Memorability and the Rate of Learning

The speed with which the memory process comes to dominate the algorithm has important implications for studies of automaticity: It suggests that automatization can occur very quickly (also see Logan, 1988; Naveh-Benjamin & Jonides, 1984; E. Smith & Lerner, 1986). This means that it is feasible to study automatization in a single session, as was done in some of the experiments reported in subsequent sections of this article.

An important question raised by the instance theory is why automatization takes so long in other experiments. In search studies, for example, several thousand trials spanning 10 to 20 sessions are often necessary to produce automaticity (e.g., Shiffrin & Schneider, 1977). The discrepancy may arise for at least three reasons: First, the criterion for automaticity is different in the previous studies. In Shiffrin and Schneider's search studies, automatization was not considered to be complete until the slope of the search function reached zero (but see Cheng, 1985; Ryan, 1983). From the present perspective, automatization may never be complete, in that each additional instance will have some effect on memory, even if its effect does not appear in the primary performance data (also see Logan, 1985b). The present perspective also suggests there may be a shift in the direction of automaticity after only a few trials, and this shift may be a more important phenomenon to study than the zero slope addressed by Shiffrin, Schneider, and others (also see Logan, 1979, 1985b).

Second, differences in the apparent rate of automatization may be artifacts of the way the data are plotted. Instance theory argues that means and standard deviations should be plotted as a function of the number of trials per stimulus because each trial potentially adds a new instance to memory. But data are usually plotted against sessions, disregarding the number of trials per stimulus. Consequently, a task with more stimuli per session (and thus fewer trials per stimulus) will appear to be learned more slowly than a task with fewer stimuli per session (and thus more trials per stimulus), even if the rate of learning per stimulus is equal.