Maybe this extends to the cortex. In numenta theory, layer 5 is possibly a displacements layer. A displacement is a difference between two positions. Likewise, in what you’ve summarized, there’s a target (“reference”; muscle length etc.) and sensory representations are used to determine the error between the current state and the target. When the target changes, a movement occurs. So in numenta theory, the cortex would be part of this hierarchy of comparators.
For displacements, the brain also needs to be able to recognize displacements caused by the external world, not just behaviors. That’s also necessary for the negative feedback for motor control, so there’s no difference there.
The only difference is, you might not produce a movement just because a displacement is non-zero. For example, you see a cup fall off a table. Your eyes could track the cup, or not.
There are many possible displacements, because there are many different reference frames. In some of those reference frames, when the cup falls, there is no displacement. If the reference frame is just the direction the eyes are pointing, then each cortical column’s location wouldn’t change, although the feature would. So if you focus on pointing your eyes at a direction in their sockets, your eyes don’t need to track the cup. If you focus on pointing your eyes at the cup, your eyes will track the cup. So every displacement is a potential behavior, if the displacement is caused by the environment. (Or the displacement is just literally a behavior, like when you move a fingertip to explore an object.) What matters is attention.
In L5tt cells are attentional (paper summaries), L5tt cells cause the brain to notice a stimulus. To do so, they burst, triggering detection via thalamus and a couple other things. If L5tt is the displacements layer, perhaps a displacement is the same as noticing something important (like information about the object’s identity). So a displacement would be for locations in the sense of, “information which makes a feature unique to an object, with a unique set of location SDRs for each object”. Not some sort of more raw location. That leads to a possible mapping to biology:
L5tt burst (detection / displacement) → thalamus (perhaps map a bunch of displacements to each destination location, if that’s useful) → L2/3 metabotropic receptors, providing persistent activation for the object layer. In the object layer, cells need to have persistence after they activate. If the metabotropic receptors are voltage-gated and on the apical dendrites, they might only produce a persistent response if proximal inputs cause the cell to fire, because the apical dendrites might require a bAP to generate a calcium spike. So the persistence in biology would also only happen after the cell activates as a result of feedforward sensory input.