The effect of structure on syllogistic reasoning

Abstract

The purpose of the present study was to investigate certain discrepancies in the error preference patterns of syllogistic reasoning in relation to the Atmosphere Effect and Caution principles established by Woodworth and Sells, and Sells. The discrepancies reported in later studies are of two types. (a) Absence of evidence for the Caution principle when syllogistic problems are presented in a Multiple-Choice format. (b) Anomalous high frequency of E Type conclusions in syllogisms involving a universal negative and a particular positive or negative as premises. Atmosphere Effect predicts a preference for O-type conclusions for these premises. Two hypotheses were tested. I. Caution predicted conclusions rank second in order preference, and consequently are not apparent when a Multiple-Choice format is used, since only first order choices are allowed in this form of presentation. II. The instructions used to remove the ambiguity of the word "some" in previous studies are incomplete and favor universal type of conclusions. The bis introduced is responsible for the anomalous increase of E conclusions, in problems where O conclusions are predicted by Atmosphere Effect. Two syllogistic tests were constructed. Test A consisted of 40 items presented in True-False format and intended to measure the logical ability of the subjects as well as the pattern of acceptance of simple conversions, for this purpose 16 of the items represented all the possible simple conversions from the four types of propositions. Test B consisted of 216 items presented in Paired-comparison, forced-choice form. Eight sets of premise-pairs were used, four related to hypothesis I and the other four to hypothesis II. Each set of premise-set was cast in the four syllogistic figures. Twenty-four items I Test B were also simple conversions but in paired-comparison format. Two groups of undergraduate students were used as subjects. Group I (N=90) received Test A and Test B in sequence. Group II (N=58) received in addition before Test B a set of "balanced" instructions in regard to the meaning of the word "some." Results supported both hypotheses. Caution predicted conclusions ranked second and with substantially higher frequencies than the other two-non-predicted conclusions. Hypothesis II was also supported by the results. When "balanced" instructions about "some" were given conclusion O was preferred as predicted by Atmosphere in the premise sets previously found to favor E conclusions. Furthermore, Group I that received no instructions at all about "some" showed the same results in accordance with Atmosphere Effect, suggestive that the "unbalanced" instructions used in previous studies were indeed biased and responsible for the anomalous preference of E type conclusions.