Branched Optimal Transport for Stimulus to Reaction Brain Mapping

A new mathematical paper by researcher Cristian Mendico introduces a novel framework for mapping how the brain routes a stimulus to produce a reaction. Titled 'Branched Optimal Transport for Stimulus to Reaction Brain Mapping,' the work addresses a core problem in systems neuroscience. Current models focus on controlling signal trajectories across a prescribed neural network. Mendico's key innovation flips this: the transport network itself—the hidden routing architecture—is the primary unknown to be inferred from data.

The framework is posed as an anisotropic branched optimal transport problem. It mathematically defines the 'inferred object' as a graph or current connecting a source (stimulus) to a target (reaction). A concavity in the flux cost function promotes solutions where signals naturally aggregate into trunks and then branch out, mimicking efficient biological transport systems. The support of the optimal solution defines the stimulus-to-reaction map. The paper also extends the model with a hybrid stochastic approach, combining branched transport with a Kullback-Leibler control cost on the induced graph dynamics, offering a more complete picture of neural propagation.

This approach provides a foundational mathematical mechanism for moving beyond analyzing traffic on a known brain road map to actually discovering the road map itself. It shifts the paradigm from controlling trajectories on a fixed substrate to inferring the latent propagation architecture that gives rise to observed brain states and behaviors. The work has been submitted to arXiv and falls under Optimization and Control mathematics, with applications in neuroscience and quantitative biology.