On Motor Performance in Virtual 3D Object Manipulation

The Challenges in Modeling Human Performance in 3D Space with Fitts’ Law

“…Stoelen and Akin, however, limited their findings without any spatial arrangements since pointing to the target was performed only across one line and furthermore performed in 2D space without the use of 3D displays [77]. The latter was accounted for by Kulik et al [44], whereby 3D displays were used to model combined transitional and rotational movements. They found a surprisingly adequate linear fit of R 2 = 0.78 when combining both translation and rotation, defined in their case as "3D Docking" [44].…”

“…Unfortunately, most literature to date use this evaluation approach exclusively [6,14,44,59,77]. While it does indeed provide an overview of how well the model does, researchers would also be advised to report the values of constants and their respective standard error as shown in Figure 1.…”

“…Its applicability has been well demonstrated in 2D tasks [19,32,52], motivating multiple other researchers with their own 2D extensions, most notably that of Hoffmann's, Welford's and Shannon's [52,87,90]. More recently, the law itself has also been tested in 3D space [44,77] and providing a robust ground for various 3D extensions [5,14,59], with lesser but still impressive predictive powers.…”

“…However, to the best of our knowledge, this is not the case. The most popular methods to date have provided an invaluable insight as to which variables are integral when spatially describing 3D space [5,14,44,52,59,77,87,90]. Consequently, these methods were tested under very specific settings and limited to certain spatial arrangements, thus progress towards a true 3D performance metric is still scattered not allowing for inter-study comparisons.…”

“…While Fitts' model was originally intended for translational tasks only, other studies demonstrated that it can adequately model 2D rotational tasks as well [15,41,42], though not exhaustively tested in the 3D domain. Furthermore, combined translational and rotational movements are severely limited and only accounted by two studies [44,77], of which movements were limited only across one line, effectively disregarding spatial arrangements. Moreover, depthrelated distances in 3D displays were only accounted for and added to a model extension in one study [5].…”