You’re right to highlight that much of our knowledge—especially sensorimotor, mi

You’re right to highlight that much of our knowledge—especially sensorimotor, mimetic, and pre-linguistic—is encoded non-verbally. But that doesn’t mean it’s unknowable, only that it’s non-propositional. It’s embodied, procedural, and episodic rather than symbolic.

The mistake is in assuming that language is the only means of encoding or transmitting knowledge. In my work, I treat language not as a container of truth but as an index into a network of operational sequences—tests, performances, transformations. We don’t need vocabulary for everything. We need operational commensurability—the capacity to represent, replicate, or verify a behavior, transformation, or inference, whether in muscle memory or machine execution.

AI does make errors when it lacks sufficient operational grounding—when it attempts to infer causality from symbolic correlation rather than from a model of demonstrated, repeatable behavior. This isn’t a failure of AI per se—it’s a limit of any system not yet trained on the relevant operational sequences. Just as a child fumbles before learning to tie shoelaces by repetition, so too does a model without feedback from embodiment or sufficient training data.

So yes—mimetic, imitative, and procedural learning is foundational. But what we call “language” is simply one layer of the stack. The deeper layer is sequence learning—motor, sensory, symbolic, or otherwise. My system emphasizes testifiability and demonstrated interest precisely to bridge this gap between symbolic and operational knowledge, and to measure whether what’s being claimed can be done, performed, or validated—not merely said.

Reply addressees: @slenchy @bryanbrey