Temperature Illusions in Mixed Reality using Color and Dynamic Graphics

Connor Wilson, Daniel J. Rea, Scott Bateman

Keywords: sensory illusion, perception, visual effects, mixed reality, augmented reality

TL;DR: We create a physical illusion of objects being hotter or colder by overlaying various graphics onto objects in mixed reality

Abstract:

Sensory illusions -- where a sensory stimulus causes people to perceive
effects that are altered by a different sensory stimulus -- have the potential to enrich mixed-reality based interactions which typically lack the ability to affect many physical senses. The well-known color-temperature illusion is a sensory illusion that causes people to, somewhat counter intuitively, perceive blue objects to feel warmer and red objects to feel colder. There is currently little information about whether this illusion can be recreated in mixed reality (MR). Additionally, it is unknown whether dynamic graphical effects made possible by mixed-reality systems could create a similar or potentially stronger effect to the color-temperature illusion. The results of our study (n=30) support that the color-temperature illusion can be recreated in MR, and that dynamic graphics can create a new temperature-sensory illusion. Our dynamic-graphics-temperature illusion creates a stronger effect than the color-temperature illusion and has more intuitive relationship between the stimulus and the effect: cold graphical effects (a virtual ice ball) are perceived as colder and hot graphical effects (a virtual fire ball) as hotter. Our results demonstrate that mixed reality has the potential to create novel and stronger temperature-based illusions and encourage further investigation into graphical effects to shape user perception.

Track: HCI/visualization

Summary Of Changes:

We thank the reviewers for their time and feedback. We believe the paper is now much stronger by addressing many of these points.

R1 had important concerns about our deviation from prior methods in terms of removing forced choice. We added additional applied and theoretical papers supporting the addition of noise and bias when using forced choice questionnaires for questions that could be reasonably have replies be neutral on effects or change. As we are measuring temperature illusions as well as a potential new illusion, it is quite reasonable that participants may see through the illusion or at least notice no difference. Thus, in addition to their being theoretical backing for our choice via additional citations, we argue that adding in forced choice makes our method more robust and allows participants to say they did not experience the illusion. This argument has been made in a new subsection in the paper, plus we have added additional justifications for choices throughout our method.

Reviewers asked for additional details about our analysis. R1 questioned our use of a white base color used in our dynamic graphics. We agree that there could be some effect, but at worst this should add additional noise to our results, and our main results do not hinge on the neutral white condition. Rather, we saw significant effects of Ice being perceived as colder than Blue and Fire being seen as warmer than Ice, despite lacking the extra power of planned contrasts. These indicate the dynamic graphics were still useful and suggest they are distinct from the color-temperature illusion and deserve more study. This argument has been added to the paper.

R2 and R3 asked for a more detailed result report, with R1 suggesting our results section was short. We have performed and reported all tests we planned to. While additional post-hoc tests could be added, we think that, methodologically speaking, they should be performed in an experiment designed for those tests. However, we do think reviewers had many important points for potential future experiments and results, and have expanded our discussion to include these. In particular, R2 and R3 raised intriguing possibilities, such as an amplification effect for actual temperature differences or further investigations into learning and sequential effects. All reviewers commented on the learning effects specifically, and we have worked their suggestions and added our own additional thoughts to the discussion section.

In addition, reviewers had questions that led us to realize many aspects of our report related to our block design was confusing. We have clarified this in our paper, but to quickly summarize: all blocks were equal, outside of counterbalancing the anchor pad order (with actual different temperatures), and the presentation of effects (where all pads were room temperature, secretly).

We thank reviewers for providing additional citations, including psychophysics and other related works. We have added these and other papers, in our related work section.

We hope we have addressed major concerns. Other feedback was all integrated into the provided draft.