Human-Computer Interaction

How Embodiment Influences the Iowa Gambling Task in Virtual Reality

This project is already assigned.

1. Motivation

Decision-Making is important. We do it everyday, in all kinds of situations. What Am I eating for dinner? Which route do I take walking home? The blue pill or the red pill?

The Iowa Gambling Task (IGT) is a tool that recreates decision-making under uncertain conditions in a laboratory setting[1]. It has originally been developed to measure differences in decision-making between healthy populations and populations with damages to the frontal cortex. Since then, it has become a tool to measure how decision-making differs in populations with e.g eating disorders[2], shizophrenia[3] and problematic gambling[4], among others.

A previous study found that the Iowa Gambling Task has significant worse outcomes when done in immersive virtual reality (VR) compared to a desktop application[5]. Why exactly this happens is still not entirely clear, but this and related factors are worth investigating, as VR is a growing technology, which is used for education[6], training[7], therapy[8] and rehabilitation[9]. And making bad decisions when it comes to these kinds of applications is…bad. Do I really need to go on why it is bad to make bad decisions which could hinder the success of education, training, therapy or rehabilitation?

The IGT and factors influencing it

While there are several different versions of the IGT, the standard test as developed by Bechara et al. consists of drawing cards from 4 decks[1]. A digital version of this task is validated[10].

There are many factors that influence decision-making in the Iowa Gambling Task: One of them are emotions, which negatively impact decision-making according to the Somatic Marker Hypothesis[11]. Others are risk-taking behavior[12], the consumption of spicy food beforehand [13] or the watching of humorous videos while doing the task[14].
Another factor is biological sex, as women have, in general, worse results in the Iowa Gambling Task when compared to men[12, 15, 16, 17]. The reasons behind this effect are not yet clear, but it might be rooted in differences of brain structure between men and women[18]. But there are also implications that it might be rooted in other factors, as level of education or culture also influence the IGT outcome[19, 20]. A recent study found that this effect of women being worse in the IGT is stronger when female participants are confronted with stereotypes (e.g that they are worse in some cognitive tasks than men) before solving the IGT [21]. This lends weight to the theory that this difference is based in societal factors and preconceived stereotypes, such as the so-called “Stereotype threat”.

IGT and VR and Embodiment

Another factor that influences the IGT outcome is immersive VR, which makes it worse. As neither the degree of immersion nor the virtual environment have an impact on the outcome in VR, this might be rooted in other factors such as the visual angle on stimuli, which is bigger in VR [22].

But besides the potential of impairing decision-making, VR brings many interesting new facets into IGT research: One of them is embodiment, where a user’s real body is substituted with a virtual body. This is sometimes also called “illusion of virtual body ownership”[23]. It heightens a users emotions, which might negatively impact decision-making [24]. As such, it stands to reason that embodiment might lead to even worse results in the IGT. Yay.

On the other hand, IGT outcomes are better when turning real cards [25]. If embodiment substitutes the real body in such a way, that a user can use their hands to turn over digital cards like real cards, then the negative impact of embodiment might be minimized. In short, embodiment in general might have a negative impact on the IGT, but drawing cards through a virtual body’s hands might offset this negative impact.

This means that research on embodiment and IGT would benefit from comparing not only embodiment vs. no embodiment, but that the interaction method might also influence the outcomes. For this, it would be interesting to compare drawing cards with button presses on a controller and drawing cards through grabbing and turning them with hands.

This direction of research might be interesting not only because embodiment is common in current VR applications, but it might also lead way to future work, in which the impact of different bodies on the IGT is explored. As the embodiment itself might change decision-making, this serves as preliminary research for further probes into embodiment and the IGT.

Embodiment and Gender effects

Another direction for research into embodiment and the IGT are the previous mentioned gender effects.

When users in VR are embodied into a body, they change their behaviour according to how they think this body should behave. This effect is dubbed Proteus Effect and works through stereotypes[26].

Gender stereotypes have an impact on the IGT through Stereotype threat. It might be interesting to know if the gender differences in the IGT persist, if users are embodied in the opposite gender. If this effect does not persist in changed gender in VR, then it supports the theories that societal factors and stereotypes have a big part in explaining the difference in gender on the IGT. This might also lead to interesting future work, in which other bodies and their impact on the IGT can be researched.

3. Methodology

This project is split into two parts:

As part two is dependent on embodiment, this is the second part of this thesis and will be done after the first part is complete and the effect of embodiment itself on the IGT is clear.

Two hypothesis are drawn from the theoretical background that relate to part one:

Concerning part two, there are two hypotheses that are drawn from the theoretical background:

Study 1 for Part One:

To investigate the effects of embodiment on the IGT, participants will be embodied in one of three conditions:

Each participant will be in only one of the conditions, making this a “between-subjects” - Design. In the conditions with full embodiment, participants are represented with a virtual body. This body is scaled according to a participants size. Male participants have a male body, female participants have a female body. In the conditions with controller interaction, participants can select the decks with the thumbsticks and draw cards by pressing a button. In the hand interaction, participants use their hands, which are tracked, and can grab them by pinching their fingers, similar to picking up cards in real life.

The main measure of this study is the IGT itself, but demographics and control measures are measured as well, among them being:

In total, 20 participants per condition should be measured.

Study 2:

This study wants to investigate how gender effects influence the IGT with embodiment in the opposite gender.

For this, the following variants of embodiment are implemented:

As each participant will only be in one of these conditions, this is also a “between-subjects” design. In total, I expect to raise at least 20 participants per condition, making it at least 80 in total. But as the Covid-19 numbers are again on the rise and it seems more and more inadvisable, wellllll…lets stay positive that this study will still be possible in the next few weeks.

The main measure of this study is again the IGT, but demographics and control variables are measured as well, among them being:

System

To test the impact of embodiment on the IGT, I developed a VR application in Unity for the Oculus Quest. This application is based on the previous system as used in several studies by Oberdörfer et al [5,22].

In this application, the user is represented through 3 kinds of embodiment (minimal embodiment, full embodiment with controller interaction, full embodiment with hand interaction). The virtual bodies for this embodiment are simple, cartoonish human bodies. They are animated from the hips upward, and hand and head movements are tracked. When a user is fully embodied with hand interaction, they can grab the cards of the IGT by using their real hands, which are tracked through the Oculus Hand Tracking. They then have to turn around the cards like real playing cards to see the result. This is a very close approximation the analogue IGT, in which real cards are drawn in a similar manner. The other two conditions use controller interaction, where a user can use the left thumbstick to select decks. On the press of the A-Button, a card is drawn and the result is shown to the participant. The system logs which decks the cards are drawn from in which order.

And this is the end of my exposee. I can’t believe you just read all of it but thank you, dear reader. If you want me to chew off your ears with more interesting things about the IGT or if you have any further questions, just write me a mail or read the papers as published by my supervisor.

4. Literature


Contact Persons at the University Würzburg

Dr. Sebastian Oberdörfer
Mensch-Computer-Interaktion, Universität Würzburg
sebastian.oberdoerfer@uni-wuerzburg.de

Follow our Social Media Channels
Contact
Lehrstuhl für
Mensch-Computer-Interaktion
Universität Würzburg
Emil-Fischer-Straße 50
97074 Würzburg
How to Find Us
University Main Page
About the University
University Contact
Legal Information