Speed-Accuracy Tradeoffs of Aimed Movements

It should not be surprising that movements made at higher speed are less accurate – i.e., have larger endpoint variability or are more likely to miss a target. Perhaps more interesting, two different ways that speed and accuracy are related have been found. Our early research (Wright & Meyer, 1983) showed that the form of the tradeoff known as Fitts’ law holds for movements with a time-minimization constraint: i.e., rapid movements to a target with a clear extent (any movement ending with the target is acceptable, but any movement outside of its boundary is an error) – consider mouse movements during text editing. By contrast, the second, so called linear tradeoff holds for movements with a time-matching constraint – consider that, when playing music, it is an error either to play the wrong note (make an error of spatial position) or to play the correct note at the wrong time (temporal errors).

Our additional early research proposed both how the linear speed-accuracy tradeoff could emerge from variability in the motor system (Meyer, Smith, & Wright 1982) and the stochastic, optimized-submovement model, which explains how Fitts’ law emerges when time-minimized movements are produced using an initial movement followed, only if necessary, by corrective submovements (Meyer, Abrams, Kornblum, Wright, &  Smith, 1988; Meyer, Smith, Kornblum, Abrams, Wright, 1990).

More recently we have been working on two follow-ups to this early research. The first explores the stability of Fitts’ law across two variations of the standard task: comparing (a) when movements are made discretely or in continuous sequences and (b) when the movement distance and target size is blocked or varied across a set of movements. These variations are of practical interest because data collected for discrete movements with blocked movement distance and target size are often used to evaluate and compare the efficiency of input devices. However, the day-to-day use of these devices typically often involves continuous movements with variation in target distance and size. Our second recent line of research explores detailed implications of the stochastic, optimized-submovement model. Critical to this research are techniques we have been developing to decompose a complex movement into its component submovements.