Conversion errors in processing artificial set inclusions

Recent studies on processing linguistically-derived, artificial set inclusions have generally found poor performance on true-false verification tests. Results have indicated logical processing errors, mainly conversions (e.g. if told All As are Bs, then responding true to the test statement All Bs are As). The present experiment attempted to reduce the occurrence of these errors by providing repetitive study of and test on the same set inclusion structure and immediate feedback on each true-false test sentence. Conversion errors were reduced, but only for about one-half of the subjects. Reaction time profiles for these subjects indicated an integrated, holistic representation of the set inclusion information. These reaction time data were like those observed for artificial linear orderings.