Complex chemotaxis behaviors of C. elegans with speed regulation achieved by dynamic neural networks

This paper explores the complex chemotaxis behaviors of C. elegans. These behaviors include finding food and avoiding toxin simultaneously under either dual-sensory mode or single-sensory mode as well as varying locomotion speed. In dual-sensory mode, the concentration difference between left side and right side is used to determine the orientation. In single-sensory mode, a memory neuron is involved to detect the concentration difference between two time steps for navigation. First, two models are explored, namely, dual-sensory model and single-sensory model. Then, an integrated model is proposed to perform all the chemotaxis behaviors synchronously. These three models are constructed biological by extracting the neural wire diagram from sensory neurons to motor neurons and can perform left turning, right turning, and speed regulation. The chemotaxis behaviors are characterized by a set of switching logic functions that decide orientation and speed. The wire diagrams are depicted as dynamic neural networks (DNN) and trained by the real time recurrent learning (RTRL) algorithm. By incorporating a speed regulation mechanism, C. elegans can stop spontaneously when approaching food or leaving toxin. Test results verify that the biological models can well mimic the chemotaxis behaviors of C. elegans.