Reasoning ability in subjects of high intelligence: An experimental study of individual differences with an appendix on Valentine's Reasoning Tests for higher levels of intelligence

In an opening review of the literature it is shown that two of the factors most widely supposed to produce error in syllogistic reasoning, namely, the 'atmosphere' of the premises and the emotional significance of the content of the argument, respectively could not, and have not been shown to, have this effect. Other factors liable to produce such errors are: misunderstanding of the universal affirmative (as implying the truth of its converse), a conflict between the truth-value of a conclusion and the validity of the corresponding argument as a whole, and an inadequate grasp of, or adherence to, the 'logical task'. In Chapter 2 the selection, on the basis of performance on two Forms of Section B of Valentine's Reasoning Tests, of a (PR) group of 22 'poor reasoners' and a (GR) group of 'good reasoners', matched one to one on a composite measure of their (superior) academic attainment and ability is described. Preliminary evidence relating to the difference between the groups is considered and the materials and procedure in the first, 'five types of statement' (5TS), experiment described. This experiment applies a form of Wason's 'card-taming task' to the three types of statement occurring most frequently in the criterion Valentine test, namely, the universal affirmative in categorical form (a), the universal negative (e) and statements of the form, 'Only X's are Y's', and to two other types incorporating logically important connectives, the universal affirmative in hypothetical form (Ah) and a universal-disjunctive (Ad).The assumption underlying the 5To experiment is that a difference between the groups in their grasp of any of the types of statement will be reflected in a difference in the success with which they tackle the task with respect to that statement-type. Significant differences were found on A and F and on one aspect of the response to Ah, but not on E or on Ad. A progressive analysis of responses to different statement-types suggests the operation of other, unsuspected, factors and throws doubt on the validity of the use of the card-turning task as a measure of the relative difficulty of different statement-types.In the later (4To) experiment, incorporating modifications to the materials and procedure, it was possible to establish (l) the persistence of the differences on A and F over a period of 12 to 18 months, (2) the superiority of the GR group in learning to make the correct response to A and (3) the significantly greater proneness of the PR group to a type of error which can only reflect a tendency to misunderstand A statements in the way described in the first paragraph. Other aspects of the task less immediately related to the difference between the groups are also considered. Chapter 4 closes with some examples, drawn from the history of psychology, of the sometimes serious practical consequences of the misunderstanding mentioned under (3) and a discussion of possible sources of this error. Chapter 5 describes the use of Wason's 'construction task' to investigate the possibility that PR subjects have significantly more difficulty than their GR counterparts with the negative particle, a very important element in all reasoning. The outcome is somewhat equivocal, there being a significant difference between the groups on the false- affirmative condition (which incorporates a single negative component, the 'semantic' notion of falsity) and when this condition is combined with the true-negative, but not on the false-negative which includes both negative components. This last result, it is suggested, may be due to the conspicuous instability of the response-time measure of difficulty in this case. Chapter 5 also considers evidence (from scores on the E.P.I.) about the extraversion and emotionality of the two groups, it having been argued that differences in either of these dimensions of personality might be partly responsible for difficulties with the negative particle. Although the differences are in the expected direction none is large enough to be significant on a two-tailed test. A small but significant correlation between scores on the N scale of the E.P.I. and latency of response to the true-negative condition when both groups of subjects are taken together lends, again in the absence of corroboration from the false- negative condition, rather uncertain support to the Eifermann hypothesis of a relationship between emotionality and slowness of response to negative statements,A concluding chapter reviews the outcomes of the above researches, considers an alternative interpretation of the 5TS results and indicates some possible points of departure for future research in this area