A framework for analyzing C. elegans neural activity using multi-dimensional hyperbolic embedding

Area of Science:

• Neuroscience
• Computational Biology
• Systems Biology

Background:

• Neural activity patterns underlie sensory processing and behavioral output.
• Understanding the dimensionality and geometry of neural activity is crucial for deciphering sensorimotor transformations.
• The contribution of network activity versus sensory input to non-sensory neuron firing remains unclear.

Purpose of the Study:

• To investigate the dimensionality and geometry of neural activity in Caenorhabditis elegans (C. elegans) in response to bacterial stimuli.
• To characterize the temporal dynamics of sensory and motor neuron clusters.
• To determine if neural activity can be represented in hyperbolic space and its relationship to sensorimotor transformation.

Main Methods:

• Recorded neural activity from most head neurons in C. elegans exposed to bacterial stimuli.
• Classified active neurons into functional clusters (sensory and motor/command).
• Estimated stimulus selectivity and used hyperbolic embedding to analyze neural dynamics.

Main Results:

• Identified six functional neuron clusters: two sensory and four motor/command.
• Sensory neurons responded maximally within 15 seconds, while motor/command neurons responded tens of seconds later.
• Neural dynamics were best described by an eight-dimensional hyperbolic space, outperforming Euclidean space.
• The hyperbolic space revealed distinct sensory and motor (forward-backward, dorsal-ventral) components.

Conclusions:

• C. elegans neural activity in response to bacterial stimuli can be effectively represented in a low-dimensional hyperbolic space.
• This hyperbolic representation aids in understanding sensorimotor transformations.
• The findings provide a scalable method for analyzing neural activity and its link to behavior in complex systems.