This is an excerpt from Neurophysiological Basis of Motor Control-3rd Edition by Mark L. Latash & Tarkeshwar Singh.
Several experimental findings prevent one from considering preprogrammed reactions as a kind of long-latency stretch reflex because preprogrammed reactions do not demonstrate a consistent dependence between the magnitude of the response and changes in muscle length. Depending on the instruction to the subject, preprogrammed reactions can be observed not only in stretched muscles but also in a muscle shortened by the perturbation or even in a muscle whose length is not changed by the perturbation at all (Marsden et al. 1979; Nashner et al. 1979; Nashner and Cordo 1981). In addition, the amplitude of the preprogrammed reactions does not correlate with the amplitude of the applied perturbation if the latter cannot be predicted by the subject. Thus, the compensation of the perturbation by preprogrammed reactions can vary in different trials from 0 to 100% or even to overcompensation (Houk 1979).
The fact that these responses are independent of the perturbation magnitude suggests that the perturbation represents a nongraded signal for the response generation, a trigger, and the response magnitude is defined prior to the stimulus on the basis of other factors. Then, certainly, these reactions can be called triggered or preprogrammed.
Consider the following scheme for the generation of preprogrammed reactions (figure 19.3). The instruction to keep a joint position against a load requires the subject to generate a voluntary command to muscles controlling this joint. If the subject knows that a perturbation can occur, a corrective command can be prepared in advance and is ready to be triggered by an appropriate peripheral signal. Note that this scheme implies preparation of a preprogrammed reaction by some “higher” center, such as the cortex, while the loop of the reaction may be limited to the spinal cord or brainstem.