Fitts argued that if two S-R ensembles have the same value on the “population stereotype” scale, they would generate the same size S-R compatibility effects (i.e. RT difference between congruent and incongruent mapping). According to the DO model, this need not be the case: two ensembles could have the same value of population stereotype, but differ in their degree of dimensional overlap. In that case, because the compatibility effect is jointly determined at the set and at the element levels, the ensemble with the higher degree of dimensional overlap would generate the larger compatibility effect. This is the proposition that was tested in this study.
Three stimulus and three response sets were combined to forms nine S-R ensembles. The population stereotype and the level of dimensional overlap were obtained for each ensemble. The population stereotype was obtained by asking subjects to give what they thought would be the best mapping between elements of the stimulus set onto elements of the response set. The dimensional overlap was obtained by asking subjects to compare pairs of S-R ensembles, and chose that pair that had the better matching stimulus and response sets. Dimensional overlap was quite strong for some ensembles, and population stereotype was strong in others, thus providing the precise conditions needed to test our proposition.
These nine ensembles were used to construct choice RT tasks with two mapping instructions. One mapping was congruent, and corresponded to the population stereotype; the other was incongruent (and there were three different incongruent mappings).
The fastest mean RT was obtained for the congruent mapping, and as the degree of dimensional overlap decreased, the mean congruent RT increased. The RT for incongruent mapping was, of course, longer and it increased with increasing degrees of dimensional overlap. Both, the decrease in the case of the congruent mapping, and the increase in the case of the incongruent mapping, are in accord with the model.
Assuming that our measures of the level of the population stereotype, and the degree of dimensional overlap are valid, the results of this study confirm our proposition.