Predictive control of intersegmental tarsal movements in an insect

Area of Science:

• Neuroscience
• Biophysics
• Robotics

Background:

• Intersegmental reflexes are crucial for animal movement and posture but remain poorly understood.
• Mathematical modeling offers a way to analyze these complex responses beyond simple input-output descriptions.

Purpose of the Study:

• To develop a novel method for measuring and quantifying intersegmental tarsal reflexes in locusts.
• To apply Artificial Neural Networks (ANNs), specifically Time Delay Neural Networks, to model these reflex dynamics.

Main Methods:

• Stimulation of the femoro-tibial chordotonal organ in locust hind legs.
• Utilizing a novel measurement technique to record tarsal appendage movement.
• Employing Time Delay Neural Networks for response analysis and prediction.

Main Results:

• ANNs accurately predicted tarsal movement, outperforming linear models when trained on reflex responses to Gaussian White Noise.
• The ANN model demonstrated the ability to predict individual animal responses and responses to other stimuli like sinusoids.

Conclusions:

• Artificial Neural Networks provide a powerful tool for modeling complex biological reflexes.
• Understanding these reflexes can inform the design of advanced prosthetics, functional electrical stimulation therapies, and bio-inspired robots.