Individual differences in working memory and reasoning-remembering relationships in solving class-inclusion problems

Abstract

In the present experiment, we evaluated the effects of individual differences in reading span and variation in memory demands on class-inclusion performance. One hundred twenty college students whose reading spans ranged from low to medium to high (as indexed by a computerized version of the Daneman and Carpenter [1980] reading-span task) solved 48 class-inclusion problems. Half of the subjects had the solution information available when the problems were presented; the other half performed a detection task between solution information and problem presentation. The results from both standard statistical analyses and from a mathematical model indicated that differences in reading span and memory load had predictable, similar effects. Specifically, the sophistication of reasoning strategies declined when memory demands increased or when reading spans decreased. Surprisingly, these effects were primarily additive. The results were interpreted in terms of global resource models and fmdings from the developmental literature.In a recent series of experiments, we have used classinclusion tasks to study reasoning-remembering tradeoffs in the development of human problem solving across the lifespan (Howe & Rabinowitz, 1996;Rabinowitz, Howe, & Lawrence, 1989). In general, the class-inclusion task involves the presentation of two subclasses: a major subclass (e.g., there are six robins) and a minor subclass (e.g., there arefour swallows). Following this, subjects are asked an inclusion question involving the superordinate class (birds) and the major subclass (Are there more robins or more birds?). Additional questions can be asked involving the subclasses (Are there more robins or more swallows?) or the superordinate class and the minor subclass (Are there more swallows or more birds?).When solving class-inclusion problems, subjects can use a variety of strategies. The most appropriate reasoning strategy reflects an understanding that the numerosity of the superordinate class has to be greater than or equal to the numerosity ofthe subclasses. This is known as classinclusion reasoning. Alternatively, subjects can use an inappropriate strategy in which they treat the superordinate class as the subclass that is not specified in the problem. To illustrate, ifthe problem is a class-inclusion one ofthe form "Are there more robins or more birds?" and the subject treats "birds" as "swallows" (and remembers the rela-

Individual differences in semantic short-term memory capacity and reading comprehension

“…It is known that individual differences in WM predict the ability to solve verbal problems, such as syllogistic analogies (Gilhooly, Logie, Wetherick, & Wynn, 1993;Gilhooly, Logie, & Wynn, 1999) and classinclusion problems (Howe, Rabinowitz, & Powell, 1998). WM enables verbal problem solving not only by supporting the comprehension of the sentences that constitute a problem but also by providing a computational workspace in which the solution to a problem can be generated and tested (Carpenter, Just, & Shell, 1990;Greeno, 1973).…”

L'interprétation des discordances compétence-performance dans le développement cognitif. L'exemple de la quantification de l'inclusion

“…L'évolution des performances à l'épreuve demeure un fait développemental majeur, au point que la plupart des auteurs postpiagetiens testeront la capacité de leur modèle théorique à en rendre compte, sans jamais parvenir à l'appréhender totalement. À en juger par la diversité des hypothèses théoriques convoquées, cette épreuve semble avoir peu à peu pris statut de paradigme (Howe, Rabinowitz et Powell, 1998). Elle constitue, aujourd'hui encore, un cadre privilégié pour l'étude du raisonnement logique et des phénomènes de discordances qui caractérisent son développement.…”