Human Sensitivity to Variability Information in Detection Decisions

Abstract

Three studies investigated the ability of observers to extract and use reliability information from graphical elements presented on a visual display. The first experiment assessed observers' accuracy in detecting differences in source reliability based on differences in the temporal variability of the sources. Observers' accuracy improved as the sample size and the difference in the underlying source variability increased. The second study investigated observers' efficiency in using the information about source variability to weight individual sources in forming a decision. Observers were able to differentially weight sources based on the underlying source variability. However, observers' adjustments of the weights tended to be less than ideal, particularly when reliable sources were not spatially contiguous and additional aids (e.g., observation periods and highlights) were unavailable to help observers identify differences in source variability. Finally, evidence in the third study indicated that information about differences in source variability reduced uncertainty about the locations of reliable sources but did not help observers to correctly adjust weights to match the ideal. In application, display enhancements (e.g., highlights) aid decision making when only general differences in reliability need to be recognized, but more advanced decision support is recommended when reliability information has to be incorporated into a decision statistic.

Effects of Warning Validity and Proximity on Responses to Warnings

“…In all conditions, and for both green and red warnings, participants used higher-than-optimal cutoffs that risked high miss rates. The failure to reach optimal cutoffs and the continued response to nonvalid warnings are in line with previous results that showed that participants failed to weigh information sources optimally, although they differentiated between the weights they gave to information from different sources (e.g., Montgomery, 1999;Montgomery & Sorkin, 1996).…”

Benchmarking Aided Decision Making in a Signal Detection Task

“…Given these heavy demands, the failure to match the performance of the OW model is unsurprising; researchers have long recognized that limits on information and information-processing abilities place bounds on human cognition that can prevent human decision makers from reaching putatively normative performance (Simon, 1955; Tversky & Kahneman, 1974). Nonetheless, human decision makers can at least approximate the performance of a linear cue combination rule (Einhorn, Kleinmuntz, & Kleinmuntz, 1979), and although they tend not to weight cues optimally (e.g., Johnson, Cavanagh, Spooner, & Samet, 1973; Montgomery, 1999, 2001; Montgomery & Sorkin, 1996), their deviations from normative weighting are likely to have modest effects on performance (Dawes & Corrigan, 1974; Wainer, 1976). Comparing the earlier predictions of the OW and UW models, for instance, shows that an equal-weighting rule for combining human and automation judgments would have approached the performance of the optimal-weighting rule.…”

“…Nonetheless, human decision makers can at least approximate the performance of a linear cue combination rule (Einhorn, Kleinmuntz, & Kleinmuntz, 1979), and though they tend not to weight cues optimally (e.g., Johnson, Cavanagh, Spooner, & Samet, 1973;Montgomery, 1999Montgomery, , 2001Montgomery & Sorkin, 1996), their deviations from normative weighting are likely to have modest effects on performance (Dawes & Corrigan, 1974;Wainer, 1976). Comparing the predictions of the OW and UW models above, for instance, shows that an equal-weighting rule for combining human and automation judgments would have approached the performance of the optimal-weighting rule.…”

Sustainable entrepreneurship and peculiarities of media-based learning

“…Besides the media effects there are also other individual aspects, starting with the widely differing perceptive abilities (Montgomery 1999). Visual learners are better off when they are presented with graphical displays; whilst auditive learner types ought to acquire information rather acoustically (Ferrari and Sternberg 1998).…”