Academic journal article Memory & Cognition

Theory-Based Categorization under Speeded Conditions

Academic journal article Memory & Cognition

Theory-Based Categorization under Speeded Conditions

Article excerpt

It is widely accepted that similarity influences rapid categorization, whereas theories can influence only more leisurely category judgments. In contrast, we argue that it is not the type of knowledge used that determines categorization speed, but rather the complexity of the categorization processes. In two experiments, participants learned four categories of items, each consisting of three causally related features. Participants gave more weight to cause features than to effect features, even under speeded response conditions. Furthermore, the time required to make judgments was equivalent, regardless of whether participants were using causal knowledge or base-rate information. We argue that both causal knowledge and base-rate information, once precompiled during learning, can be used at roughly the same speeds during categorization, thus demonstrating an important parallel between these two types of knowledge.

Previous work on categorization has frequently focused on how people use similarity (i.e., featural overlap between instances and category representations; e.g., Nosofsky, 1986; Rosch & Mervis, 1975) or naive theories (i.e., general knowledge about how the world works; e.g., Murphy & Medin, 1985) to make category judgments. Typically, similarity use has been characterized as fast, automatic, and generally primary, whereas theory use has been characterized as slow, deliberate, and capable of overriding similarity only under particularly reflective situations (e.g., McRae, 2005; Sloman, 1996; see Kahneman, 2003, for a more general summary of this division). However, recent evidence has suggested that theory-based categorization can be fast (e.g., Lin & Murphy, 1997; Palmeri & Blalock, 2000).

Understanding when and why theory-based categorization can be fast allows us to place constraints on its underlying mechanisms. Our main theoretical assumption is that differences in the speed of categorization are due to differences in the complexity of the underlying processes, rather than differences in the type of information (i.e., similarity vs. theory). Within this framework, the present study examines whether the effect of theories can be observed during rapid categorization. In this introduction, we first review previous work that examined fast and slow categorization using the traditional distinction of similarity versus theory. We then present two possible ways in which theory-based feature weighting can take place, one of which suggests simple, and therefore fast, processing.

Slow and Fast Use of Similarity and Theory Information

Evidence for similarity's automaticity comes from work showing that similarity is often used in situations in which it should not be. For example, Centner and Toupin (1986) showed that participants would often use nonoptimal problem-solving strategies when problems shared superficial properties. Egeth (1966) showed that properties of pairs of stimuli were more rapidly identified as matching if other, irrelevant properties also matched. Similarly, Brooks, Norman, and Alien (1991) showed that participants were strongly influenced by feature overlap even when given an explicit classification rule. In contrast, theory use has been characterized as slow and deliberate. For example, Smith and Sloman (1994) failed to replicate a previously reported demonstration of theory use (Rips, 1989) simply by requiring participants to respond as quickly as possible. Only when participants were instructed to talk aloud while categorizing did the original, theory-based results reemerge. Similarly, Baraffand Coley (2003, described in Coley, Shafto, Stepanova, & Baraff, 2005) found that, with category-based induction, experts tended to rely more on specific domain theories than did novices. However, when placed under time pressure, even experts relied more on taxonomic similarity.

At a more theoretical level, Sloman (1996) construed theory-based categorization as a type of rule-based reasoning. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.