Working memory and human reasoning : an individual differences approach
Experiments 1-3 investigated the relationship between working memory and syllogistic and five-ten-n series spatial inference. A secondary aim was to replicate the findings of Shah and Miyake (1996) who suggested the use of separate central resources of working memory for spatial and verbal ability. The correlational analysis showed that the complex verbal and spatial working memory span tasks were associated together and consistently predicted reasoning performance in both verbal and visual modalities. The confirmatory factor analysis showed that three factors best accounted for the data -a verbal, a spatial, and a general resource. All the span tasks and most of the reasoning tasks significantly and consistently loaded the general factor. Experiments 4-6 investigated the relationship between working memory and a range of reasoning tasks - identified as either propositional. spatial, or quantifiable tasks. These experiments were based on the work of Stanovich and West (1998) who found that a range of reasoning tasks were predicted by cognitive ability and a reasoner's thinking style. The correlational anaylsis showed that the complex verbal and spatial working memory span tasks were associated together and consistently predicted reasoning perforinance. Two clusters of reasoning task emerged from the correlational analysis - one cluster related to the propositional and simple spatial reasoning tasks, whilst the other related to the quantifiable and complex spatial reasoning tasks. The confin-natory factor analysis showed that four factors best accounted for the data -a verbal, a spatial, a general, and a thinking style resource. All the span tasks and the reasoning tasks loaded the general factor, and most of the reasoning tasks further loaded the thinking disposition factor. These results are discussed in light of models of workino memory, theories of reasoning, and how to best characterise factor 3 (executive function) and factor 4 (thinking style) from tile factor analysis.