Speed and Accuracy in the Processing of False Statements About Semantic Information

Data from eight experiments on semantic verification are presented. Two procedures were used, a standard reaction time procedure and a response signal procedure. Data from the response signal procedure showed that for false category-member statements (e.g., all birds are robins), there was an increasing tendency to respond yes early in processing, replaced later in processing by an increasing tendency to respond no. For statements involving antonym relationships (all mothers are fathers), data from the response signal procedure showed that there is no greater tendency to respond yes than for anomalous statements (a chair is a wall). Results in the standard reaction time procedure could not be predicted from the results in the response signal procedure. This suggests that simple models of the semantic verification task that assume a single yes/no dimension on which discrimination is made are not correct. Some suggestions are made as to the kinds of properties that would be required for an adequate model.