The relation between latency and accuracy in the identification of visual stimuli. I. The effects of task difficulty

Abstract

Existing results concerning the relation between latency and accuracy are reviewed and it is concluded that when the stimuli are easily identified errors are faster than correct responses, and when the stimuli are difficult to identify errors are slower than correct responses. Theories are examined which explain the first type of error only or the second type of error only, but no theory encompasses both. The accumulator theory of A UDLEY and P IKE (1965) is taken as an example of a theory which explains the second type of error; it predicts that, if change in identifiability is assumed to affect the probability of obtaining a correct reading from the stimulus, the latency of such errors will increase as identifiability increases, while the latency of correct responses decreases. Experiment 1 used a difficult ten-choice identification task. Latency increased with the extent of the misjudgment, then declined on extreme misjudgments. Experiment 2 used a task with five stimuli and three responses. Latency again was greater on errors than on correct respinses. Correct responses to more discriminable stimuli were faster and incorrect responses to such stimuli were slower than those to less discriminable stimuli, as predicted by the accumulator theory. If identifiability was reduced by increasing luminance or the interval between the stimulus and a masking field, however, the latency of the correct responses was slightly reduced and that of incorrect responses was greatly reduced. This is explained in terms of the accumulator theory by postulating a higher value of the criterion for response when the stimulus is less ‘adequate’. An interaction between discriminability and adequacy in their effect on the latencies of correct and incorrect responses was also explicable in these terms.