PaperlayerView exact revision on arXiv

Source-linked AI summary

Virtual Reality for Emotion Elicitation -- A Review

Rukshani Somarathna, Tomasz Bednarz, Gelareh Mohammadi

arXiv:2111.04461v1cs.HC

TL;DR

Prior reviews largely emphasize passive emotion elicitation, leaving limited comprehensive guidance on VR as an active mechanism and on suitable VR materials. This review surveys VR media, datasets, and sensing interfaces, concluding that VR can elicit a broad range of emotions and support ecologically valid affective research, while facial-expression measurement remains constrained.

  • Problem

    Prior reviews have largely emphasized passive elicitation, leaving limited comprehensive evidence on VR as an active emotion-induction mechanism and on suitable VR materials.

  • Method

    The review analyzes VR visual and audio-visual stimuli, games and tasks, 360-degree media, public datasets, and emotion-sensing interfaces across affective-computing research.

  • Results

    The survey concludes that VR media can elicit both discrete and dimensional emotions across a wide range of affective experiences.

  • Takeaways & Limitations

    VR’s immersion, presence, interactivity, and active participation make it a promising ecologically valid paradigm for studying emotions.

  • Takeaways & Limitations

    VR emotion research remains constrained in measuring full facial expressions, and available sensing interfaces have limited evidence and require further validation.

Abstract

from arXiv · show

Emotions are multifaceted phenomena that affect our behaviour, perception, and cognition. Increasing evidence indicates that induction mechanisms play a crucial role in triggering emotions by simulating the sensations required for an experimental design. Over the years, many reviews have evaluated a passive elicitation mechanism where the user is an observer, ignoring the importance of self-relevance in emotional experience. So, in response to the gap in the literature, this study intends to explore the possibility of using Virtual Reality (VR) as an active mechanism for emotion induction. Furthermore, for the success and quality of research settings, VR must select the appropriate material to effectively evoke emotions. Therefore, in the present review, we evaluated to what extent VR visual and audio-visual stimuli, games, and tasks, and 360-degree panoramas and videos can elicit emotions based on the current literature. Further, we present public datasets generated by VR and emotion-sensing interfaces that can be used in VR based research. The conclusions of this survey reveal that VR has a great potential to evoke emotions effectively and naturally by generating motivational and empathy mechanisms which makes it an ecologically valid paradigm to study emotions.

1 Introduction

The review addresses the limited literature-informed guidance on using VR for active, ecologically valid emotion elicitation. It surveys VR media, emotion representations, datasets, and sensing interfaces to clarify how VR can support affective research.

  • The review evaluates VR visual and audio-visual stimuli, games and tasks, 360-degree media, and mixed reality as emotion-induction materials.
  • The survey covers 74 papers and presents public VR-generated datasets and measurement guidance intended to support future affective-computing research.
  • Passive elicitation dominates prior work, whereas active methods offer greater ecological validity and immersivity through interactive participation.
  • VR studies have elicited both discrete emotions and dimensional affect, including valence and arousal, using immersive environments and games.
  • VR’s immersion, presence, interactivity, and participant isolation support controlled experiments designed to evoke emotionally engaging experiences.

4 Elicitation material related to Virtual Reality

The review compares VR elicitation materials, emphasizing how interactivity, immersion, and stimulus type shape emotional experience. Games and tasks provide active involvement, while 360-degree media support immersive dimensional emotion studies.

  • VR studies use visual and audio-visual stimuli, games and tasks, immersive environments, and 360-degree media to elicit emotions.
  • Visual and Audio-Visual Stimuli: Static images are easy to present in VR but evoke weaker, shorter-lived responses than videos, virtual environments, and games.Image-based responses are often collected through subjective reports and are less suited to discrete emotion representations.
  • Games and Tasks: Games actively engage players through control, decision-making, challenges, and changing events, supporting motivational tendencies and variable emotional states.These properties make games relevant for studies of appraisal processes and physiological emotion monitoring.
  • Games and Tasks: Compared with traditional video games, VR games provide greater immersion and subjective involvement by allowing users to encounter and influence events in mediated environments.Traditional games can separate participants from narration through avatars, whereas VR games are described as a more fully dimensional medium.
  • 360-degree Panoramas and Videos: 360-degree environments provide panoramic exploration, and architectural panoramas achieved 75% arousal and 71.21% valence prediction accuracy.The environments varied illumination, colour, and geometry to target valence-arousal space.

5 Different emotions in Virtual Reality

The review finds that VR materials can elicit both discrete and dimensional emotions, although the literature is uneven across emotion categories and media types. It also surveys VR datasets supporting affective research.

  • Across reviewed publications, VR showed potential for studying emotions and triggering physiological changes through diverse media types.
  • Discrete Emotions: Fear, joy, sadness, anger, relief, and amusement were frequently induced, whereas pride, compassion, regret, and several other emotions were rarely reported.The review attributes this imbalance to emotional complexity, limited research attention, or insufficient VR content.
  • Discrete Emotions: Visual and audio-visual stimuli were used more often for happiness, sadness, and anger, while games and tasks more often targeted amusement, pleasure, contentment, relief, fear, and tension.Games were used less for many negative emotions, and 360-degree media appeared less frequently overall.
  • Dimensional Emotions: Visual and audio-visual stimuli, games, and tasks effectively supported valence, arousal, and dominance research, although evidence for dominance remained limited.Audio-visual content and games were more common than 360-degree media for eliciting valence and arousal.
  • Public Datasets: Public VR datasets include 73 annotated immersive video clips, 87 rehabilitation-game recordings with ratings and physiological signals, and databases based on everyday environments and 360-degree panoramas.The datasets support research using valence, arousal, dominance, EEG, eye tracking, and other physiological measures.

7 Measuring subjective experience in Virtual Reality

VR studies commonly measure subjective experience through participant self-reports, using paper, digital, verbal, or VR questionnaires. These evaluations address discrete, dimensional, and appraisal emotion models.

  • VR research commonly evaluates subjective experience with paper-based, digital, verbal, or VR self-report questionnaires.Studies assess discrete, dimensional, and appraisal models of emotion.

8 Emotion sensing interfaces for Virtual Reality

The review describes physiological and wearable interfaces integrated with VR headsets for emotion and activity recognition. These systems expand measurement possibilities, but their psychological and neuroscientific evidence remains limited.

  • VR emotion sensing combines wearable physiological signals, headset-integrated sensors, and real-time acquisition interfaces for affect and activity recognition.
  • Physiological Interfaces: Facial electromyography is suited to VR because headsets largely cover the face, limiting conventional computer-vision analysis of expressions.
  • Integrated Sensor Systems: Interfaces such as PhysioHMD, LooxidVR, LooxidLink, and retrofitted EEG headsets support configurable EEG, EMG, EOG, EDA, eye-tracking, or PPG acquisition with VR systems.
  • Limitations: VR-embedded biosignal interfaces remain under-supported by neuroscience and psychology evidence and require further attention and improvement.

9 Practical Implications in using Virtual Reality

Practical VR research should account for motion sickness, prior immersive experience, novelty bias, and the limited measurement of full facial expressions.

  • Motion sickness should be assessed early, with participant filtering and tools such as VRSQ supporting safer data collection.The review recommends evaluating kinetic environments across the average population and using established questionnaires.
  • Prior VR experience or training can reduce novelty bias, unexpected frustration, and mismatches between expected and reported emotions.Participants may report positive emotions despite negatively intended material, while training improves familiarity with VR hardware and controllers.
  • VR interfaces have limited ability to capture detailed facial expressions because they provide only a few physiological signal points.Future studies should verify whether these interfaces fit diverse facial structures effectively.

10 Discussion

The discussion presents VR as an increasingly used active medium for studying emotions across content types, physiological responses, and emotional models. It also identifies dataset scarcity, positive-emotion bias, and immature multimodal methods as important boundaries for future work.

  • 10 Discussion: The review surveyed VR emotion induction, public datasets, emotion-sensing interfaces, subjective measurements, and limitations across the literature.Its scope centers on visual and audio-visual stimuli, games, tasks, and 360-degree panoramas and videos.
  • 10 Discussion: VR media, games, and 360-degree materials have elicited a wide range of emotions across discrete and dimensional models.The review highlights first-person role-play, virtual collaboration, empathy, and motivational tendencies as distinctive features of VR games.
  • 10 Discussion: VR can trigger physiological changes and motivational tendencies, supporting study of emotional physiology and motivation over extended experiences.Games are linked to motivation through tendencies such as trying to win and may support evaluation of the Component Process Model.
  • 10 Discussion: Active VR elicitation is presented as a pragmatic direction for obtaining more natural feelings than predominantly passive approaches.The discussion connects active participant involvement with reliable affective-computing research and calls for benchmark datasets.
  • 10 Discussion: Public VR datasets remain scarce, existing contents favor positive emotions, and multimodal haptic-emotion research remains immature.The review identifies limited negative-emotion content and calls for deeper investigation of data-collection methodology.

11 Conclusion

The survey presents VR as a promising medium for emotion elicitation across multiple content types and identifies datasets for future VR-based research.

  • The review concludes that VR can elicit emotions using visual and audio-visual stimuli, games, tasks, and 360-degree media.It also discusses discrete and dimensional emotion models.
  • The survey presents public datasets generated through VR and sensor interfaces for future research.
  • The authors identify potential for future real-time adaptive VR designs based on users’ emotions.

Supplementary Material

The supplementary material catalogs visual and audio-visual VR studies, including their research aims, participant information, equipment, and analytic methods.

  • Table 3 assesses visual and audio-visual VR materials used in affective computing, psychology, and human-computer interaction.The table includes physiological measures, virtual environments, and statistical or machine-learning methods.
  • The surveyed studies examine VR for emotion analysis, physiological responses, immersion, empathy, and affective environments.
  • Reported analysis approaches include machine learning, clustering, and qualitative analysis alongside physiological sensing with PPG, EEG, and related measures.

Supplementary material B

The supplementary material catalogs VR games and tasks used to study emotions, trust, aggression, immersion, decision making, personality, and player experience.

  • Table 4 assesses games and tasks used as VR materials in affective computing, psychology, and human-computer interaction.It covers statistical analysis and machine-learning approaches used across the studies.
  • The studies investigate trust under cognitive load, aggression, empathy, personality needs, risk environments, and emotional experience.
  • Other studies compare immersion or medium effects on decision making, horror games, player experience, and emotional intensity.

Supplementary material C

Table 5 assesses 360-degree panoramas and videos used as VR material in affective computing, psychology, and human-computer interaction.

  • The table covers 360-degree panoramas and videos alongside EEG, SVM, KNN, naïve Bayes, random forest, GBM, and DNN methods.

Supplementary material D

Table 6 catalogs VR games and experiences used to elicit varied emotional experiences, including fear, relaxation, thrill, and engagement.

  • Table 6 overviews VR games used in the literature alongside their targeted or reported emotional experiences.
  • Games and experiences include shooting, survival horror, bomb defusal, fruit slicing, obstacle avoidance, and puzzle-based activities.
  • Several scenarios use collaboration, remote partners, or interactive challenges to structure the emotional experience.
  • Other materials provide exploration, meditation, museum, zero-gravity, skyscraper, and thrill-ride experiences.

Supplementary material E

Table 7 compiles VR emotion studies by materials, emotional targets, physiological or self-report measures, and reported outcomes across discrete and dimensional models.

  • Table 7 summarizes VR studies spanning discrete emotions, valence-arousal-dominance dimensions, emotional challenges, and physiological classification.
  • VR elicited fear, anxiety, sadness, happiness, surprise, excitement, empathy, compassion, anger, guilt, and relaxation across varied scenarios.
  • Accuracy reached 86.03% for arousal detection from CNN and 82.5% for valence detection from facial EMG using SVM.
  • Physiological signals supported emotion classification, including 95.61% F1 for excited and 95.50% for relaxed states.
  • VR sometimes produced higher presence without greater emotional intensity, and one museum comparison found physiological changes in real but not virtual scenarios.
Loading 2111.04461v1…