Computing Instantaneous Throughput For Spatially Constrained Aiming Tasks

Abstract Motor throughput is a standard measure of how rapidly information can be transmitted through an input device (such as a mouse or wand) by a human user carrying out point-to-point movement (also known as a Fitts' task.) As low-cost motion tracking systems become more prevalent, engineers are presented with increasing opportunities to control operations using hand gestures. However, accurate analysis of motor throughput in such cases may be difficult, as muscle tremors may prevent midair motions from coming to an absolute stop, and visual obstructions may prevent endpoint data from being reliably captured. A lack of accurate temporal and spatial endpoint information precludes any determination of motor throughput. Inspired by Fitts' information channel model, we derive a measure of instantaneous motor throughput. Observation of this instantaneous measure demonstrates how the traditional measure of information bandwidth, Fitts' motor throughput, emerges during the course of an aimed movement. Further, an experimental study demonstrates that instantaneous throughput (ITP) profiles exhibit a common shape when plotted against normalized distance. We select the moment of peak variance to be the most convenient landmark along the ITP trajectory for computing an overall throughput value, and demonstrate that this new method compares favorably with traditional throughput measures.