A feedback information-theoretic transmission scheme (FITTS) for modeling trajectory variability in aimed movements

Optimal Feedback Control for Modeling Human-Computer Interaction

“…Under signal-dependent noise, the application of smaller controls in the second half of the movement (during the correction phase, which follows the surge) results in a decreasing movement variability towards the end of the movement. This is in accordance with the two-phase positional variance profiles that are typically observed in aimed movements [47,51,73]. From an informationtheoretic perspective, the decaying rate of these profiles can be explained by a user-specific channel capacity [47].…”

“…This is in accordance with the two-phase positional variance profiles that are typically observed in aimed movements [47,51,73]. From an informationtheoretic perspective, the decaying rate of these profiles can be explained by a user-specific channel capacity [47]. In the considered LQG model, the idea of user-specific control and observation noise levels affecting the (expected) end-effector variability provides a different interpretation.…”

Positional Variance Profiles (PVPs): A New Take on the Speed-Accuracy Trade-off

“…1. In this work (red), we transpose our behavior model [16] called the Variance Model, to propose a novel 2D random dualminimization protocol, as well as the 2D PVP method, which extracts features from the standard deviation of a set of trajectories. When evaluating human performance, this provides an alternative to classical Fitts' law experiments, which use time-minimization protocols, and endpoint information only.…”

“…In this work, we build on our recent information-theoretic transmission model with feedback, which we call the Variance Model [16], and the corresponding data analysis method called the method of PVPs (Positional Variance Profiles) to provide an alternative to Fitts' law experiments for input performance evaluations. The method of PVPs is a data aggregation method, where a time series of the variance of a set of trajectories is computed to obtain a variance profile.…”

On Computer Mouse Pointing Model Online Identification and Endpoint Prediction

“…A different view on modeling worth mentioning is through the information theory. In the recent work [25], the two-phase process is viewed as a communication problem where the information is transmitted from a source (initial/current point) to a destination (endpoint) over a so-called channel, perturbed by Gaussian noise with a presence of feedback (in a second phase). Despite the diversity of approaches for modeling the pointing task, we believe that searching for a simple, and at the same time, an accurate model is yet a relevant research direction.…”