Unveiling the underlying structure of awe in virtual reality and in autobiographical recall: an exploratory study

The study investigates whether the multidimensional latent structure of the awe experience—operationalized via the Awe Experience Scale (AWE-S)—is consistent across two common induction techniques (autobiographical recall and VR), and whether this structure replicates in an Italian context. Awe, often defined by vastness and need for accommodation, has been linked to self-transcendent features (time distortion, small self, connectedness) and positive outcomes (wellbeing, prosociality). While AWE-S has been validated primarily with US participants, cultural and linguistic differences (notably in Italian) may affect appraisals, regulation, and expression of emotions. The research aims to translate and validate AWE-S in Italian, examine its factor structure via autobiographical recall (Study 1), and test the same structure under a standardized VR induction (Study 2), addressing whether induction method and intensity affect the latent structure of awe.

The paper reviews conceptualizations of awe, including Keltner and Haidt’s account highlighting vastness and need for accommodation, and the emotion’s varied ‘flavors’ (e.g., admiration, elevation, the sublime). Awe is associated with self-transcendent qualities (time alteration, connectedness, small self) and beneficial outcomes (wellbeing, prosociality, inspiration, authenticity). Induction methods range from everyday awe to intense, transformative experiences akin to the sublime, with VR increasingly used to elicit profound awe in controlled settings. Cultural differences between US and Italian contexts in emotion appraisals and language suggest the need for psychometric validation of awe measures beyond English. Prior research reveals diverse elicitors (nature, skills, religion, monuments, leaders, theories, art, music, epiphany, childbirth), and proposes grouping elicitors into domains (spiritual engagement, music, dance, psychedelics; collective inductors). The central unresolved question is whether awe’s latent structure varies across induction techniques and intensity levels.

Two studies were conducted. - Study 1 (Autobiographical recall): N=350 adults (255 females; mean age females=31.8 SD=14.7; 95 males; mean age males=29.9 SD=15.4), mostly educated (14 years avg schooling), living mostly with family; about half reported Christian Catholic religion. Recruitment via social networks; online survey (Qualtrics). Measures: Italian translation of AWE-S (30 items; 7-point Likert; six dimensions: Vastness, Need for Accommodation, Connectedness, Self-diminishment/Small self, Physical Sensations, Time), PANAS (state), DPES (trait positive emotions; Awe subscale), ERQ (trait emotion regulation: Reappraisal, Suppression). Translation used forward–back translation by two bilingual translators per Borsa et al. guidelines. Procedure: Participants recalled and described a personal awe experience (instructions per Yaden et al.), then completed AWE-S, DPES, PANAS, ERQ. Analyses: Exploratory Factor Analysis (EFA) using Principal Axis Factoring with promax rotation; Parallel Monte Carlo analysis to determine number of factors; reliability (Cronbach’s alpha); Pearson’s correlations for concurrent validity. Additionally, qualitative coding of awe elicitors (per Yaden et al. categories plus Social Connection) and typologies of nature (Blue/Green/White spaces; Mountains; Ephemeral phenomena; Brown environments; Other). - Study 2 (VR induction): N=106 adults (74 females; mean age=27.2 SD=9.2; males mean age=29.6 SD=11.3); 39.7% with prior VR experience. Platform: AltspaceVR (non-immersive, screen-based), using validated awe-inspiring scenarios (“Deep Space” and “Tall Forest”) with ~5-minute narratives, accessed in counterbalanced order within a 1-hour training including tutorials, interaction practice, guided reflection. Measures: Italian 29-item AWE-S from Study 1, DPES Awe subscale, ITC-SOPI Sense of Presence (subscales: Sense of Physical Space, Engagement, Ecological Validity, Negative Effects). Procedure: Participants navigated scenarios under facilitator guidance, then completed questionnaires on Qualtrics. Analyses: Confirmatory Factor Analysis (CFA) in Mplus using MLM estimator (robust to non-normality); fit indices (CFI, TLI, RMSEA, SRMR, AIC, BIC); stepwise model modifications based on modification indices to allow selected residual correlations; internal reliability (Cronbach’s alpha); correlations between AWE-S subscales and DPES Awe; correlations between AWE-S subscales and ITC-SOPI dimensions; inter-factor correlations among AWE-S dimensions.

Study 1 (Autobiographical recall): - Parallel analysis supported a six-factor solution; eigenvalues >1 for six factors; total variance explained=61%. - Adequacy: KMO=0.834; Bartlett’s test χ2(425)=6625, p<.01. Items showed expected non-normality (kurtosis) for intense awe items. - Item 29 did not load and had low communality; removed. Final Italian AWE-S=29 items: five items per factor except Need for Accommodation with four. - Reliability (Cronbach’s alpha): Time=0.87; Self-loss (Small self)=0.94; Connection (Connectedness)=0.85; Vastness=0.80; Physical Sensations=0.84; Need for Accommodation=0.88. - Concurrent validity (selected results): Dispositional Awe (DPES) correlated positively with AWE-S Time, Connection, and Physical Sensations during autobiographical recall; ERQ Reappraisal correlated positively with most AWE-S dimensions; ERQ Suppression correlated positively with Need for Accommodation and Self-loss; PANAS Negative Affect correlated positively with Self-loss and negatively with Connection and Physical Sensations. Happiness and Pride (DPES) showed significant negative correlations with Self-loss. - Awe elicitors (N=227 coded): Natural Scenery=114; Grand Theory/Idea=15; Birth=15; Other=28; Great Skill=9; Building/Monument=8; Epiphany=8; Art=7; Social Connection=6; Meditation=4; Encounter with God=3; Great Virtue=3; Death=3; Music=2; Powerful Leader=2. - Nature typologies (N=114): Blue space=47; Ephemeral phenomena=13; Green space=14; White space=10; Brown environment=10; Mountain=9; Other=11. Study 2 (VR induction): - CFA of 29-item 6-factor model: Full model fit indices: χ2(362)=687.532, p<0.001; RMSEA=0.092 (90% CI 0.082–0.103); CFI=0.899; TLI=0.887; SRMR=0.081. Successive models with added item residual correlations improved fit, with final Model M4: AIC=8649.8; RMSEA=0.061; CFI=0.957; TLI=0.951; SRMR=0.057 (best fit). - Reliability (Cronbach’s alpha): Time=0.956; Self-loss=0.970; Connectedness=0.960; Vastness=0.955; Physical Sensations=0.891; Need for Accommodation=0.890. - Inter-factor correlations (selected): Time with Self-loss r=0.664***; with Connection r=0.503***; with Vastness r=0.503***; with Physical Sensations r=0.345***; with Need for Accommodation r=0.422***. Connection with Vastness r=0.690***; Physical Sensations with Vastness r=0.638***; Connection with Physical Sensations r=0.602***. - DPES Awe correlations with AWE-S subscales: Time r=0.216 (p<0.01); Connection r=0.262 (p<0.01); Self-loss r=−0.194 (p<0.01); Vastness r=0.150 (ns); Physical Sensations r=0.092 (ns); Need for Accommodation r=0.137 (ns). - ITC-SOPI Ecological Validity showed negative correlations with Time r=−0.223 (p<0.05), Connectedness r=−0.254 (p<0.01), Physical Sensations r=−0.191 (p<0.05).

Across both induction methods, the original English six-factor structure of AWE-S replicated in Italian, suggesting that the multidimensional latent structure of awe is robust to variations in elicitation technique (autobiographical recall vs. VR) and medium. However, the trait–state relationships were modest and varied by context: dispositional awe related positively to Time and Connection in both studies, but showed a negative relation to Self-loss in VR and lost the Physical Sensations association in VR, implying that simulated environments may engage embodied aspects differently and potentially attenuate individual differences. The Self-loss dimension aligned with negative affect and suppression, while Connection and Vastness aligned with more positive affective profiles, supporting awe’s mixed-valence nature. Item 29 (Need for Accommodation) was removed due to low communality, echoing challenges in operationalizing accommodation; future work could leverage expectancy violation paradigms to refine this construct. Presence findings suggest that greater perceived ecological validity in non-immersive VR was associated with reduced time distortion, connectedness, and bodily sensations typical of awe; paradoxical or fantastical content may drive awe more than realism in such contexts. Elicitor analyses reaffirmed nature’s prominence but also highlighted conceptual and social elicitors among Italians, pointing to cultural and contextual variability in what triggers awe.

The Italian translation and validation of the AWE-S demonstrated a stable six-factor structure that held under both autobiographical recall and standardized VR induction. While the latent structure overlaps across English and Italian contexts, trait–state correlations were generally low, indicating potential cultural and methodological influences on awe’s components. The validated Italian AWE-S can be used to probe dimensions contributing to awe’s positive and negative nuances. Future research should refine lexical and operational definitions of awe in non-English contexts, explore expectancy-violation mechanisms for accommodation, examine immersive versus non-immersive VR effects, and investigate gender and cultural moderators.

- The Italian translation used a periphrasis for “awe” (profonda meraviglia), and Study 2 did not explicitly anchor the term in the questionnaire, potentially affecting participants’ interpretation. - Lack of a validated Italian state measure of awe limited convergent/divergent validity assessment; comparisons relied on trait measures (DPES, ERQ) and PANAS. - Gender imbalance (more female participants) may influence generalizability. - Presence was assessed in VR but not in relation to autobiographical memory vividness. - Non-immersive, screen-based VR may elicit awe differently than immersive HMD-based VR, limiting ecological generalizability of VR findings.