That is an interesting approach, but it also suggests that the (many!) important tasks and skills which we can do but can't easily describe and reason about (for example, image recognition, and why exactly a human would decide that this image is X but not Y; or how to ride a bicycle without falling down) likely shouldn't be solved by symbolic logic.