In [2], the control variables have been defined in two different ways. The theory is formulated in terms of a variable called lambda, from which it derives its name, that is defined as a threshold at which the tonic stretch reflex will start to activate a motor neuron pool. This is a bioelectrical definition. Lambda is also described as the angle at which the torque-angle relation of the tonic stretch reflex crosses the abscissa and muscle torque equals zero, a biomechanical definition. Two new variables called Reciprocal (R) and Coactivating (C) central commands are then defined respectively as the sum and the difference of the lambdas of the antagonist muscles. R and C are also described in terms of their mechanical properties; R being the determinant of the equilibrium position (EP) of the joint and C modulating it’s stiffness. To move, the theory postulates that the CNS specifies the temporal patterns of the central commands R and C and of another central command variable called m that controls feedback gain. These commands, in combination with peripheral feedback signals from muscle proprioceptors, activate the muscles. According to [2, p 731], "a change in the R command gives rise to a shift in the EP of a joint" and if external forces permit, movement follows. "The C command [influences joint stiffness but] does not affect the EP of the joint." An implicit assumption here is that the transformation of lambda into muscle tension, a transformation that involves the full spectrum of motoneuron, muscle fiber and joint biomechanical properties, is identical for both flexors and extensors. The antagonist muscles/reflexes must have, in the model’s terminology, identical static "invariant characteristics" and identical dynamical joint torque (e.g. twitch and force/velocity) characteristics. At few (and perhaps no) joints will these conditions be met, the ankle dorsi/plantar flexors and the jaw openers/closers being just two examples of joints with strong asymmetries. Compounding the effects of muscle asymmetries is the fact that the dynamical properties of shortening and lengthening muscles are very different from each other. Without those assumptions, these definitions of R, C and lambda are incompatible. As the control variables are presently defined, the C command does affect the EP of the joint because it will produce a change in net joint torque. While it may in principle be possible to address this by redefining the relationships between R/C and the lambdas, the computational problems this would raise appear daunting and physiologically implausible but the issue has never been addressed. Except under idealized conditions, R does not define an EP.

The most attractive feature of the lambda hypothesis is that it lets us conceptualize movement control in terms of positional concepts such as a virtual trajectory. This is simple, elegant and seductive. However, it ignores the issue that leads to our third criticism. Control is not effected by the construction of a virtual trajectory or R command alone but only by the added construction of the equally important C command as well as others that control the strength of feedback signals (e.g. m). Despite 35 years of development including the articles cited above, there are still no explicit, systematic rules for any variable other than R ([1], fig 3). The theory and the simulations give us little guidance as to how C or m change to move a single joint over different distances, at different speeds or with different external loads. The rules for C and m are neither obvious nor simple and it is impossible to define their behavior without explicit consideration of the dynamics of the moving system. Yet it is the claimed ability to describe control in positional terms or shifting "frames of reference," rather than in dynamical terms, that has made the equilibrium point hypothesis so attractive.

In light of the above, we suggest that it is not useful to think of controlling movement in terms of this proposed formulation of R, C and m commands. As yet the published record does not tell us in a reasonable way how this happens. This version of the equilibrium point hypothesis assumes parameters with values that are unphysiological, it inaccurately describes the effects of its control variables on movement except under implied and unphysiological assumptions, and by defining the behavior of only one of its control variables, it has obscured how incomplete a description it offers of movement control.