Sensorimotor delays in tracking may be compensated by negative feedback control of motion-extrapolated position.

Sensorimotor delays dictate that humans act on outdated perceptual information. As a result, continuous manual tracking of an unpredictable target incurs significant response delays. However, no such delays are observed for repeating targets such as the sinusoids. Findings of this kind have led researchers to claim that the nervous system constructs predictive, probabilistic models of the world. However, a more parsimonious explanation is that visual perception of a moving target position is systematically biased by its velocity. The resultant extrapolated position could be compared with the cursor position and the difference canceled by negative feedback control, compensating sensorimotor delays. The current study tested whether a position extrapolation model fit human tracking of sinusoid (predictable) and pseudorandom (less predictable) targets better than the non-biased position control model, Twenty-eight participants tracked these targets and the two computational models were fit to the data at 60 fixed loop delay values (simulating sensorimotor delays). We observed that pseudorandom targets were tracked with a significantly greater phase delay than sinusoid targets. For sinusoid targets, the position extrapolation model simulated tracking results more accurately for loop delays longer than 120 ms, thereby confirming its ability to compensate for sensorimotor delays. However, for pseudorandom targets, this advantage arose only after 300 ms, indicating that velocity information is unlikely to be exploited in this way during the tracking of less predictable targets. We conclude that negative feedback control of position is a parsimonious model for tracking pseudorandom targets and that negative feedback control of extrapolated position is a parsimonious model for tracking sinusoidal targets.

A systematic evaluation of the evidence for perceptual control theory in tracking studies.

Perceptual control theory (PCT) proposes that perceptual inputs are controlled to intentional 'reference' states by hierarchical negative feedback control, evidence for which comes from manual tracking experiments in humans. We reviewed these experiments to determine whether tracking is a process of perceptual control, and to assess the state-of-the-evidence for PCT. A systematic literature search was conducted of peer-review journal and book chapters in which tracking data were simulated with a PCT model (13 studies, 53 participants). We report a narrative review of these studies and a qualitative assessment of their methodological quality. We found evidence that individuals track to individual-specific endogenously-specified reference states and act against disturbances, and evidence that hierarchical PCT can simulate complex tracking. PCT's learning algorithm, reorganization, was not modelled. Limitations exist in the range of tracking conditions under which the PCT model has been tested. Future PCT research should apply the PCT methodology to identify control variables in real-world tasks and develop hierarchical PCT architectures for goal-oriented robotics to test the plausibility of PCT model-based action control.

Perceptual control models of pursuit manual tracking demonstrate individual specificity and parameter consistency.

Computational models that simulate individuals' movements in pursuit-tracking tasks have been used to elucidate mechanisms of human motor control. Whilst there is evidence that individuals demonstrate idiosyncratic control-tracking strategies, it remains unclear whether models can be sensitive to these idiosyncrasies. Perceptual control theory (PCT) provides a unique model architecture with an internally set reference value parameter, and can be optimized to fit an individual's tracking behavior. The current study investigated whether PCT models could show temporal stability and individual specificity over time. Twenty adults completed three blocks of 15 1-min, pursuit-tracking trials. Two blocks (training and post-training) were completed in one session and the third was completed after 1 week (follow-up). The target moved in a one-dimensional, pseudorandom pattern. PCT models were optimized to the training data using a least-mean-squares algorithm, and validated with data from post-training and follow-up. We found significant inter-individual variability (partial η2: .464-.697) and intra-individual consistency (Cronbach's α: .880-.976) in parameter estimates. Polynomial regression revealed that all model parameters, including the reference value parameter, contribute to simulation accuracy. Participants' tracking performances were significantly more accurately simulated by models developed from their own tracking data than by models developed from other participants' data. We conclude that PCT models can be optimized to simulate the performance of an individual and that the test-retest reliability of individual models is a necessary criterion for evaluating computational models of human performance.