How people experience virtual rooms: self-report, wearable physiology and the limits of spatial affect prediction in a VR pilot study
Neuroarchitecture promises evidence about how people will experience a space before it is built, often by combining virtual reality (VR) with wearable physiology. We report a three-experiment pilot in which 10 architects viewed 30 rendered rooms as 360° panoramas on a Meta Quest 3 headset (160 exposures) while two-channel forehead EEG and wrist ECG were recorded at 500 Hz, and rated each room afterwards. Rooms differed reliably in reported valence and arousal: room identity accounted for 27–41% of rating variance, and the reliability of room-mean ratings was 0.64–0.81 (ICC(3,k)). Comfort ratings of form-only rooms did not differ reliably (ICC 0.03). The daylit version of a room with identical plan and ceiling height was rated more pleasant and calmer than its night version by every participant, and, in a post hoc comparison, within each pair of rooms of the same function the larger room was rated more pleasant (5 of 5 pairs). Physiological features showed no consistent association with reported experience; the only interval excluding zero ran opposite to the theoretically expected direction. Room attributes predicted ratings of unseen rooms better than a person-mean baseline only when a strong shared factor (day versus night) was present. We argue that, at pilot scale, self-report is the more trustworthy signal for design, and derive concrete requirements for future VR–physiology studies. Preprint, not peer reviewed. Use of AI tools: AI assistants (OpenAI Codex; Anthropic Claude) were used for software development, analysis code, reference-metadata checks, figures and drafting; the author reviewed all analyses and text and is responsible for the content. Data and code: https://github.com/manideepmamidala-sys/Predictive-Atmospheres_Mark-3
Authors
- Manideep Mamidala (ORCID: https://orcid.org/0009-0009-4248-6698)
Institutions
- Institute for Advanced Architecture of Catalonia (ES)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-14
- DOI
- https://doi.org/10.5281/zenodo.24219078
- Primary Topic
- Virtual Reality Applications and Impacts
- Type
- preprint