The Effect of Intermediate Trust Ratings on Automation Reliance
Abstract
As automated systems are increasingly capable of augmenting human decision-makers, appropriate reliance on automation has the potential to increase safety and efficiency in several high-stake domains. To that end, a solid understanding of how and under what conditions people rely on automation is needed to design decision aids that allow people to rely on them appropriately. Previous studies have used regular trustratings during human automation interactions to examine how trust develops and evolves, but such intermediate judgments might affect subsequent reliance decisions. This dissertation addresses a knowledge gap by empirically exploring how intermediate trust ratings affect automation reliance in human automation interactions. A laboratory experiment in which 118 participants, supported by automated decision aids,identified UAVs in images was conducted to determine whether trust rating frequency, automation reliability, and participant motivation affected participant reliance behavior. Findings show that intermediate trust ratings increased automation reliance and retrospective trust ratings but did not affect response time.This dissertation proposes an extended theoretical model that might help explain and predict automatio nreliance. Additionally, it suggests that intermediate trust ratings might be suitable for calibrating automation reliance but not for research that seeks to measure trust without influencing reliance behavior.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 2023
- Accession Number
- AD1225623
Entities
People
- Linus Torner
Organizations
- Naval Postgraduate School