The inverse kinematics problem deals with the question of how the nervous system coordinates movement to resolve redundancy, such as in the case of arm reaching movements where more degrees of freedom are available at the joint versus hand level. In particular, this work focuses on determining which coordinate frames can best represent human movements, allowing the motor system to solve the inverse kinematics problem in the presence of kinematic redundancies. In particular, in this work we used a multi-dimensional sparse source separation method called FADA to derive sets of basis functions (here called sources) for both the task and joint spaces, with joint space being represented in terms of either the absolute or anatomical joint angles. We assessed the similarities between the joint and task sources in each of these joint representations. We found that the time-dependent profiles of the absolute reference frame’s sources show greater similarity to those of the corresponding sources in the task space. This result was found to be statistically significant. Hence, our analysis suggests that the nervous system represents multi-joint arm movements using a limited number of basis functions, to allow for simple transformations between task and joint spaces. Importantly, joint space seems to be represented in terms of an absolute reference frame to achieve successful performance and simplify inverse kinematics transformations in the face of the existing kinematic redundancies. Further studies will be needed to determine the generalizability of this finding and its implications for neural control of movement.