|
Article Excerpt
Reasoning is a fundamental cognitive activity, implicit in all aspects of everyday problem solving. Recent research has suggested that individual differences exist in the strategies people use when solving reasoning problems. For instance, Bacon, Handley, and Newstead (2003) and Ford (1995) have suggested that while some people prefer to represent and manipulate information in a visuospatial way, others are more likely to use a verbal-propositional strategy. Both these studies involved reasoning with syllogisms, logical arguments comprising two premises and a conclusion, for example:
Premise 1 All teachers are psychologists Premise 2 Some writers are teachers Conclusion Some writers are psychologists
Each syllogism contains three terms (teachers, psychologists and writers in the above example), one of which is common to both premises (in this case, teachers). Each premise contains one of four possible quantifiers (either All, Some, None or Some ... not) which describe the relationship between terms. The classic syllogistic inference is to determine, from the information given in the two premises, the one relationship which is not explicitly stated, that between the two other terms (e.g. psychologists and writers) which forms the conclusion (i.e. Some writers are psychologists). Such problems encapsulate many aspects of everyday reasoning such as deciding what conclusion, if any, can be drawn from assumptions about category membership, using stored knowledge and evaluating arguments. See also Evans, Newstead, and Byrne (1993) and Garnham and Oakhill (1994) for further details of syllogistic reasoning.
Johnson-Laird and Byrne (1991) identified 27 syllogistic forms with logically valid conclusions. Bacon et al. (2003) presented participants with the premises of each of these valid syllogisms and collected verbal and written protocol data as evidence of the strategies adopted while generating conclusions. Figure 1 illustrates some typical examples of verbal and spatial protocols collected by Bacon et al. and summarizes the main differences between strategies.
Individuals they termed verbal reasoners seemed to manipulate the information in its abstract form, frequently simply swapping around the terms in the premises. This substitution strategy involved obtaining a value for the common term from the first universal (All) premise encountered and simply substituting that value for the common term in the other premise to reach a conclusion. Other verbal reasoners used a rule which defined conclusions as associated with particular combinations of quantifiers, for instance, All + None = a None conclusion. As conclusions reached by these methods necessarily contain the same quantifier as at least one of the premises, verbal reasoners tended therefore to have most difficulty with problems where the logical conclusion had a different quantifier to either premise. Their verbal reports frequently referred to actions such as replacing, substituting and cancelling syllogistic terms.
Spatial reasoners, on the other hand, produced protocols that showed a more explicit representation of the relationships between terms. Their written protocols showed terms within shapes (usually circles) placed in differing spatial relationships to represent the relationship of the terms within and between the premises. Verbal protocols often described the terms, and their interrelationships, as groups or subsets. Spatial reasoners first construct a representation of the relationship between the first two terms (premise 1) and then augment this by adding information from premise 2 regarding the third term. Although these diagrams resemble traditional Euler circles (e.g. Erickson, 1978; Stenning & Oberlander, 1995) there is little evidence that they attempt to represent all logical possibilities as Euler circle theory would propose. The syllogisms which have more possible representations also happen to be those with different-form conclusions. Hence, similar to verbal reasoners, spatial reasoners also have most difficulty with these problems, but for different reasons.
These differences in strategy are not just limited to syllogistic reasoning. Recent research (Bacon, Handley, & Newstead, 2004) has shown that they extend to reasoning on transitive inference tasks. Transitive inference is a form of relational reasoning in which terms can be ordered in a single dimension, for instance a scale where Tom is stronger than Harry and Harry is stronger than John. Hence, Tom (the strongest) must be stronger then John (the least strong)-recognition of this relationship can be described as a transitive inference. Such inferences have usually been studied in the form of three-term series problems which allow terms to be arranged in a linear sequence according to their relative properties. The above example may be expressed in terms of the following three-term series problems:
Tom is stronger than Harry or, Harry is not as strong as Tom Harry is stronger than John John is weaker than Harry
Both describe the same array of terms unambiguously. Experimental tasks typically present such pairs of premises and ask participants to determine the relationship between two particular terms. Bacon et al. presented similar problems containing a variety of relational properties and with abstract terms represented by single letters. They found that people who used a spatial strategy for syllogisms (as Figure 1) seem to require a more explicit visual image of the meaning of the premises in order to generate a conclusion to three-term series problems. Verbal reasoners are able to work with material in the abstract propositional form in which it was presented.
Furthermore, Bacon et al. (2004) present some evidence to suggest differences between the two strategies in terms of how they draw on working memory resources. They report a confirmatory factor analysis which suggests that while verbal reasoners draw almost entirely on verbal working memory resource (the phonological loop), spatial reasoners draw significantly, and to a...
|
