Researchers Build a Pre-Experimental Framework to Measure Emotions in Built Environments

Before any brain scans or eye-tracking sessions begin, researchers studying how buildings make people feel need to know whether their measurement tools actually hold together. A new open-access study published in the journal City and Built Environment presents a preliminary experimental framework for what the authors call visual emotionscape research, the emotionally shaped perceptual experience of built environments as evoked by their visual qualities.

The study centered on the Kitakyushu area and a multilingual online questionnaire distributed in English, Japanese, Chinese, and Bahasa Indonesia. The authors explain that multilingual distribution improved semantic precision, because certain emotional descriptors were too nuanced to rely solely on English translation; the English version served as the primary reference and all translations were reviewed by fluent speakers. Data were collected over a two-month period from May to June 2025 and involved 64 participants with diverse demographic backgrounds. The sample was predominantly young: 60.94 percent belonged to Generation Z and 37.5 percent to the millennial generation, with only 1.56 percent from Generation X. Men made up 59.38 percent of respondents, women 35.94 percent, and 4.69 percent chose not to disclose their gender.

The questionnaire was designed to gather foundational information to refine a subsequent experimental protocol rather than to test hypotheses at this stage. It assessed personality profiles using the Big Five, or OCEAN, model covering Openness, Conscientiousness, Extraversion, Agreeableness, and Neuroticism. Crucially, the researchers did not assign participants to a single dominant personality category, which they considered methodologically reductive. Instead, they calculated the percentage contribution of each trait, allowing personality to function as a weighted analytical variable. The instrument also collected physical and sensory conditions such as vision status, behavioral tendencies across biological, psychological, and sociocultural dimensions, personal experience, residential context, and preferences for travel and long-term residence.

Emotional responses were organized into three intensity levels, high, neutral, and low, across four primary emotion categories: fear, anger, sadness, and happiness. This structure follows a modified application of Russell’s circumplex model of affect, adapted for environmental evaluation, and the authors stress that the model is employed as an analytical structuring framework rather than a clinical diagnostic instrument. Raw emotional scores were standardized using Z-score normalization, a procedure that transforms individual responses into relative deviations from the overall sample mean, allowing responses measured on different scales to be compared across participants and emotional variables.

A central technical component of the study was a reliability analysis using Cronbach’s alpha, calculated for each emotion subscale and for the full scale based on participants’ binary indoor/outdoor preference responses across twelve emotion item pairs. The Happiness subscale, with five item pairs, demonstrated satisfactory reliability at alpha = 0.742. The Sadness subscale reached acceptable reliability at alpha = 0.629, while the Fear subscale fell below the conventional exploratory threshold at alpha = 0.588, indicating that its items require expansion. The Anger subscale could not be evaluated at all because it contains only one item pair, and Cronbach’s alpha requires a minimum of two items. The full twelve-item scale showed overall acceptable reliability at alpha = 0.661. The authors candidly note that Cronbach’s alpha is sensitive to the number of items, so the higher reliability of the Happiness subscale partly reflects its larger item count rather than a fundamental difference in scale quality.

The descriptive findings offer a portrait of how this sample relates emotionally to its surroundings. A striking 95.31 percent of participants reported that their emotions were influenced by their surrounding environment. Animated elements such as people and animals were identified as emotionally influential by 53.35 percent of participants, while 46.65 percent reported similar effects from static elements. Outdoor environments were reported as slightly more emotionally comforting than indoor ones, at 53.25 percent, with outdoor preferences mainly associated with spring, fall, and dry tropical climates. Spring was the most preferred season at 70.31 percent, and while a majority preferred driving over public transportation, most participants reported routinely walking distances of 500 to over 800 meters in daily life.

After normalization, the comparative analysis indicated that indoor environmental perception involved a broader range of affective evaluation, with positive indicators such as joy and enthusiasm showing wider dispersion and negative responses including tiredness and sadness fluctuating more pronouncedly. Outdoor evaluations, by contrast, displayed relatively more consistent positive emotional tendencies associated with calmness, openness, and relaxation. Across urban area types, residential environments exhibited comparatively higher variability, commercial areas were more frequently associated with energizing emotional tendencies, and industrial environments showed increased deviation values related to fatigue and boredom. For architectural forms, structures with greater visual mass and enclosure were associated with relatively higher deviation values in sadness- and tension-related indicators, while more visually open forms corresponded with stronger tendencies toward joy and enthusiasm.

The personality-weighted analysis suggested that trait characteristics were associated primarily with variations in the magnitude of emotional response rather than shifts in overall emotional direction. Openness, Conscientiousness, Extraversion, and Agreeableness demonstrated broadly parallel response tendencies across environmental contexts, while Neuroticism showed comparatively higher sensitivity in negatively valenced conditions, including relatively higher values in fear- or alertness-related dimensions and the most pronounced negative deviations in industrial areas. Anger-related emotions remained consistently low across all trait groups. The authors frame this pattern as consistent with trait-based affect reactivity perspectives, which hold that personality influences emotional intensity and vulnerability without fundamentally altering situational emotional appraisal.

When the framework was re-applied to stimuli from the team’s earlier pilot project, the multidimensional assessment revealed that emotional profiles of several architectural stimuli did not fully correspond to earlier categorical assumptions. Stimuli previously interpreted as predominantly sadness-related showed a broader distribution of affective indicators, suggesting that single-category labeling captured only a partial range of the emotional experience associated with those spatial configurations. The authors interpret these associations as indicative perceptual interpretations rather than deterministic design prescriptions.

The researchers are careful about what these results do and do not establish. The sampling approach was intentionally broad to establish a calibration baseline rather than population representativeness, and the sample of 64 participants is insufficient for inferential group-level comparisons. Roughly half of the participants were residing in the Kitakyushu case study area at the time of the survey, and the authors acknowledge the absence of a control condition comparing residents with non-residents in a matched design. The findings should be understood as indicative perceptual distributions rather than confirmatory evidence of environmental emotional performance, and the observed consistencies function as indicators of preparatory adequacy rather than evidence of causal environment-emotion relationships. Twenty-seven participants expressed willingness to take part in the subsequent experimental stage, which will incorporate EEG-based neural indicators of emotional valence and arousal together with eye-tracking measures of visual attention, and will expand the Fear and Anger item pools before formal psychometric validation. Within those limits, the study offers a documented template for how survey-based perceptual data can be structured, weighted, and stress-tested before a laboratory ever switches on its electrodes.

Subject of Research: A preliminary questionnaire-based framework for calibrating emotional measurement scales in built environment research

Article Title: A preliminary experimental framework for visual emotionscape research in built environments

Article References: Perwira, S. A., Dewancker, B., & Indra, P. (2026). A preliminary experimental framework for visual emotionscape research in built environments. City and Built Environment, 4(1), Article 8. https://doi.org/10.1007/s44213-026-00075-w

Image Credits: AI Generated

DOI: 10.1007/s44213-026-00075-w

Keywords: visual emotionscape, built environment, urban perception, Big Five personality, Cronbach's alpha, Z-score normalization, Kitakyushu, neuro-architecture, emotional scaling, questionnaire survey, EEG, eye-tracking