However, there are serious problems with the modal view. Some investigators questioned the evidence that automatic processing is free of attentional limitations (e.g., Cheng, 1985; Ryan, 1983). Others found evidence of attentional limitations in tasks that are thought to be performed automatically (e.g., Hoffman, Nelson, & Houck, 1983; Kahneman & Chajzyck, 1983; Paap & Ogden, 1981; Regan, 1981). The single-capacity view of attention, in which the modal view is articulated, has been seriously challenged by multiple-resource theories, which argue that many resources other than attention may limit performance (e.g., Navon & Gopher, 1979; Wickens, 1984)? Others argued that performance may not be limited by any resources, attentional or otherwise (e.g., Allport, 1980; Navon, 1984; Neisser, 1976). Moreover, there is growing dissatisfaction with the idea that automatization reflects the gradual withdrawal of attention (e.g., Hirst, Spelke, Reaves, Caharack, & Neisser, 1980; Kolers, 1975; Spelke, Hirst, & Neisser, 1976). Critics argue that the idea is empty unless the learning mechanism can be specified.

The purpose of this article is to propose a theory of automaticity that describes the nature of automatic processing and says how it may be acquired without invoking the single-capacity theory of attention or the idea of resource limitations. The theory is first described generally, then a specific version of the theory is developed to account for the speed-up and reduction in variability that accompany automatization. The theory is then fitted to data from two different tasks--lexical decision and alphabet arithmetic--and experiments that test the learning assumptions of the theory are reported. Finally, the qualitative properties of automaticity are discussed in detail, implications of the theory are developed and discussed, and the theory is contrasted with existing theories of skill acquisition and automatization. Automaticity as Memory Retrieval

The theory relates automaticity to memorial aspects of attention rather than resource limitations. It construes automaticity as a memory phenomenon, governed by the theoretical and empirical principles that govern memory. Automaticity is memory retrieval: Performance is automatic when it is based on singlestep direct-access retrieval of past solutions from memory. The theory assumes that novices begin with a general algorithm that is sufficient to perform the task. As they gain experience, they learn specific solutions to specific problems, which they retrieve when they encounter the same problems again. Then, they can respond with the solution retrieved from memory or the one computed by the algorithm. At some point, they may gain enough experience to respond with a solution from memory on every trial and abandon the algorithm entirely. At that point, their performance is automatic. 2 Automatization reflects a transition from algorithm-based performance to memory-based performance.

The idea behind the theory is well illustrated in children's acquisition of simple arithmetic. Initially, children learn to add single-digit numbers by counting (i.e., incrementing a counter by one for each unit of each addend), a slow and laborious process, but one that guarantees correct answers, if applied properly. With experience, however, children learn by rote the sums of all pairs of single digits, and rely on memory retrieval rather than counting (Ashcraft, 1982; Siegler, 1987; Zbrodoff, 1979). Once memory becomes sufficiently reliable, they rely on memory entirely, reformulating more complex problems so that they can be solved by memory retrieval.