Vestibular Spacetime Tensor (VST) Theory: A Geometric Framework for Neural Computation and Motor Control

The brain faces a fundamental computational challenge: constructing a stable perceptual representation of the world from sensory inputs that are continuously distorted by self-motion. This paper introduces the Vestibular Spacetime Tensor (VST) theory, a novel framework positing that the nervous system resolves this challenge by treating the vestibular system not merely as a sensory organ, but as the source of a foundational, four-dimensional spacetime coordinate system. We propose that this input, representing the dynamic state of the head through acceleration and its temporal derivative (jerk), forms a covariant sensory vector. Neural computation, particularly for sensorimotor control and spatial cognition, is conceptualized as a geometric transformation within this intrinsic manifold. Specifically, we argue that the cerebellar microcircuit instantiates a metric tensor, mathematically transforming the covariant vestibular state vector into a contravariant motor command vector, thereby generating predictive and compensatory actions. This transformation is continuously refined through a predictive coding mechanism, where climbing fiber inputs signal a sensory prediction error that drives synaptic plasticity, effectively updating the components of the metric tensor to minimize future error. VST theory offers a unifying framework that integrates principles from theoretical physics, computational neuroscience, and robotics, providing a neurobiologically plausible mechanism for state estimation, motor learning, and spatial awareness. It recasts the cerebellum as a predictive engine that models the physics of movement, offering novel, testable hypotheses regarding the neural basis of perception and action.