Is There an Equilibrium Point Hypothesis?

Abstract

The notion of studying internal control variables instead of externally observable parameters is certainly attractive. It is what motivates neurophysiologists to stick their electrodes into various parts of the nervous system. The problem is knowing a control variable when you see one. Recorded neural activity reflects a mixture of outputs representing some kind of control to various lower centers plus inputs representing both the neurons’ own control from higher centers as well as sensory feedback and efference copy from lower centers. So inevitably we guess at what might be a control variable at any particular level, look for correlations and hope for causality. If we guess that an observable parameter is actually the controlled variable (e.g. end-point trajectory in extra-personal coordinates), then the exercise is straightforward, albeit perhaps pointless as discussed below. If the hypothesized controlled variable must be computed from various parameters of the task, then it seems essential to justify the choice and to suggest how it might be computed by the nervous system. The equilibrium point (EP) hypothesis suggests that there are two control variables, CV1 related to the balance between antagonistic muscles controlling an equilibrium posture, and CV2 related to cocontraction of antagonistic muscles thereby modulating the impedance (stiffness) of the posture. While this idea obviously resonates with early research on control of single-joint movements and reciprocally organized stretch reflexes in antagonist muscles, it is remarkable how vague it becomes when extended to more realistic, multiarticular movements and musculature (see list below). This presumably relates to the “lack of tools to study patterns of control variables”, a shortcoming that the proponents of the EP hypothesis must resolve if their hypothesis is to be testable.