Effect of Prior Knowledge of Signal Probabilities on Vigilance Performance at a Two-Signal Task

Abstract

Forty-eight subjects each performed for 1 hr on a visual vigilance task with two kinds of signal which differed in conspicuity and also in probability of occurrence. Prior to testing, 24 subjects were explicitly informed of the signal probabilities (0.05 and 0.15); the remaining 24 were not. Results indicate that the prior knowledge tends to eliminate or reduce the extent of the within-session criterion shift and, hence, to stabilize performance over the session. Even with this multiple-signal task, the findings offer support for the hypothesis that, in vigilance, the familiar within-session decline is primarily dependent on an initial disparity between the rate at which signals are presented and the rate with which affirmative responses are made and that subjects adapt their rate of affirmative responding to reduce this disparity. As predicted, the initial disparity tended to be greater for the lower signal probability and to be reduced by the prior information. By the end of the session, the aggregate affirmative response rate closely matched the aggregate signal rate of 0.20 (0.05 + 0.15), but there was some indication that subjects were unable to maintain the objective separation between the two signal rates, so that the differential response rates were more similar than the matching hypothesis predicts.