR system to present step-changes in position.

In Aim 3, we investigate the neural basis of multisensory integration, in a bottleneck layer between the brain and the motor circuits. We characterize the anatomy of this bottleneck using Transmission Electron Microscopy. Further, we use multichannel, extracellular electrodes to record neural activity in this neck-connective while presenting visual and mechanosensory stimuli with robotic flowers. Finally, we develop a novel behavioral electrophysiology technique, allowing the moths to flap their wings during electrophysiology. We test two alternative behavioral measurement techniques for this setup, one based on force-sensing, and another based on kinematics. These approaches bring us closer to having behaving animals on an electrophysiology rig, for studying closed-loop feedback control in small nervous systems.