The widely-held view that Fitts' law expresses a speed/accuracy trade-off is presumably correct, but it is vague. We outline a simple resource-allocation theory of Fitts' law in which movement time and error trade for each other. The theory accounts quite accurately for the data of Fitts' (1954) seminal study, as well as some fresh data of our own. In both data sets we found the time/error trade-off to obey a power law. Our data, which we could analyze more thoroughly than Fitts', are consistent with a squareroot function with a single adjustable constant. We suggest that the resource-allocation framework should help combine information and energy considerations to allow a more complete account of Fitts' law.

No abstract