Introduction

Mental disorders such as mood and anxiety disorders are problems that are, and will continue to be, prevalent in the current world. The virtual world can confer many therapeutic benefits over naturalistic or traditional therapeutic situations, because it can provide a controlled environment where feelings of sadness or fear can be elicited for purposes such as understanding how an individual experiences and reacts in these situations. Virtual Reality (VR), the use of technology to create a simulated environment, has been shown to be an effective medium for psychotherapeutic techniques and can evoke the same reactions and emotions as a real experience in a stressor environment with the ability to manipulate the environment of the patient in an easy and safe way. VR has also started to become more prevalent in the entertainment industry and the elicitation of emotion can help in storytelling, making VR applicable in this context as well.

One of the advantages of using VR for rehabilitation compared to purely visual mediums, such as video, is the ability for interaction, the degree to which users of a medium can influence the virtual environment (VE), yet there has surprisingly been minimal research on its effect in evoking emotions. Being able to manipulate emotions not only gives us a better understanding of the psychology around them, but also shows us how we can provide better therapeutic treatment or better entertainment experiences.

When agents are present in a VE it is important to have a believable Artificial Intelligence (AI), which in this context is the use of algorithms to generate responsive and adaptive behaviors in non-player characters. For the behaviors to be believable, they need to be emergent, where the AI has different reactions to events depending on the context of the environment and the user’s behavior. These emergent behaviors are vital for making the reactions of the VE seem more realistic.

Project Goals

This project focused on Interaction and AI in already existing environments, one created to elicit fear and the other created to elicit sadness. Both environments were developed to achieve the following goals:

1) Create a VE which can manipulate the amount of interaction between the participant and the environment, allowing the user to choose between three varying levels of interaction.

2) Create a believable AI in the VE that reacts to the participant’s actions by dynamically altering the environment or the AI’s behavior.

Procedure

We followed a user centred design and development paradigm, followed by an open form and expert heuristic evaluation as our testing framework.

Each environment was developed using the Unity game engine which mainly uses C# to run scripts and was designed for use with the high-resolution HTC Vive headset as well as controllers, that act as virtual hands. An agile, user-centred design model was followed which focuses on iterative software development cycles and constant user feedback for each cycle. At each iteration of development, through prototyping and demos, feedback was provided from users in psychology and computer science fields, and the environment modified based on this feedback.



Once the environments were completed, each level of interaction underwent a VR Heuristic Evaluation, a method for finding usability problems in a user interface design by having a small number of evaluators, three in this case, examine the interface against a set of usability principles, the heuristics. The evaluator was presented with a few open form questions both before and after the evaluation finding out their emotional states and how the environment influenced them. The Heuristic Issues were analysed and discussed, and possible fixes were suggested for each problem that occurred. A few of these fixes were tested to see if the suggested remedial action was viable.

About us

Fear environment:
Jordan Taschner | TSCJOR001
tscjor001@myuct.ac.za
Sadness environment:
Brent van der Walt | VWLBRE001
vwlbre001@myuct.ac.za