Table 1.

Computational representation of an internal model in terms of expected changes in the angular orientation of the EMG, as represented by a spatial function in hand centered coordinates, for each muscle

Model	Moment arm matrices (cm) J_m^T = (dλ/dθ)^T	Expected rotation in the muscle’s EMG function when fieldB₁ is learned (°)
Model	Moment arm matrices (cm) J_m^T = (dλ/dθ)^T	Biceps	Triceps	Ant del	Post del
A		−26.6	−26.5	−14.1	−13.6
B		−26.6	−26.5	−13.2	−13.0
C		−26.6	−26.6	−13.7	−13.6
D		−26.6	−26.6	−19.7	−20.2
E		−26.6	−26.5	−13.1	−16.6
F		−26.6	−26.5	−12.7	−15.2

Learning of an internal model is predicted to result in a specific rotation in each muscle EMG function. The magnitude of the predicted rotation depends on the muscle moment arms. Because of a lack of consensus regarding the magnitude of each muscle’s moment arm, we have considered six different models for J_m. In each case, λ is a vector of muscle lengths [biceps, triceps, anterior deltoid (Ant del), and posterior deltoid (Post del)], and θ is a vector of joint angles (shoulder and elbow). The consequence of learning an internal model of field B₁ should be a rotation in the resultant vector of each muscle’s EMG functions from that observed in the null field.