Recognition of the stimulus degradation in the word frequency effect

Summary An experiment was designed to distinguish between two explanations of the word frequency effect, each accounting for the effect in terms of response bias. One explanation assumes that the bias affects a viewer's percept, the other a decision about what to report. Subjects were shown common and uncommon words, degraded to various degrees. Their task was to state what word was presented and to estimate its degree of degradation. A word frequency effect was demonstrated: the degree of actual degradation at which some arbitrary proportion of common words was correctly reported was greater than that at which the same proportion of uncommon words was correctly reported. In addition, subjects correctly assessed that the common words correctly reported were more degraded. The result was discussed with reference to the two versions of response bias theory.The discussion in this paper was materially assisted by comments made by Donald Broadbent, whose help is gratefully acknowledged.