Academic journal article Memory & Cognition

Metacognition and Abstract Reasoning

Academic journal article Memory & Cognition

Metacognition and Abstract Reasoning

Article excerpt

Published online: 22 November 2014

© Psychonomic Society, Inc. 2014

Abstract The nature of people's meta-representations of deductive reasoning is critical to understanding how people control their own reasoning processes. We conducted two studies to examine whether people have a metacognitive representation of abstract validity and whether familiarity alone acts as a separate metacognitive cue. In Study 1, participants were asked to make a series of (1) abstract conditional inferences, (2) concrete conditional inferences with premises having many potential alternative antecedents and thus specifically conducive to the production of responses consistent with conditional logic, or (3) concrete problems with premises having relatively few potential alternative antecedents. Participants gave confidence ratings after each inference. Results show that confidence ratings were positively correlated with logical performance on abstract problems and concrete problems with many potential alternatives, but not with concrete problems with content less conducive to normative responses. Confidence ratings were higher with few alternatives than for abstract content. Study 2 used a generation of contrary-to-fact alternatives task to improve levels of abstract logical performance. The resulting increase in logical performance was mirrored by increases in mean confidence ratings. Results provide evidence for a metacognitive representation based on logical validity, and show that familiarity acts as a separate metacognitive cue.

Keywords Conditional reasoning . Metacognition . Abstract reasoning

An important question in the study of human reasoning is how people evaluate the adequacy of their own inferential processes. To date, much of the debate about people's logical abilities has focused on the kinds of cognitive processes that are used when making inferences, for example mental models (Johnson-Laird & Byrne, 1991, 2002) versus probabilistic models (Oaksford & Chater, 2003). In this paper, we shift focus to a different level of analysis, namely the nature of people's metacognitive skills and the processes that they use to evaluate the adequacy of their reasoning performance.

Metacognition and reasoning

Ackerman and Thompson (2014) have developed a framework for understanding metacognitive processes in the context of reasoning. They used the phrase "meta-reasoning" to refer to the processes that monitor and control reasoning, problem solving and decision-making. Meta-reasoning refers to second-order processes that monitor and control object level cognitive processes (c.f., Nelson & Narens 1990). In other words, the study of meta-reasoning is not necessarily concerned with inferential processes per se, but with processes that determine how satisfied one is with a conclusion, as well as those that determine how one allocates resources to a problem.

Indeed, it is important to distinguish between debates about the nature of inferential processes and debates about the processes involved in people's metacognitive representations. Specifically, the nature of people's underlying inferential processes determines the kind of information that is used to make specific inferences and potentially whether these inferences will be consistent with the norms of logical validity or not. However, the nature of these processes need not have any direct bearing upon the nature of metacognitive understanding. Indeed, in keeping with other metacognitive theorists, we assume that people do not necessarily have access or insight into their cognitive processes when making metacognitive judgments, and that, instead, such judgments are inferential and based on cues (See Dunlosky & Tauber, 2014; Koriat, 2007 for reviews). Consequently, from a meta-level perspective, the questions are not necessarily about inferential processes per se, but about the cues that are used to evaluate those inferences. From this perspective, one of the more important questions would be what determines whether reasoners are content with a given inference, which involves understanding the processes by which people evaluate their confidence in reasoning outcomes. …

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.