Latency operating characteristic: III. Temporal uncertainty effects.

Abstract

The purpose of this study was to determine the effect of temporal uncertainty on the latency operating characteristic (LOG), denned as the trade-off between speed and accuracy. Foreperiods were either constant (1,000 msec.), one of two equally likely but unnoticeably different durations (975 and 1,025 msec.), or one of two equally likely and noticeably different durations (750 and 1,250 msec.). Mean RT increased with temporal uncertainty, but this was not always so for the LOC. Most efficient performance was obtained when the foreperiod was 975 msec., and poorest performance occurred in the 1,250-msec. condition. Better performance with the early stimulus (at 975 msec.) is contrary to Anticipation and Preparatory Set models for the effects of temporal uncertainty, but is consistent with a Diluted Sample model in which temporal uncertainty affects the temporal integration of sensory data. Reaction time (RT) increases with temporal uncertainty. Increasing either the variability (e.g., Karlin, 1959) or the length (Posner & Boies, 1971; Woodrow, 1914) of the foreperiod between warning signal and stimulus reliably produces longer RTs. The oldest and most widely accepted explanation for this phenomenon is that RT is inversely related to 5s preparedness for perceiving and responding to the stimulus, and this preparedness presumably decreases with S's expectancy for the stimulus presentation. The explanation will be termed the Preparatory Set model. Following the rationale elaborated by Thomas (1967), the S's expectancy will be represented as a hazard function, i.e., the subjective probability that the stimulus will be presented at time t given that it has not yet been presented. For the present experiment, in which the foreperiod is selected with equal probability from a small set of alternatives, the principal characteristic of this representation is that S's expectancy is an increasing function of time from the onset of the trial. Prepara1 Preparation of this report was partially supported by Grant MH-21105 from the National Institute of Mental Health. The results were also presented at the Midwestern Conference on Mathematical and Theoretical Psychology in Bloomington, Indiana, April 1972.