Academic journal article Memory & Cognition

Induction with Uncertain Categories: When Do People Consider the Category Alternatives?

Academic journal article Memory & Cognition

Induction with Uncertain Categories: When Do People Consider the Category Alternatives?

Article excerpt

These three experiments examined how people make property inferences about exemplars whose category membership is uncertain. Participants were shown two categories and a novel exemplar with a feature that indicated that the exemplar was more likely to belong to one category (target) than to the other (nontarget). Participants then made categorization decisions and property inferences about the novel exemplar. In some conditions, property inferences could be made only by considering both target and nontarget categories. In other conditions, predictions could be based on both categories or on the target category alone. Consistent with previous studies (e.g., Murphy & Ross, 1994, 2005), we found that many people made predictions based only on consideration of the target category. However, the prevalence of such single-category reasoning was greatly reduced by highlighting the costs of neglecting nontarget alternatives and by asking for inferences before categorization decisions. The results suggest that previous work may have exaggerated the prevalence of single-category reasoning and that people may be more flexible in their use of multiple categories in property inference than has been previously recognized.

(ProQuest: ... denotes formula omitted.)

Inductive inference involves generalizing information from known to novel exemplars. One powerful form of induction makes use of our understanding of categories as a basis for property inference. For example, if told that a lion has some novel property, people are likely to generalize this property to other members of similar categories, such as tigers, but less likely to generalize to dissimilar kinds, such as swans (Osherson, Smith, Wilkie, López, & Shafir, 1990; Rips, 1975; Sloman, 1993).

Most research on category-based induction has focused on cases where the category membership of the base or premise exemplar is known with certainty. There are many situations, however, where inductive inferences about an object are required before its category membership has been determined. Consider a person hiking through wilderness who hears rustling in the scrub near her feet. Although the hiker knows of many animals (e.g., snakes, small marsupials, birds) that could make the noise, she cannot be certain about the actual source. Nevertheless, she may wish to make an immediate prediction about how likely it is that the animal is dangerous. In a similar vein, after an initial examination a physician may have several possible diagnoses in mind about what is really troubling a patient. Before a final diagnosis is made, however, the physician may have to make some clinical inferences and suggestions to the patient about managing symptoms.

According to Bayesian models of categorization and inference (e.g., Anderson, 1991; Tenenbaum, Kemp, & Shafto, 2007), people should consider each of the uncertain category alternatives when making feature inferences. Anderson's Rational model, for example, proposes that people make property inferences about objects with uncertain category membership by estimating the probabilities of the predicted property for each category alternative, then combining these estimates and weighting each according to the likelihood of the object being in that category. More precisely, according to the Rational model, the prediction that an object with observed features F has an unobserved feature j involves the computation of a weighted sum across each category k:

... (1)

So in our hiking example, the conditional probabilities of the animal being dangerous would be estimated for each of the candidate categories of snakes, birds, and so on, and these would be weighted according to the likelihood that the animal comes from each category. Note that in Equation 1, the probability of the predicted feature j is assumed to be conditionally independent of the observed features F in each category. In other words, the Rational model assumes that people assess the frequencies of the given and predicted features independently, ignoring co-occurrence of feature pairs or higher order combinations of features. …

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

Oops!

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.