A model of human operator behavior during pursuit manual tracking-what does it reveal?

Theoretical modeling results are used to provide insight into the origin of changes in human operator (HO) behavioral patterns during pursuit manual tracking of unpredictable targets of various bandwidths. A set of random ternary signals that elicit transient tracking behavior were utilized in the experiment. Autoregressive moving-average (ARMA) models, with two output lags and no input lags, provided the best linear least-square-error approximation to the input-output (target-HO response) data. Estimated model parameters reflected changes in HO dynamics with target characteristics. The estimated HO response delay decreased with increasing target bandwidth. A block-structured nonlinear model of the HO during pursuit manual tracking is proposed. The model is based on the observation that target directional reversals act as the main trigger for the HO to change tracking strategies. Assuming constant overdamped muscle dynamics, time variations in the tracking strategies used by the HO at target directional reversals are also inferred. The simulated HO response, using the inferred time-varying strategies, closely matches actual HO response