VIRTUES:
VIRTUal Ethics Simulations

A runaway train car is speeding down a train track towards five construction workers who cannot hear it approaching. If it continues on its present course, it will hit and kill them. You are standing on a footbridge above the scene in between the approaching train car and the construction workers but you are too far away to warn them. A person is standing in front of you on the bridge. If you push this person, they would fall on the track below and the train car would hit them. They would be killed but the train car would derail and would not reach the five construction workers. Do you push the person sacrificing one life in order to save five lives?

These classical moral dilemmas allow researchers to test people's willingness to endorse a characteristically "utilitarian" response (sacrificing one life for many) versus the characteristically "deontological" response (refusing to endorse harm regardless of the outcome). Typically, these moral dilemmas are presented in text format and people rely on their imagination to visualise the scene and the events taking place. 

In our research, we have found that people respond differently to moral dilemmas when they are presented in Virtual Reality (VR) compared to text. People are more likely to simulate a utilitarian action (sacrificing one life to save many) in Virtual Reality  moral dilemmas. These findings suggest that increasing the feeling of authenticity in a dilemma influences moral decision-making. However, studies to date have only included one or two hypothetical dilemmas like the one above.

The purpose of the "VIRTUES" project has been to translate several moral dilemmas into VR so that we can investigate this pattern of responses across several different scenarios.

Below I outline what we have been doing in the project focusing on the design of the scenarios:

Scene design

When reading a text-based version of a moral dilemma, we rely on imagination to picture the events being described. Building moral dilemmas in Virtual Reality (VR) removes imagination and instead, we decide what a user will see and experience.

The scene design starts with basic structures to determine the sequences of events that are going to take place. In the screen shot above you can see a very simple outline of the footbridge dilemma.

Once the terrain is built, the objects are placed and the sequences of events are correct, we can start to increase perspectival fidelity and context realism. You can see the increased contextual saliency of the footbridge dilemma above.

User experience

We also have to consider the user's experience and how they will interact with the environment in a way that feels authentic. In the video below, we try to ensure that the user's movement in the environment would be similar to a real life version of the same space. For example, blocking the bright sunlight with their hand feels natural as it would in a real environment.

Variety

In the VIRTUES project, we have been working on several other moral dilemmas including some that are considered more plausible or realistic in nature. We have been developing scenarios refined by Christensen and Gomilla (2012) and adapted by Francis and McNabb (2022) for repeated-measures studies.  

All moral dilemmas involve the same form of moral decision-making - deciding whether it is better to sacrifice one life to save many more lives. 

For example, this moral dilemma is set in a hospital environment. This case is impersonal in that the user must decide whether to push a button to redirect toxic fumes from one room to another.

This scenarios involves a person seeking shelter and protection. However, if they are allowed to enter your settlement, they bring danger with them (in the form of disease or enemy soldiers for example). 

This scenario is set in a large city. The fidelity of this scene has been improved through several iterations of development. In this scenario, the user decides whether to sacrifice one person in order to stop them from harming several others on the streets below.

Since August 2023,  we have been collecting data in proof-of-concept studies. Scenarios are continuously refined to ensure efficient and engaging usability. This is a crucial first step prior to experimental testing.

Outputs & Outreach

To date, the project has produced the following outputs and the work has been presented at a number of events:

Publications

The moral dilemmas developed to be constructed in VR for repeated measures designs, were used in a longitudinal project during the COVID-19 pandemic:

The first set of VR moral dilemmas constructed in this project are undergoing proof-of-concept testing. The data will be included in the following publication in-prep:

VR Moral Dilemmas

The project has produced Virtual Reality versions of a number of moral dilemmas originally refined by Christensen and Gomilla (2012). 

These VR moral dilemmas have been created for either: 1) Oculus Rift S or 2) Oculus Quest 2.

When the scenarios have been tested (proof-of-concept, usability, and experimental), they will be openly accessible to other researchers.



Please get in touch if you would be interested in learning more about this project and getting involved in future work.


Outreach & Next Steps

This project has benn presented at several events including:



This early work has been used to support larger grant applications to ESRC, the Templeton Foundation, and the Leverhulme Trust. If you would be interested in collaborating on a research project in this area, please do get in touch. 



This project has supported us in developing a number of VR moral dilemmas (beyond the footbridge dilemma) that we can now start to validate in experimental research in moral psychology. After user testing, we hope to make these materials freely available to the scientific community so that researchers with an interest in moral decision-making can incorporate these into their own research.


I would like to thank both the British Academy and the Leverhulme Trust for funding this project.