Date of Award

Spring 5-2022

Document Type

Thesis

Degree Name

Master of Science (MS)

Department

<--Please Select Department-->

First Advisor

Mary Fendley

Abstract

Decision aids are developed to ease cognitive load on operators interacting with complex automation systems; however, critical human components are often ignored during design. Finding an appropriate balance of automated assistance and operator trust is paramount in achieving optimal output from the human-automation interaction. Establishing a consistent metric of trust measurement will enhance the functional design of automated decision aids, especially as the use of eye tracking opens the field to the use of real-time measurements. This study will task participants to make measurements, assisted by a decision aid system, within a bone defect model image. The study will test for correlation between eye tracking data and participant trust survey answers. The researchers hypothesize that results from this experiment will indicate an inverse relationship between self-reported trust and gaze data, as the research participants will fixate fewer times and for shorter durations on the provided decision aids they show more self-reported trust in. Results from this study do not indicate significant correlations between trust and eye tracking metrics; however, negative relationships are seen. Percentage splits of fixation duration and fixation count on decision aids (when compared to overall fixation data between both AOIs) rise as decision aid reliability decreases. In summary, these results support eye tracking’s potential as a real-time measurement of human trust in automation.

Download

Share

COinS