Modeling,Simulation,and Mechanism Analysis Of Nervous System Underlying Two Behaviors In Caenorhabditis Elegans

Understanding how the nervous system generates behavior is an objective of modern neuroscience.Caenorhabditis elegans(C.elegans)is an ideal model organism for exploring the relationship between the nervous system and behavior,owing to its compact,small is scale,easily understood nervous system,and a nearly fully mapped connectome.Elucidating the neural mechanisms underlying the behavior of C.elegans not only contributes to understanding brain function in higher organisms,including humans,but also provides ideas for the advancement of machine intelligence.Despite extensive knowledge of the connectome and behavior of C.elegans,the understanding of the neural mechanisms underlying its behavior at the system level remains limited owing to insufficient fine-grained neurophysiological data,such as synaptic types,intrinsic dynamics and communication patterns of neurons.Simulation models can compensate for the lack of experimental data and overcome the limitations of observation technology to a certain extent,thereby helping to explore the mechanisms within the nervous system,and providing biologically plausible insights and hypotheses.Moreover,they offer theoretical support for the application of biological methods to engineering disciplines,such as robotics.This dissertation focuses on the undulatory locomotion and chemotaxis behavior(orientation in response to external chemical stimuli)of C.elegans.Based on its existing biological knowledge,the nervous pathways underlying these two behaviors in C.elegans are modeled,simulated,and analyzed using biological neuron models and network optimization technology.The research focuses on the neural mechanisms underlying klinotaxis and random walking during the chemotaxis behavior of C.elegans,the rhythm generation mechanisms underlying the backward undulatory locomotion of C.elegans,and the motion control and navigation decisions inspired by C.elegans.The main contents of this dissertation can be summarized as follows.1.Neural model of klinotaxis and random walking behavior in C.elegans and mechanism analysis.Klinotaxis is a chemotactic strategy in C.elegans,and random walking is evident during this process.As yet,mechanisms by which the nervous system of C.elegans generates these behaviors remain unclear.To address this issue,a neural model of klinotaxis and random walking behavior in C.elegans is established and the neural mechanisms underlying these behaviors are explored.First,an ASEL sensory neuron model with all-or-none depolarization characteristics is constructed to provide biologically realistic sensory input.The network model is constructed based on the C.elegans connectome,and klinotaxis spontaneously emerged during network evolution.Using dynamic system analysis,a state-dependent gating mechanism for klinotaxis is identified.Further,simulating the stochastic nature of biological synapses results in autonomous generation of random walking in the neural model;this verifies that random walking behavior is mainly generated by synaptic stochasticity,providing a new hypothesis as to the neural mechanism underlying random walking behavior.The neuronal ablation results suggest a certain similarity between this neural model and the biological network.2.Neuromuscular model of backward undulatory locomotion in C.elegans and mechanism analysis.Caenorhabditis elegans locomotes in a body undulatory manner,and oscillatory activity in the ventral nerve cord(VNC)underlies its motor rhythm.Currently,the origin of C.elegans backward motor rhythm has not been fully elucidated.To address this issue,a neuromuscular model of backward undulatory locomotion in C.elegans is established,and the rhythm generation mechanism is explored.The model integrates descending input from command interneurons and proprioceptive feedback from muscles in the backward VNC circuit,generating the muscle rhythmic pattern corresponding to backward locomotion and qualitatively reproducing key neurophysiological experimental findings that are not specifically designed to fit in the model.Dynamic system analysis shows that the C.elegans backward VNC circuit can generate distributed rhythms in different regions even in the absence of pacemaker neurons,either via intrinsic network oscillators or reflexive network oscillators driven by proprioceptive feedback.A series of predictions regarding this process are provided.3.An integrated model of body undulatory and chemotaxis behavior of C.elegans.C.elegans exhibits two complementary chemotaxis strategies(klinotaxis and klinokinesis)along with undulatory locomotion pattern,using a compact nervous system.Thus,C.elegans makes precise navigation decisions by simultaneously utilizing chemical gradients parallel to and perpendicular to the travelling direction,despite sensing concentration information at a single point;therefore,it is suitable for use in the development of simple,efficient robotic motion control and navigation schemes.However,existing models often fail to fully harness the advantages of the complete chemotaxis behavior of C.elegans due to their emphases on simulating a single strategy.To address this issue,an integrated model of the body undulatory and chemotaxis behavior of C.elegans is established,providing a theoretical control prototype for worm-like navigational robots.This model incorporates a rhythm generator and the proprioceptive feedback mechanism to achieve biologically realistic body undulatory locomotion.Additionally,the model incorporates the dynamic adaptivity of sensory neurons,state-dependent gating mechanism of klinotaxis,and sensory-output mapping of klinokinesis,enabling the autonomous searching for the target source of environmental variables(e.g.,chemical concentration,radiation,etc.)in directions close to the steepest gradients in environments with different gradient scales under the assumption of a single sensor.The simulation results verify the effectiveness and superiority of the proposed model.4.A spiking neural network for autonomous search and contour tracking inspired by C.elegans.A spiking neural network model is developed to address the problem of searching and tracking the specific intensities of environmental variables(referred to as set points)in complex scenarios,inspired by the experience-dependent chemotaxis behavior of C.elegans and the Lévy walk search strategy.The model comprises multiple network modules for various behavioral strategies and can autonomously switch to the appropriate navigation mode according to local environmental information.Specifically,in scenarios with gradients,the model leverages experience-dependent klinotaxis and klinokinesis strategies,enabling the autonomous searching for set points in directions close to the steepest gradients and subsequent tracking of their contours with small deviations,in environments with different gradient scales,under the assumption of a single sensor.For situations without gradients,the model exploits the Lévy walk search strategy to search for gradients,increasing the encounter rate with gradients.For local extremal traps,the model adaptively makes decisions regarding escape and escape termination based on the gradient transition characteristics of the traps,so as to exit the traps timely.The simulation results verify the effectiveness of the proposed model.