ABSTRACT
Classification experiments, those in which one stimulus from a set of possible stimuli is presented on each trial, are considered for the special case of stimuli lying only on one perceptual dimension. Two-response classification is a generalization of the yes-no experiment to many stimuli. Values of sensitivity, d′, for any two adjacent stimuli can be found by subtracting z-transformed response proportions. Bias is measured by criterion location. Cumulative sensitivity is the d′ distance between any stimulus and the endpoint stimulus and total sensitivity is the sum of all adjacent values of d′. When each stimulus to be judged is preceded by a standard, calculation of sensitivity and bias is unchanged. The psychometric function gives the relationship between cumulative sensitivity and stimulus value. The midpoint of this function is called the (empirical) threshold or point of subjective equality; the inverse of the slope of the function is a sensitivity measure, the just noticeable difference. For the same stimulus set, discrimination is superior to classification. One model for this effect attributes the discrepancy to the need for context memory in classification but not in discrimination.
