ON A RELATIONSHIP BETWEEN EVALUATION OF SIMPLE UNIVERSAL PROPOSITIONS AND SEMANTIC MEMORY

Abstract

When a person judges whether a simple sentence such as “An S is a P”(S: subject, P: predicate) is true,(1)reaction time (RT) varies directly with category size of P if S is held constant (subset effect) and (2)semantic similarity between S and P speeds up positive decisions and slows down negative decisions (similarity effect). These two phenomena have been pointed out repeatedly in many studies of semantic memory. Traditionally, the experimental material used to examine these effects has almost been limited to universal affirmative propositions (UA's) and there seems to be no prior data as to universal negative propositions (UN's). Since everyday semantic judgements, however, are performed in UN form as frequently as in UA form. UN form cannot be ignored. Hence, the main purpose of the present study was to examine the effects of negation (“not”) on these two widely accepted phenomena using a standard sentence verification task.
The main results were as follows;
1. For both UA and UN, there appeared no evident ordinary subset effect, and reverse subset effect was rather dominant as a whole. These observations were interpreted in terms of the complementary set relations between stimulus categories appearing as P and were also considered from the viewpoint of a model similar to Meyer's two-stage model.
2. It was revealed that there existed the same pattern of ordinary similarity effect in UN as in UA. RT was greater in UN than in UA over the full range of semantic similarity. But the difference in RT between UA and UN was greater when S and P were in disjoint relation than when S was a subset of P. A tentative model which assumed differential amount of cognitive load conditional on dual predictive representations (“isa base” and “isn'ta base”) was proposed to account for this interaction.