An experimental comparison of conservative versus optimal collision avoidance warning system thresholds

In the distributed signal detection theoretic (DSDT) model, the human opera tor and the warning mechanism are independent decision makers who work toge ther as a team. The DSDT demonstrates that the optimal warning threshold, i n general, differs from the signal detection theoretic (SDT) threshold, whi ch assumes a single decision maker. This prediction was tested in an experi ment where drivers received monetary rewards for making safe passing decisi ons on a driving simulator. The experiment focused on evaluating the qualit y of the decision making of the drivers, and not on perceptual issues. A co llision avoidance system provided a warning when the probability of an inad equate overtaking gap exceeded a threshold. Three thresholds were tested. T he control threshold resulted in no detections or false alarms. The DSDT th reshold resulted in some misses but no false alarms. The SDT threshold resu lted in no misses but frequent false alarms. As predicted, (1) drivers perf ormed the best when the warning system used the DSDT threshold, and (2) use of the SDT threshold improved performance over the control threshold, even though four of the 10 drivers occasionally ignored the warning and made ri sky passing attempts in the SDT conditions, possibly because of earlier fal se alarms. These findings support the conclusion that the DSDT model is a u seful, quantitative tool that should be used by warning designers. (C) 2000 Elsevier Science Ltd. All rights reserved.