Models of behaviors when detecting displacements of joints.

This report describes two models of human behavior when detecting displacements of joints that allow one to compare and integrate findings from different proprioception tests in a quantitative way. Results from various tests have led to different and often conflicting conclusions about proprioceptive behaviors and their underlying neural mechanisms. However, it has been impossible to compare data and conclusions in any meaningful way due to lack of a suitable analytical framework to accommodate important differences in procedures used in the various tests. These models can provide one such framework. The models, developed using data from proprioception tests reported in the literature, describe how the amplitude and velocity of joint excursions, and the subject bias expressed as false alarm rate, affect the detectability of displacements of joints. Two models were needed to represent observed behaviors: one based on velocity signals alone (the velocity model) and the other based on both velocity and positional signals (the displacement-velocity model). To simulate the detection-decision process subjects used to determine whether a joint was displaced, we adapted strategies from signal detection theory. The models characterized reported behaviors from disparate proprioception tests remarkably well, requiring only 3 degrees of freedom in the velocity case, and 4 in the displacement-velocity case.