On rates of improvement with practice

A survey of practice effects for several sensory-motor tasks shows that no one linear equation relating performance time to practice gives a good description of them all. However, most of the data are well described in terms of equations derived from signal-detection theory, which postulate that, during each trial, activation of a correct response has to be made to an extent that it is discriminable from others. The activation is built up in part during the trial concerned and in part from a carry over from previous trials. In some cases, practice effects are proportional to the time taken by the first trial, but in others they are not. The question is considered of whether effects should be treated in terms of trials or of time. The extra information gained by comparing the regressions for two or more conditions in a task or for groups of subjects, rather than looking at one alone, is noted.