1. KINEMATICS/KINETICS CORRELATIONS OF ARM MOTOR CONTROL DURING CORIOLIS PERTURBATIONS. A. Pierobon, S.B. Bortolami, J.R. Lackner*, P. DiZio. Ashton Graybiel Spatial Orientation Laboratory and Volen Center for Complex Systems, Brandeis University, Waltham, MA. Introduction The CNS generates feed-forward signals that control kinetics and kinematics of movements. Brain and spinal cord mechanisms generate torque adjustments in response to unexpected external loads, such as Coriolis forces. We measured the arm kinematics and computed torques with an inverse dynamics model. We used a variational approach to isolate the torque corrections and accumulated kinematics errors in response to the Coriolis perturbation. Conclusions & Discussion We discovered that during reaching in a Coriolis force field the neuromuscular correction measured in the first 200ms at the shoulder joint, is fully velocity dependent since the whole torque corrections follow the joint error rates and precede the joint position errors. In other words, the torque correction could not be a consequence of the joint errors given the causality of the physiological system. Torque variations precede position variations and follow velocity variations. Assuming that kinetics variations result from spinal control actions as a consequence of some kinematics variations, due to Coriolis-like perturbations, torque variations must be elicited by velocity variations. The statistical origin of cross-correlation analysis allows to assign a significance to the computation of lead/lag. Acknowledgements GRANT XY Materials and methods Enclosed 22ft diameter Slow Rotation Room (SRR), Optotrak tracking system. 30 subjects 5 rotation speeds (5,10,15,20,25 rpm) 3 subjects tested @ 5rpm, 6 subjects @ 10rpm, 7 subjects @ 15rpm, 7 subjects @ 20rpm, 7 subjects @ 25rpm PROCEDURE 2 pre-rotation blocks (8 reaching movements) 1 per-rotation reach that embedded the effects of Coriolis Forces on arm DATA PROCESSING Arm marker filter. For each subject, we calculated the baseline reach by averaging together, sample by sample, the repetitions within the last block of the pre-rotation reaches. The first per-rotation reach was a unique sample, which could only be acquired once within the same position frequency content was negligible above 20 Hz. Arm marker positions were numerically differentiated twice with a five-point central approximation formula to obtain marker velocities and accelerations and filtered before and after each differentiation with a seven point averaging experimental section. The first per-rotation reach embed the effects of unexpected Coriolis forces. Data analysis Variational approach: BASELINE = Feed-Fwd Signal + Servo Feed-Back CORIOLIS PERTURBED = Feed Fwd Signal + Servo Feed-Back +  Cross Correlation Analysis The computational origin of the cross-correlation function is similar to that of the correlation coefficient, although the former is not normally used as a tool for analyzing normally distributed series, but rather time variant, not normally distributed signals. Speculations about the sign and the magnitude of the maximum of the correlation function’s absolute value are not meaningful unless the source signals are normally distributed (generally not the case). It could be possible to define a “p-function”, computed with statistical analysis tools, which can be associated to the correlation function, computed as a signal analysis tool. Results 989.11 References p>0.05, and therefore the correlation function is not statistically meaningful The angle rate signal naturally leads the angle signal, being its derivative. This results in a cross correlation graph as two symmetrical maximums. Variations in the torque signal, at both shoulder and elbow, follow angule rate variations but precede angle variation. These differences are significant for the shoulder but not the elbow. The average duration of the baseline unperturbed reaching movement was 369 ± 48ms (SD), the average duration of the first perturbed reaching movement was 351 ± 90ms (SD). TORQUE variations ANGLE variationsANGLE RATE variations CONSEQUENTLY AR/T LEADA/T LAG AR/A LEAD The lead of the Angle Rate Variation signal and the lag of the Angle Variation signal, with respect to the Torque variation signal, sum up to the natural gap between the two signals independently of the Torque Variation Signal. Cross Correlation of the variation signals is used to evaluate possible I/O relationships of the arm reach controller. Cross-correlation function Correlation coefficient as a function of delay Torque correction are the response of the spinal servo to the kinematic variations, which are deviations from the desired trajectory due to the effects of the Coriolis forces acting on the arm.