A Proposed Resolution of Curious Conflicts in the Literature on Linear Syllogisms.

摘要

Students of reasoning have engaged in a vigorous debate regarding the representations and processes used by subjects solving linear syllogisms. Meaningful communication between proponents of the various positions has been hampered by the appearance of curious conflicts in reported data sets for the linear syllogism problems. The present experiment was intended to isolate the source of these conflicts in the literature. Eighteen adult subjects received linear syllogisms under instructions designed to yield speeds commensurate with error rates of about 10%. Latency and error data were analyzed both separately (via multiple regression) and jointly (via canonical regression). These data were also analyzed via pseudo-deadlines, according to which responses were counted as correct if they were correct and fell below a given pseudo-deadline, and were counted as erroneous if they were incorrect or fell above a given pseudo-deadline. The analyses revealed that the source of the conflicts in the literature is the failure of researchers to appreciate the complex interrelationships between latency and error rate. When these interrelationships are taken into account, the conflicts disappear. (Author)