Understanding steering behaviors in paths with non-linear width changes

Steering through width-changing paths is fundamental in graphical user interface (GUI) tasks such as lasso selection. Researchers in ergonomics have modelled movement times by path shape, but most studies examined only linear width changes. We investigated the generalisability of the Steering Law by measuring movement times in paths where the width changes according to exponential, quadratic, and sigmoid functions. The global steering model yielded R-squared values ranging from 0.9304 to 0.9934, but for quadratic paths, a model that approximated the width change as linear showed improved goodness-of-fit indices. This is presumably because, according to speed profile and cursor trajectory analyses, participants were unable to effectively utilise the bulging part of the path for acceleration. Our work helps to clarify the behaviours in GUI operations and also offers the possibility of model simplification in applications using the Steering Law, such as difficulty setting in racing video games.