Self-Driving Vehicles Against Human Drivers: Equal Safety Is Far From Enough

Abstract

We examined the acceptable risk of self-driving vehicles (SDVs) compared with that of human-driven vehicles (HDVs) and the psychological mechanisms influencing the decision-making regarding acceptable risk through 4 studies conducted in China and South Korea. Participants from both countries required SDVs to be 4-5 times as safe as HDVs (Studies 1 and 4). When an SDV and an HDV were manipulated to exhibit equivalent safety performance, participants' lower trust in the SDV, rather than the higher negative affect evoked by the SDV, accounted for their lower risk acceptance of the SDV (Studies 2 and 3). Both lower trust and higher negative affect accounted for why participants were less willing to ride in the SDV (Study 3). These reproducible findings improve the understanding of public assessment of acceptable risk of SDVs and offer insights for regulating SDVs. (PsycInfo Database Record (c) 2020 APA, all rights reserved).

Full text

Self-Driving Vehicles Against Human Drivers: Equal Safety Is Far
From Enough
Peng Liu
Tianjin University and University of Oxford
Lin Wang
Incheon National University
Charles Vincent
University of Oxford
We examined the acceptable risk of self-driving vehicles (SDVs) compared with that of human-driven
vehicles (HDVs) and the psychological mechanisms influencing the decision-making regarding accept-
able risk through 4 studies conducted in China and South Korea. Participants from both countries
required SDVs to be 4 –5 times as safe as HDVs (Studies 1 and 4). When an SDV and an HDV were
manipulated to exhibit equivalent safety performance, participants’ lower trust in the SDV, rather than
the higher negative affect evoked by the SDV, accounted for their lower risk acceptance of the SDV
(Studies 2 and 3). Both lower trust and higher negative affect accounted for why participants were less
willing to ride in the SDV (Study 3). These reproducible findings improve the understanding of public
assessment of acceptable risk of SDVs and offer insights for regulating SDVs.
Public Significance Statement
This study suggests that people require self-driving cars to be safer than conventional cars. Further,
it explains that people trust less in self-driving cars than conventional cars with equivalent safety
performance, which in turn leads them to be less willing to accept the risk of self-driving cars.
Keywords: acceptable risk, trust, affect heuristic, self-driving vehicles
Policymakers, scientists, and road safety organizations are en-
thusiastic about the potential for widespread adoption of self-
driving vehicles (SDVs) to reduce traffic accidents, traffic conges-
tion, and air pollution and to increase fuel efficiency, space
utilization, and human mobility (Anderson et al., 2016;National
Highway Traffic Safety Administration, 2016;Waldrop, 2015).
However, SDVs also pose risks and challenges related to safety,
security, liability, and regulation (Anderson et al., 2016;Bonnefon,
Shariff, & Rahwan, 2016;Fagnant & Kockelman, 2015;Liu,
Yang, & Xu, 2019b;Nunes, Reimer, & Coughlin, 2018;Xu et al.,
2018). Among dozens of studies on public perceptions, attitude,
and acceptance, some found that participants held positive atti-
tudes toward SDVs (e.g., Penmetsa, Adanu, Wood, Wang, &
Jones, 2019;Schoettle & Sivak, 2014), whereas others reported
participants’ resistance and negative attitude to SDVs (e.g.,
Nielsen & Haustein, 2018;Smith & Anderson, 2017). In particu-
lar, participants were concerned about potential risks of SDVs
(e.g., hacking; Liu et al., 2019b). The present series of studies
builds on these earlier findings to consider what people would
regard as an acceptable level of risk for SDVs, an issue that has so
far not been well addressed.
Removing control from human drivers is assumed to make
SDVs much safer than conventional human-driven vehicles
(HDVs; Mervis, 2017), but this has not yet been confirmed (Ba-
nerjee, Jha, Cyriac, Kalbarczyk, & Iyer, 2018). Then, how safe is
safe enough for SDVs? This popularized question has been widely
debated (Halsey, 2017;Hook, 2017;Mervis, 2017). Some law-
makers and regulators have been reported to consider allowing
SDVs to be deployed on roads provided they are deemed either as
safe as human drivers (Mervis, 2017) or twice as safe as human
drivers (Demers, 2018). Others claim that SDVs need to be mul-
tiple times (between 2 and 100 times) safer than HDVs, but, so far,
this claim lacks a scientific foundation (Shladover & Nowakowski,
2019). Policy researchers (Kalra & Groves, 2017) argued that a
less stringent policy (e.g., allowing SDVs to be slightly safer than
the average human driver) should be considered to save more
human lives. From a utilitarian standpoint, this policy seems
This article was published Online First March 23, 2020.
XPeng Liu, College of Management and Economics, Tianjin Univer-
sity, and Department of Experimental Psychology, University of Oxford;
Lin Wang, Department of Library and Information Science, Incheon Na-
tional University; Charles Vincent, Department of Experimental Psychol-
ogy, University of Oxford.
Peng Liu and Lin Wang contributed equally to this work. Lin Wang was
supported by Incheon National University Research Grant in 2018. Charles
Vincent was supported by the Health Foundation.
Correspondence concerning this article should be addressed to Peng Liu,
College of Management and Economics, Tianjin University, Building 25A,
No. 92 Weijin Road, Nankai District, Tianjin 300072, China. E-mail:
pengliu@tju.edu.cn
This document is copyrighted by the American Psychological Association or one of its allied publishers.
This article is intended solely for the personal use of the individual user and is not to be disseminated broadly.
Journal of Experimental Psychology: Applied
© 2020 American Psychological Association 2020, Vol. 26, No. 4, 692–704
ISSN: 1076-898X http://dx.doi.org/10.1037/xap0000267
692