Abstract

Causal Status and Explanatory Goodness in Categorization Jason Jameson & Dedre Gentner Northwestern University Department of Psychology, 2029 Sheridan Road Evanston, IL 60208 USA psychology, the view that category knowledge comprises important explanatory information is not new; and our proposal takes off from the theory- or knowledge-based framework of conceptual structure (e.g., Carey, 1985; Keil, 1989; Murphy & Medin, 1985). We also draw on a specific instantiation of this view, the causal status hypothesis (Ahn, 1998; Ahn, et al., 2000). Murphy and Medin (1985) argued persuasively that categorization can be viewed as an inference to the best explanation (see also Rips, 1989). On the relationship between theoretical knowledge and centrality, Murphy and Medin defined the weight associated with features as “[d]etermined in part by importance in the underlying principles” (p. 298, 1985). This characterization is best viewed as a starting point, for it leaves open the critical question of how to define “importance.” One prominent answer to this question is the causal status hypothesis, according to which (1) the role that a feature plays in a concept—either as a cause or as an effect—partly determines its centrality; and (2) causal features are more important than effect features for categorization judgments: e.g., a potential category members’ possession of causal features is more diagnostic of category membership than possession of effect features. This hypothesis has received support in a variety of contexts and for several cognitive tasks (see Ahn & Kim, 2000, for a review). Our central claim is that the causal status effect derives from the supporting role that causal information plays in explanation. Although this idea is a long-standing theme in the study of categorization (e.g., Murphy & Medin, 1985; Rips, 1989), relatively little research has directly examined how explanatory goodness relates to categorization. In this paper, we study the relationship between explanatory goodness and categorization. We demonstrate that the magnitude of the causal status effect depends on the quality of the explanation in which the causal information is embedded. We first briefly review the causal status hypothesis and present arguments that the causal status effect depends on explanatory goodness. Then we present a study to support this claim. Finally, we discuss the implications of this study for categorization more broadly. Abstract Much research (e.g., Keil, 1989; Murphy & Medin, 1985; Rips, 1989) has emphasized the critical role that domain knowledge plays in categorization judgments. Recent instantiations of this view (e.g., Ahn, et al., 2000; Rehder & Hastie, 2001) have focused on characterizing how causal knowledge supports categorization decisions. We suggest that a more satisfactory account of categorization can be gained by considering the broader role that causal information plays in processes of explanation. An explanation-based perspective treats categorization as an inference to the best explanation (Murphy & Medin, 1985; Rips, 1989). This suggests that a critical source of constraints on categorization may come from a direct investigation of explanatory goodness. We present evidence that the causal status effect (e.g., Ahn, et al., 2000)—i.e., the phenomenon in which causes tend to be more heavily weighted than effects in categorization judgments— depends on the goodness of the explanation in which the causal information is embedded. Keywords: Categorization, explanatory goodness. causal status, explanation, Introduction Categorization processes are fundamental in making sense of the world. Not surprisingly, the study of categorization is of central importance in cognitive science; researchers have long sought to characterize the nature of the conceptual structures that best support classification decisions. There is general agreement that some parts of a concept are more important—more conceptually central— than others for categorization (e.g., Ahn, 1998; Ahn, et al., 2000; Medin & Shoben, 1988; Rehder & Hastie, 2001). For example, as Medin and Shoben (1988) demonstrated, the property of “curvedness” is more central for the category of boomerangs than for the category of bananas. That is, “straight bananas” are better examples of bananas than “straight boomerangs” are of boomerangs, because “curvedness” plays a more central role in the theoretical principles supporting one’s knowledge of boomerangs. We argue here that conceptual centrality is closely related to the goodness of an explanation: specifically, that the centrality of a conceptual component derives from its role in the best explanation of what holds category members together. On this view, a good conceptual structure is one that contains important explanatory information. The study of explanation spans several disciplines, ranging from philosophy and computer science (e.g., Thagard, 1989) to psychology (Read & Marcus-Newhall; Lombrozo, 2006; Lombrozo & Carey, 2006; Rips, 1989). In Conceptual centrality Intuitively, some components of a concept do indeed appear more important for categorization judgments. As Medin and Shoben (1988) point out, it is much easier to imagine a robin that is not red than it is to imagine a robin that lacks the appropriate genetic structure of robins: