Dynamic Thermal Feedback in Highly Immersive VR Scenarios: a Multimodal Analysis of User Experience

Thermal feedback is critical to a range of Virtual Reality (VR) applications, such as firefighting training or thermal comfort simulation. Previous studies showed that adding congruent thermal feedback positively influences User eXperience (UX). However, existing work did not compare different levels of thermal feedback quality and mostly used less immersive virtual environments. To investigate these gaps in the scientific literature, we conducted a within-participant user study in two highly-immersive scenarios, Desert Island (n=25) and Snowy Mountains (n=24). Participants explored the scenarios in three conditions (Audio-Visual only, Static-Thermal Feedback, and Dynamic-Thermal Feedback). To assess the complex and subtle effects of thermal feedback on UX, we performed a multimodal analysis by crossing data from questionnaires, semi-structured interviews, and behavioral indicators. Our results show that despite an already high level of presence in the Audio-Visual only condition, adding thermal feedback increased presence further. Comparison between levels of thermal feedback quality showed no significant difference in UX questionnaires, however this result is nuanced according to participant profiles and interviews. Furthermore, we show that although the order of passage did not influence UX directly, it influenced user behavior. We propose guidelines for the use of thermal feedback in VR, and the design of studies in complex multisensory scenarios.

💡 Research Summary

This paper investigates how ambient thermal feedback influences user experience (UX) in highly immersive virtual reality (VR) environments. While prior work has shown that adding congruent temperature cues can improve presence, most studies have been limited to low‑fidelity visual‑audio settings, have used only a single quality level of thermal feedback, and have relied primarily on post‑experience questionnaires. To address these gaps, the authors designed two richly detailed scenarios—Desert Island (hot) and Snowy Mountains (cold)—and recruited 49 participants (25 for the hot scenario, 24 for the cold scenario) to experience each scenario under three conditions: (1) Audio‑Visual only (no thermal feedback), (2) Static thermal feedback (temperature changes that are congruent with the environment but do not adapt to user actions), and (3) Dynamic thermal feedback (temperature that updates in real time based on the user’s position and actions). The study employed a within‑subject design with counterbalanced condition order to control for learning effects.

Data were collected from three complementary sources. Standardized questionnaires measured general presence, physical presence, sense of agency, and haptic experience after each condition. Semi‑structured interviews were conducted after the final session to capture nuanced, qualitative impressions of the thermal cues. Objective behavioral metrics—including physical and virtual travel distance, number of teleportations, and movement entropy—were automatically logged using the PLUME toolbox, providing synchronous, low‑cost insight into participants’ in‑scene behavior.

Statistical analysis revealed that any form of thermal feedback significantly increased presence scores compared with the audio‑visual baseline (p < 0.01), confirming that temperature cues remain a potent immersion enhancer even in high‑fidelity VR. However, the difference between static and dynamic thermal feedback was not statistically significant on questionnaire scales (p = 0.27). Qualitative interview data clarified this ambiguity: participants with higher self‑reported thermal sensitivity described dynamic feedback as “more natural” and “more immersive,” whereas participants with lower sensitivity reported little distinction. This suggests that the benefit of dynamism is contingent on individual thermal perception rather than being universally advantageous.

The order of exposure did not affect questionnaire outcomes, but it did produce a clear trend in behavioral metrics. Across both scenarios, participants traveled shorter distances, performed fewer teleportations, and exhibited reduced movement entropy in later sessions, indicating a learning or habituation effect that streamlined navigation over time.

Technically, the authors employed a non‑wearable ambient thermal system consisting of infrared lamps and fans arranged around a 2 × 2 m tracked play area. The system was integrated with Unity to receive real‑time positional data and adjust temperature output. Static feedback replayed a pre‑programmed temperature profile, while dynamic feedback altered the profile within roughly 800 ms of a positional change. Although this latency is acceptable for gradual environmental temperature changes, the hardware’s thermodynamic limits constrained rapid temperature swings, a limitation the authors acknowledge for future hardware development.

The paper’s contributions are fourfold: (1) it validates that ambient thermal cues enhance presence even in high‑fidelity, task‑oriented VR; (2) it provides the first direct comparison of static versus dynamic thermal feedback, highlighting the importance of user‑specific thermal sensitivity; (3) it demonstrates the value of a multimodal analysis framework that triangulates questionnaires, interviews, and behavioral data to uncover subtle UX effects; and (4) it offers methodological recommendations for designing and evaluating multisensory VR studies, including guidelines for hardware selection, experimental control, and data synchronization.

Future work suggested includes incorporating physiological measures (EDA, ECG, skin temperature) to objectively map affective and cognitive states, exploring higher‑performance thermal actuators (e.g., high‑power Peltier devices or electrical nerve stimulation) to reduce latency and increase temperature range, extending the participant pool to capture cultural and age‑related differences in thermal perception, and investigating synergistic effects of combined sensory modalities such as wind, humidity, and scent. By doing so, researchers can refine the parameters that make thermal feedback a truly adaptive, user‑centric component of immersive VR experiences.