From Motility to Mind: Why Large Language Models Lack the Architecture of Consciousness

This paper derives cognitive architecture from two theses: motility precedes sensation, and the brain is an extension of the sensory surface rather than its cause. From these we trace a causally continuous sequence of structural emergents culminating in the internal representational canvas — the shared multimodal space mediating between sensation and action, grounded in the thalamocortical loop and pan-vertebrate since the lamprey (~560 Mya). Basal activation emerges from concurrent signal density. Prioritization operates at three non-interchangeable levels — architectural, training, and fine-tuning — where the architectural level defines the fundamental character of the mind. We specify four functional requirements for a synthetic canvas and propose a Recurrent Bidirectional Hub geometry. Large language models are dialoguers without fully developed canvas: they occupy the final linguistic layer without the underlying layers that generated it. The gap is architectural and solvable by engineering, not by metaphysics.