The Case for Implicit Category Learning
Smith, Edward E., Cognitive, Affective and Behavioral Neuroscience
This article evaluates the evidence regarding the claim that people can learn a novel category implicitly-that is, by an implicit memory system that is qualitatively different from an explicit system. The evidence that is considered is based on the prototype extraction task, in which participants are first exposed to a set of category exemplars under incidental learning instructions and are then required to categorize novel test items. Knowlton and Squire (1993) first reported that memory-impaired patients performed normally on the prototype extraction task while being impaired on a comparable recognition task. Several studies have replicated these results, but other articles have criticized the evidence for implicit category learning on both methodological and theoretical grounds. In this article, we consider five of these criticisms-for example, that the normal performance of the patients is due to intact working memory mechanisms (see, e.g., Palmeri & Flannery, 1999) or to the lesser cognitive demands of prototype extraction rather than recognition (e.g., Nosofsky & Zaki, 1998). For each of the five criticisms, we offer counterevidence that supports implicit category learning.
Psychologists and neuroscientists began to recognize the existence of an implicit memory system during the 1960s and early 1970s when they discovered that a medial temporal lobe (MTL) amnesic, H.M., could learn a new perceptual motor skill as readily as neurologically intact participants. Soon after, researchers discovered that MTL patients showed intact perceptual priming, which indicated that implicit memory is not confined to motor skills (see, e.g., Schacter, 1987, and Squire, 2004, for reviews of these early developments). These findings led to a characterization of implicit memory as a system that can use past experience without intention or awareness and that does not depend on MTL.
In 1993, Knowlton and Squire published the first article claiming to demonstrate that memory-impaired patients had an intact ability to learn a novel category. This is known as implicit category learning. These findings were important because they demonstrated a kind of implicit learning that extracts common features and supports generalizations to novel items, which seemed to go beyond perceptual priming or other known forms of implicit learning. However, in recent years, the evidence for implicit category learning has come under sharp criticism, primarily from researchers who have focused on behavioral and mathematical analyses of category learning (e.g., Nosofsky & Zaki, 1998; Zaki & Nosofsky, 2001 ). The purpose of this article is to review and evaluate the evidence for and against postulating an implicit category-learning system, particularly in light of recent findings on this topic (e.g., Bozoki, Grossman, & Smith, 2006; Reber, Gitelman, Parrish, & Mesulam, 2003). The argument will be made that these recent articles provide new evidence that (1) people can indeed learn a category implicitly, (2) patients with compromised MTL function learn roughly as well as normals, and (3) the learning involved may be a kind of perceptual fluency.
Initial Evidence for Implicit Category Learning
In their seminal study, Knowlton and Squire (1993) used a prototype extraction task. Both memory-impaired patients and normal controls were presented a series of 40 dot patterns (see Figure IA for examples). All of the patterns had been created by starting with a prototype pattern and then transforming it to some degree by moving some proportion of the dots (after Posner & Keele, 1968). During training, nothing was mentioned about a category, since both patients and controls were instructed simply to point to the center dot in each pattern. After training, all participants were informed that the patterns they had just seen "all belonged to a single category" and that they now had to determine which of a sequence of test patterns also belonged to that category. …