Humans working with un-reliable automation: Reverse psychology versus disuse Model

Abstract

This paper investigated people’s behavior model and performance when working with an un-reliable automation. Automation with reliability higher than 70% is beneficial to human-automation system. Nevertheless, how automation with reliability far less than 70% affects human’s behavior and performance is yet unknown. Two types of behavior models: the reverse psychology model and the disuse model could occur when automation reliability is low but little research has investigated them. In this study, fourteen participants performed a two alternative forced choice (2AFC) memory test with an un-reliable automation with 20% or 40% reliability. Results showed that the number of occurrences of each behavior model was significantly affected by reliability level. When reliabil ity was 20%, there was a linear relationship between occurrence number of the reverse psychology model and performance improvement. Our findings provided a reference to better understand human behavior in a complex human-automation system where automation reliability could be low.

Adapting to Imperfect Automation: The Impact of Experience on Dependence Behavior and Response Strategies

“…Human interaction with diagnostic automation, a form of automation that analyzes raw information and infers the status of the world (Wickens & Dixon, 2007), is considered decision- making under uncertainty (Manzey, Gérard, & Wiczorek, 2014; Meyer, 2004; Sorkin & Woods, 1985; Wang & Yang, 2022; Yang, Unhelkar, Li, & Shah, 2017). As the most basic form of diagnostic automation, binary diagnostic automation categorizes the world into ‘signal present’ and ‘signal absent’ and alerts the human (often in the form of an alarm) once a signal is detected.…”