We describe an alternative approach to the system identification analysis presented in (Krouchev et al., SFN Abstr. 26:1482, 2000) to interpret neuronal data from (Kalaska & Sergio, SFN Abstr. 23:1554, 1997). It is based on the neural network model in (Lukashin et al., Biol. Cybern. 74:469, 1996). We enhanced the model to account for time-varying kinetic and neuronal data in an isometric force-ramp paradigm. By analyzing the computability of equilibrium point (EP) conditions for arbitrary points in the workspace, we also provide for handling kinematic data in an arm movement paradigm. We are thus able to address: 1) The direct problem: Given an EP position, a direction of motor output and time-varying single unit activity, estimate generated forces at the hand. 2) The inverse problem: Given kinematics and kinetics, estimate neural activation. We compare the predictions of this model based on the population-coding hypothesis with experimental data from behaving primates and with the results of our previous ...