Models that apply observer theory to the brain's control of dynamic body motion. The general model (Merfeld et al. 1993) of active motor control (*A*) was simplified to a model for sensory estimation of passive motion (*B*). *A*: “desired motion” (*left*) is subtracted from the current motion to determine the motion required to reach that desired. Motor planning in the brain then determines the specific muscle activations required to create the motion. Muscle activity determines motion according to “body dynamics”–e.g., limb mass determines how quickly it moves. Deterministic motion is perturbed by “nondeterministic” disturbances (e.g., due to unstable ground), yielding total motion, which is sensed by organs such as the vestibular organs. Afferent signals from sensory organs are contaminated by noise, yielding the “noisy afferent measurement.” The “observer” (shaded area) estimates total motion using knowledge of the muscle commands received via “efference copy” and noisy information about total motion received via “noise afferent measurement.” The internal models of body dynamics are applied to the “efference copy of muscle commands” to estimate motion. “The internal model of the sensors” predicts the “expected measurement.” The difference between the actual and expected measurements can be due to disturbances, motion, measurement noise, and imperfect internal models. This difference, the “sensory conflict signal” (Oman 1982), steers the estimated state of the system toward the actual state. *B*: the structure of an observer model to estimate yaw angular velocity. A single sensory organ (the SCC) measures angular velocity, including process noise and other motion that is part of an experiment (“experimental external motion”). “Noisy measurement,” the SCC output with added measurement noise, inputs to the observer model and corresponds to SCC afferents. Both the process and measurement noises are Gaussian with zero mean and variances *Q* and *R*, respectively. The observer contains a model of the sensor, which estimates angular velocity. The sensory conflict between the estimated and actual measurements is multiplied by the observer gain and becomes the feedback that guides the internal model.

## PubMed Commons