Towards Understanding Brain Function by Uncovering the Conceptual Structure of C.elegans Sensorium

From International Center for Computational Logic

Towards Understanding Brain Function by Uncovering the Conceptual Structure of C.elegans Sensorium

Talk by Emma Reyner-Fuentes
The framework of Embodied and Embedded cognition suggests that animals engage in a continuous sensory-motor feedback loop with their environment, mediated by their nervous systems. A significant challenge in the study of nervous systems lies in the lack of a comprehensive model for the sensorium - the mechanism by which organisms perceive their surroundings.

In this presentation, I will introduce two information theory-based approaches that enable exploration of the sensorium of Caenorhabditis elegans, examining its connectivity, responsiveness, and neuronal characterization. The first approach demonstrates how K-Formal Concept Analysis (K-FCA), an information-theoretical tool, can be applied for Exploratory Data Analysis (EDA) of this sensorium. Through K-FCA, we can identify specific neurons’ responses to various odorants in an immediate and visual manner. Additionally, this method allows us to observe clusters of neurons based on responsiveness to odorants, as well as clusters of odorants according to their perceptual representations within the nematode’s sensorium. The second approach focuses on characterizing neurons based on their neural traces, using Convolutional Neural Networks (CNNs), Explainability techniques, and FCA. Although this approach remains a work in progress, I will present preliminary findings on how neurons can be characterized through their neural traces, revealing distinct neuronal profiles that enable their identification.

Overall, this talk will showcase tools from Information Theory, Deep Learning, and Artificial Intelligence in general, to support the exploration, analysis, and understanding of the structural organization of C. elegans’ sensorium. This will, therefore, allow us to begin to deeply understand brain functioning.