Evaluating human decisions during time and actor independent variables

Walker, Charles (2020) Evaluating human decisions during time and actor independent variables. PhD thesis, University of Lincoln.

Evaluating human decisions during time and actor independent variables
PhD Thesis
Walker, Charles - Computer Science - June 2020.pdf - Whole Document

Item Type:Thesis (PhD)
Item Status:Live Archive


The Trolley Problem, a well known thought experiment comparing decisions during
life or death circumstances, was applied by the Moral Machine. This gained 40
million decisions from millions of participants. Whilst accepted and praised for its
success, investigation is available of participants position in the environment and the
effect of time-pressure on the decisions made.
To answer this, a web study was conducted to gain a quantitative understanding of
participants likelihood to make life or death decisions under the effect of the independent variables via generalised estimating equation. The effects of which proved
to be non-significant across both independent variables. Time-pressure showed self-sacrifice to be twice as likely when under time-pressure (B = 0.512, p = 0.012). This
effect was studied via a quantitative and qualitative virtual reality study, understanding whether the significance is repeatable. The results indicate the opposite,
showing regardless of the independent variable, participants are likely to sacrifice
themselves. The explanation of the prior studies findings being concluded as 5%
false positive in regards to significance.
The implications of both studies provide validation into the Moral Machine’s results, showing the independent variables not chosen by the Moral Machine had little
significance on participants decisions. This provides understanding around the development of a Trolley Problem algorithm in autonomous vehicles and the effects
that would occur in the world. The research also provides a recommendation that
research is required to understand the time taken to make a decision during both
time and non-time pressure decisions. This would be to see if non-time pressure is
being treated as such.

Keywords:Autonomous Vehicles, robotics, decision-making
Subjects:G Mathematical and Computer Sciences > G400 Computer Science
Divisions:College of Science > School of Computer Science
ID Code:48492
Deposited On:08 Mar 2022 15:53

Repository Staff Only: item control page