Date of Award

Fall 2019

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Psychology

Program/Concentration

Human Factors Psychology

Committee Director

Yusuke Yamani

Committee Member

Holly Handley

Committee Member

Jing Chen

Abstract

The present work examined the effects of two types of decision support systems in a simulated luggage screening task: An input aid and an output aid. An input aid supports an operator’s information gathering. An output aid supports decision making and action selection. A Time-Accuracy Function (TAF) analysis was applied to isolate processing time from performance asymptote, which conventional performance measures such as sensitivity and response time do not distinguish one from the other. Sixty participants performed a luggage screening task unaided (manual condition), with the assistance of an input aid (spatial aid), and with the assistance of an output aid (decision aid) across different stimulus exposure durations of 250 ms, 500 ms, 1000 ms, 2000 ms, or 3000 ms. Participants were asked to judge the presence of a knife in each of the bags and either “stop” the bag or “pass” the bag. Reliability of the automated aids was 90% in Experiment 1 and 60% in Experiment 2. Experiment 1 showed that sensitivity increased with the assistance of both the input and the output aids as the stimulus exposure duration increased. The performance improvement was greater for the input aid than the output aid condition. Though processing times did not differ across the conditions, asymptotic performance level was higher when participants had the assistance of the input aid compared to the unaided condition. Experiment 2 and cross-experimental analysis demonstrated that the unreliable aids eliminated the benefit of the reliable aids. TAF analysis further showed that, although asymptotic performance can differ, processing times can remain constant regardless of DOA. The results imply that the input aid elevates asymptotic performance without influencing processing times, perhaps allowing operators to crosscheck their decisions within the restricted area of the search field identified by the aid. The present findings are inconsistent with the lumberjack hypothesis (Onnasch et al., 2013) and future research directions are provided.

DOI

10.25777/as82-8503

ISBN

9781392612538

Share

COinS