A Systematic Review of Extended Reality (XR) Applications in Cultural Heritage

Abstract

This systematic review examines how Extended Reality (XR) technologies, i.e., Virtual (VR), Augmented (AR), Mixed (MR), and Spatial Augmented Reality (SAR) are designed, implemented, and evaluated in cultural heritage (CH) applications, addressing five research questions: (RQ1) How were XR technologies applied in CH between 2021 and 2025? (RQ2) What interaction paradigms are used, and how do they shape engagement and meaning making? (RQ3) What user experience outcomes are reported in XR CH applications? (RQ4) What evaluation methods are employed and what methodological gaps remain? (RQ5) What challenges persist across XR heritage implementations? Peer-reviewed, English-language studies reporting on implemented XR systems in CH contexts with empirical or evaluative data were included; conceptual articles without a described implementation, non-English publications, and studies published before January 2020 were excluded. Scopus, Web of Science, IEEE Xplore, and the ACM Digital Library were searched for publications dated January 2020 through March 2025, complemented by manual proceedings screening (SIGGRAPH, CHI, IMX, VRCAI) and backward/forward citation tracking. All databases were last searched in March 2025. Two independent researchers screened all records and extracted data; disagreements were resolved through structured discussion. Bias toward positive novelty outcomes was mitigated by including conference proceedings alongside journal articles to broaden the evidence base. A qualitative thematic synthesis was employed, as methodological heterogeneity across studies precluded statistical meta-analysis. Findings were organized inductively into four thematic domains through iterative coding and inter-author consensus. From an initial corpus of 359 records, 287 unique records were retained after deduplication; following title/abstract screening and full-text eligibility assessment, 64 studies were included in the final synthesis. The majority (60/64) were published between 2021 and 2025, with study sample sizes ranging from small expert cohorts (n ≈ 6) to large public deployments (n > 125). The thematic analysis across technology, interaction design, user experience, and evaluation reveals trends toward participatory, multiuser, and multimodal XR designs, reporting benefits including immersion, engagement, learning, and accessibility, alongside recurring challenges such as cost, usability, cybersickness, content authenticity, and lack of longitudinal evaluation. Beyond thematic description, using a cross-domain analytical synthesis, we identify the Design Coherence Framework for XR Heritage (DCF-XR); this is a four-dimensional interpretive model spanning technology, interaction design, user experience, and evaluation, which provides an original diagnostic lens for understanding the conditions under which XR effectively serves cultural heritage goals. A typology of four recurring design failure modes, derived inductively from the corpus, demonstrates that the most persistent shortcomings in the field arise not from the weakness of individual dimensions but from their misalignment with one another. Evidence is limited by the predominance of small convenience samples, single-session laboratory evaluations, and the absence of domain-specific standardized assessment instruments for XR in CH, which constrains the generalizability of reported outcomes. Targeted recommendations for rigorous, ethical, and inclusive XR practice in CH are presented, highlighting the need for longitudinal studies, open datasets, and standardized evaluation frameworks. This review received no external funding. This review was not pre-registered in a prospective register.

1. Introduction

Protecting, preserving, and presenting both tangible artifacts and intangible cultural practices are enduring responsibilities that sustain historical continuity and cultural identity [1,2]. Traditional forms of displaying artefacts, such as static displays, text-based representations, and digital imagery, do not serve contemporary, technology-literate audiences well, particularly younger visitors who expect interactive, inquiry-driven experiences [2]. At the same time, these forms fail to convey embodied, performative, and contextual dimensions that are closely related to their creation since they encompass the socio-historical context that gives meaning to their creation.

As a possible solution to these limitations, digitally enabled technologies have the potential to offer new opportunities to safeguard heritage, broadening public access to cultural heritage (CH) by providing new and richer forms of engagement [3,4,5]. Museums and heritage sites remain indispensable, yet many artifacts are fragile or inaccessible, with their presentation and access constrained by conservation requirements [6,7,8]. Immersive and interactive Extended Reality (XR) systems give the power to extend access beyond the physical limits of an installation and thus can be used to support deeper interpretive and educational experiences.

XR denotes a spectrum of technologies that blend physical and virtual realities [9,10]. Virtual Reality (VR) fully immerses users in simulated environments; Augmented Reality (AR) overlays digital content onto the physical world; Mixed Reality (MR) allows interactive virtual objects to coexist with and respond to the real environment; and Spatial Augmented Reality (SAR) projects imagery onto physical surfaces [9,11,12,13,14,15]. Each of these can be considered as an alternative interaction and visualization modality that offers distinct affordances for CH. For example, VR is well-suited for site reconstruction and history-driven narratives; AR can support contextual overlays in situ; MR is appropriate for hands-on manipulation of virtual replicas; SAR is applicable in creating shared, large-scale installations as well as museum-based applications that convey information through projections and the selective illumination of physical or 3D-printed models [13,15,16].

While previous reviews have examined digital technologies in CH, significant gaps remain in the literature. Recent reviews have focused on single aspects of the field [17,18,19], such as the AR use, immersive VR reconstructions, or head-mount-display-based (HMD-based) systems; they have not studied evidence from the full spectrum of XR. Moreover, existing reviews cover primarily pre-2021 literature [17] or adopt broad conceptual frameworks [20] that do not systematically use empirical evidence from implemented CH applications. In addition, although several reviews analyze technology and user experience, none systematically analyze interaction paradigms and evaluation methodologies. This review goes beyond evidence mapping and descriptive aggregation. Building on the four thematic domains, Section 4 develops an analytical synthesis that identifies recurring design configurations, cross-cutting tensions, and conceptual patterns emerging from the corpus. At the heart of this synthesis is the Design Coherence Framework for XR Heritage (DCF-XR), a four-dimensional model that treats technology, interaction design, user experience, and evaluation not as independent analytical categories but as mutually dependent dimensions of a unified design logic. The DCF-XR holds that the most effective XR heritage applications are those in which all four dimensions are oriented toward the same cultural and interpretive goal, and that the field’s most persistent failures arise precisely where this coherence breaks down. This framework is not imposed on the studied literature from outside; it is derived inductively from patterns that are only visible through cross-corpus synthesis, and it constitutes the primary conceptual contribution of this review beyond its descriptive and bibliographic functions.

Under this prism, this SLR aims to address these gaps by providing a recent and cross-XR analysis and synthesis. It aims to consolidate empirical evidence on how XR technologies are designed, implemented, and evaluated in CH settings. We organize the literature across four thematic domains: technology, interaction design, user experience, and evaluation to provide an account of current practice, challenges, and promising directions for research and applied work. To guide this systematic review, we formulated the following research questions:

• RQ1: How were XR technologies applied in CH between 2021 and 2025?
• RQ2: What interaction paradigms are used, and how do they shape engagement and meaning making?
• RQ3: What user experience outcomes are reported in XR CH applications?
• RQ4: What evaluation methods are employed and what methodological gaps remain?
• RQ5: What challenges persist across XR heritage implementations?

2. Methodology

This systematic review follows the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework to ensure transparency and reproducibility in the search, screening, and synthesis stages [21]. The review was conducted without a pre-registered protocol; no amendments to the procedures were made during the review process. The review protocol specified databases, exact search strings, inclusion/exclusion criteria, and a pre-registered coding scheme for the thematic analysis.

2.1. Review Process

2.1.1. Eligibility Criteria

Studies were included in this review if they met all the criteria listed in

Table 1. Description of the eligibility criteria used in this study.

Criterion	Inclusion	Exclusion
Publication type	Peer-reviewed journal articles and conference papers	Editorials, opinion pieces, posters without empirical data
Language	English only	Non-English publications
Publication date	January 2020–March 2025	Published before January 2020
Topic	XR (VR, AR, MR, SAR) applied in CH contexts	XR in non-CH domains; CH studies without XR
Study type	Implemented XR systems with empirical or evaluative data	Purely conceptual or theoretical proposals without implementation
Outcome	Reports on technology, interaction design, UX, and/or evaluation data	Studies reporting no usable empirical outcomes

. Studies were grouped for synthesis into four thematic domains based on the primary contribution of each work as identified during full-text coding: (1) technology, (2) interaction style, (3) user experience, and (4) evaluation methods.

2.1.2. Information Sources

We searched Scopus, Web of Science, IEEE Xplore, and the ACM Digital Library for publications dated January 2020 through March 2025. All four databases were last searched on 15 January 2026. We also hand-searched proceedings from SIGGRAPH, CHI, IMX, and VRCAI and performed backward and forward citation tracking. These supplementary sources were last consulted in January 2026.

2.1.3. Search Strategy

The following Boolean search strings were applied consistently across all databases, adapted to each database’s syntax where required:

• Scopus/Web of Science: [TITLE-ABS-KEY ((“Virtual Reality” OR “Augmented Reality” OR “Mixed Reality” OR “Extended Reality” OR “Spatial Augmented Reality”) AND (“Cultural Heritage” OR “Intangible Cultural Heritage”))].
• IEEE Xplore: [(“Virtual Reality” OR “Augmented Reality” OR “Mixed Reality” OR “Extended Reality” OR “Spatial Augmented Reality”) AND (“Cultural Heritage” OR “Intangible Cultural Heritage”)].
• ACM Digital Library: [Title: (“Virtual Reality” OR “Augmented Reality” OR “Mixed Reality” OR “Extended Reality”) AND Abstract: (“Cultural Heritage” OR “Intangible Cultural Heritage”)].

Filters applied across all databases: publication date (January 2020–March 2025); document type (journal article, conference paper); language (English).

2.1.4. Selection Process

The outcome of this systematic search was a corpus of 359 records; after removing duplicate entries, we obtained a corpus of 287 unique records. Both authors independently screened all 287 records at the title and abstract stage using the eligibility criteria defined in Section 2.1.1. Disagreements were resolved through structured discussion until consensus was reached, with moderation from the third author. Title and abstract screening excluded 193 records, leaving 94 full-text articles for eligibility assessment. Access to full texts was supported by institutional access via the authors’ organizations. From the 94 full-text articles, 29 were excluded for failing to meet the inclusion criteria, yielding a final set of 64 studies. The primary reasons for full-text exclusion were as follows: (1) no implemented XR system described; (2) no empirical or evaluative data reported; (3) XR application outside a CH context; (4) study falling outside the 2020–2025 date range. The selection process is shown in the PRISMA flow diagram (Figure 1). Of the 29 full-text articles excluded after eligibility assessment, the exclusion reasons were as follows: (1) no implemented XR system described, often in cases where the paper was purely conceptual or a review itself (n ≈ 12); (2) no empirical or evaluative data reported, i.e., system papers without any user study or technical evaluation (n ≈ 9); (3) XR application outside a CH context (n ≈ 5); (4) study published outside the 2020–2025 date range (n ≈ 3). The process is presented in Figure 1.

2.1.5. Data Collection Process

Data were extracted from all 64 included studies using a standardized coding form developed by the authors before extraction. The form captured the following variables for each study: XR type (VR, AR, MR, SAR), hardware platform, software framework, content-generation method, interaction paradigm, UX outcomes reported (positive and negative), evaluation instruments used, study design, sample size, and publication year. Both authors independently extracted data from each study; discrepancies were discussed and resolved by consensus. No automation tools were used in the data extraction process. No study investigators were contacted for additional data.

2.1.6. Data Items

The primary outcomes sought for each thematic domain are detailed in

Table 2. Primary outcomes sought for each thematic domain.

Thematic Domain	Primary Outcomes Sought
Technology	Hardware platform, software engine, content creation method, supporting infrastructure
Interaction Style	Input modality, interaction paradigm, narrative structure
User Experience	Immersion, presence, engagement, learning, emotional connection, accessibility, cybersickness, cognitive load
Evaluation	Evaluation instruments, study design, sample size, qualitative/quantitative methods

. Additional variables extracted for each study included the following: CH domain (tangible/intangible), geographic context, study setting (lab/in situ/museum), participant demographics, and funding, wherever such information was disclosed by the authors. Where information was unclear or missing, the item was recorded as “not reported”, and no assumptions were made.

2.1.7. Risk of Bias Assessment

Risk of bias in individual studies was assessed using a lightweight appraisal rubric adapted for non-clinical, technology-centered systematic reviews. Each study was evaluated on four dimensions: (1) clarity of system description (sufficient to replicate), (2) appropriateness of evaluation design (sample size, study setting), (3) transparency of results reporting (raw data or statistics provided), and (4) potential for novelty bias (single-session, first-use evaluations). Both authors independently rated each study; discrepancies were resolved through discussion moderated by the third author. Formal inter-rater reliability was not calculated, given the qualitative nature of the appraisal. Results of the risk of bias assessment per study are reported in Section 3 (Results).

2.1.8. Effect Measures

As this review employs qualitative thematic synthesis rather than statistical meta-analysis, no standardized effect measures (e.g., risk ratios, mean differences, confidence intervals) were calculated or reported. The heterogeneity of study designs, outcome measures, and evaluation instruments across the corpus precluded any formal quantitative aggregation.

2.2. Thematic Categorization

Each article in the final corpus was systematically coded using the data extraction form described in Section 2.1.5, and the rationalization of the coding results allowed the formulation of four central themes:

• Technology (hardware, software, content creation): represents the technological foundation that supports design choices, performance, accessibility, and scalability.
• Interaction style (input modalities and narrative structures): concerns the interaction design that affects how users engage with the software, moving from passive consumption to active exploration.
• User experience (immersion, learning, accessibility, emotion): captures outcomes and challenges from the user’s perspective.
• Evaluation (methods and instruments used to assess usability and educational impact): provides evidence of usability, effectiveness, and cultural/educational value.

Studies were assigned to one or more thematic domains based on their primary contributions, as determined during coding. Themes were derived inductively through iterative coding across both authors and consolidated through discussion with the moderation of the third author until a consensus was reached. No quantitative synthesis (meta-analysis) was performed; instead, a narrative and tabular thematic synthesis approach was adopted, presenting findings through descriptive tables, frequency counts, and interpretive summaries for each theme.

Heterogeneity across studies (in design type, sample size, CH domain, and XR technology) was explored descriptively by comparing study characteristics within each thematic domain. No formal subgroup or meta-regression analyses were conducted given the qualitative nature of the synthesis.

No sensitivity analyses were conducted, as the synthesis is qualitative and not dependent on statistical model assumptions.

Publication bias was not formally assessed using statistical tests (e.g., funnel plots). To mitigate selective reporting bias, conference proceedings and grey-literature-adjacent sources were included alongside journal articles, and both positive and negative outcomes were systematically recorded during data extraction.

Certainty of evidence was not formally graded (e.g., via GRADE), as this review synthesizes descriptive empirical findings from heterogeneous, non-clinical studies rather than intervention-comparison outcomes. The overall confidence in synthesized findings is discussed qualitatively in the Discussion Section, with explicit reference to evaluation rigor limitations.

3. Results: A Thematic Analysis of XR in CH

3.1. Overview of the Thematic Analysis

3.1.1. Characteristics of Included Studies

The characteristics of all 64 included studies are presented in

Table 3. Overview of the included studies.

Ref #	Author(s)	Year	XR Type	CH Context	Hardware Platform	Interaction Paradigm	Evaluation Method(s)	Sample	Venue Type
[1]	Xu et al.	2025	AR	Shadow puppetry (ICH)	Mobile AR	Storytelling	Custom questionnaire	NR	Conference
[2]	Su	2025	VR	Peking Opera costumes (ICH)	HMD (VR)	Exploratory navigation	Interviews, observation, photogrammetry	NR	Journal
[3]	Khalloufi et al.	2023	VR	Jemaa El-Fna, Marrakech (ICH)	HMD (Oculus Quest)	Exploratory navigation	SUS, SSQ, custom Likert	NR	Journal
[4]	Li	2025	AR	Intangible CH (general)	Mobile AR	Exploratory navigation	Custom questionnaire	NR	Conference
[5]	Mogre et al.	2025	AR	Artifact preservation (general)	Mobile AR	Exploratory navigation	System proposal (no user study)	N/A	Conference
[6]	Huang et al.	2025	XR (VR and AR)	Sanxingdui Bronzes (ICH)	HMD (Oculus Quest) and mobile	Gamification, exploratory navigation	SUS, GEQ, custom Likert	NR	Journal
[7]	Li et al.	2018	VR and AR (Hybrid)	CH objects (tangible)	HMD (HTC Vive) and mobile AR	Multiuser collaboration	Custom questionnaire	52 (26 pairs)	Conference
[8]	Petrelli and Roberts	2023	VR and TUI	Roman culture (tangible)	HMD (Google Cardboard/PSVR) and tangible	Embodied, interactive storytelling	Observation, interviews	NR	Journal
[9]	Li et al.	2022	VR and SAR	Cultural spaces (general)	Projection and Nreal Light	Exploratory navigation	System description (no formal user study)	N/A	Journal
[10]	Olaz et al.	2022	MR and SAR	In situ CH visit (tangible/intangible)	Projection (SAR)	Interactive storytelling, co-design	Participatory workshop	NR	Journal
[11]	Gatto et al.	2025	XR	Church of Madonna dell’Itri (tangible)	Mobile, HMD and tactile	Multimodal, accessibility	Custom Likert, accessibility evaluation	30	Journal
[12]	Neamțu et al.	2024	XR (VR, AR and MR)	Dacian Bronze Matrix (tangible)	HMD, mobile, and desktop	Gamification, exploratory navigation	Custom Likert, comparative	111 (37 per condition)	Journal
[13]	Xu et al.	2024	VR	Qin Dynasty Baixi (ICH)	HMD (VR)	Exploratory navigation	Interviews, observation, custom questionnaire	NR	Conference
[14]	Sun and Wang	2023	AR	ICH and cultural tourism	Mobile AR	Interactive storytelling	Custom Likert	NR	Conference
[15]	Russo et al.	2024	VR	Industrial heritage (BIM-VR)	HMD (VR)	Exploratory navigation	Custom questionnaire	NR	Journal
[21]	Fanini et al.	2021	WebXR	Archaeological site (tangible)	Web browser (WebGL/WebXR)	Exploratory navigation	System evaluation (technical)	N/A	Journal
[22]	Hulusic et al.	2023	VR, and TUI	Schoolhouse CH (tangible)	HMD (Oculus Quest) and TUI	Tangible UI, interactive storytelling	GEQ, IEQ, custom questionnaire	62 and 25	Journal
[23]	Lucifora et al.	2024	VR	Social cohesion/CH (intangible)	HMD (Oculus Quest)	Multimodal, embodied	SUS, PQ, SSQ, custom Likert	30	Journal
[24]	Rahaman et al.	2023	VR (360°)	Historic hotel heritage (tangible)	HMD and 360° panorama	Exploratory navigation, embodied	Interviews, observation, SSQ	NR	Journal
[25]	Wang and Li	2025	VR (360°)	Panoramic virtual museum (general)	HMD (360°)	Exploratory navigation	SUS, custom questionnaire	NR	Journal
[26]	Yan and Du	2025	VR	Historical districts (tangible)	HMD (Pico 4)	Exploratory navigation	Custom Likert, structural equation model	NR	Journal
[27]	Wang et al.	2025	VR	Chinese flower arrangement (ICH)	HMD (Oculus Quest)	Multiuser, exploratory navigation	Interviews, custom questionnaire	NR	Conference (CHI)
[28]	Xhako et al.	2024	VR and MR	Antiquity dresses/museum (tangible)	HMD (HoloLens) and VR	TUI, multiuser	Expert evaluation (11 experts), custom Likert	11	Journal
[29]	Jangra et al.	2025	VR	Museum experience (general)	HMD (Oculus Quest) and mobile	Exploratory navigation	Custom questionnaire, comparative	80	Journal
[30]	Kong	2024	VR	Diabolo (ICH)	HMD (VR)	Gamification, participatory design	Co-design workshop	NR	Conference
[31]	Liang	2024	XR	CH experience (general)	HMD (VR, AR, and MR)	Exploratory navigation	Doctoral colloquium (design research)	N/A	Conference
[32]	Bekele	2021	MR (Cloud)	Virtual heritage (tangible)	HMD (HoloLens) and cloud	Multiuser, multimodal	Expert evaluation	11	Journal
[33]	Bekele et al.	2021	MR (Cloud)	Cultural learning (virtual heritage)	HMD (HoloLens) and cloud	Multiuser, multimodal	Custom questionnaire, interviews	NR	Journal
[34]	Boray et al.	2025	MR	Falconry heritage (ICH)	HMD (HoloLens 2)	Embodied, interactive storytelling	System description (no formal user study)	N/A	Conference
[35]	Du et al.	2024	XR (AR, MR)	CH tourism (tangible)	HMD (Nreal) and TOLED	Exploratory navigation	Expert interview (6 specialists)	6	Conference (SIGGRAPH)
[36]	Ch’ng et al.	2023	AR (Social)	CH objects (tangible)	Mobile AR	Multiuser, social	Observation, interviews	30 (10 groups)	Journal
[37]	Prandi et al.	2025	AR	Cultural phenomenon (ICH)	Mobile AR, physical installation	Interactive storytelling, gamification	Survey, focus group	73 and 125	Journal
[38]	Souropetsis and Kyza	2025	AR	CH sites (tangible)	Mobile AR	Gamification, interactive storytelling	Custom questionnaire	NR	Journal
[39]	Vilar et al.	2025	XR	Cultural gaming/interactive narratives	Mixed (web-based)	Gamification, interactive storytelling	SUS, custom questionnaire	NR	Journal
[40]	Wu	2023	VR	ICH digital inheritance (general)	Desktop/Web VR	Interactive storytelling, virtual guide	Custom questionnaire	NR	Conference
[41]	Zhang	2023	VR	ICH digital display (general)	Desktop VR	Gamification, adaptive narrative	Custom Likert	NR	Conference
[42]	Serres et al.	2023	AR	Digital humanities/CH (general)	Mobile AR (WebAR)	Exploratory navigation	Custom questionnaire	NR	Journal
[43]	Guimarães et al.	2015	AR	Public garden heritage (tangible)	Mobile AR	Gamification, interactive storytelling	Custom questionnaire	NR	Conference
[44]	Gao et al.	2018	AR	Campus heritage (tangible)	Mobile AR	Exploratory navigation, TUI	Custom questionnaire	10 and pilot	Conference
[45]	Chen et al.	2024	VR, and SAR	Cultural exhibition (tangible)	HMD (VR), projection (SAR)	Multiuser, interactive storytelling	Custom questionnaire, comparative	16 (8 pairs)	Conference
[46]	Nikolakopoulou and Koutsabasis	2025	MR	ICH in museums (intangible)	MR setup	Embodied, co-design	Participatory workshop	NR	Conference
[47]	Nikolakopoulou et al.	2022	MR and TUI	Tinian marble crafts (ICH)	MR and physical controller	TUI, gamification	Custom questionnaire, interviews	NR	Journal
[48]	Zhang et al.	2023	VR	NVSHU (ICH characters)	HMD (VR)	Adaptive narrative, interactive storytelling	System description	N/A	Conference (SIGGRAPH)
[49]	Muñoz et al.	2025	XR (MR and TUI)	CH engagement (general)	MR and tangible (antilatency)	TUI, multisensory	SUS, Focus group (6 experts), Formative (9 staff)	6 and 9	Journal
[50]	Schauer and Sieck	2023	MR	Virtual reconstruction CH (tangible)	MR (Vuforia LiDAR)	TUI, exploratory Navigation	System description (technical)	N/A	Conference
[51]	Xiong et al.	2025	MR	Sugar painting (ICH)	MR (Meta SDK)	TUI, interactive storytelling	System description	N/A	Conference (SIGGRAPH)
[52]	Di Feola and Rostami	2025	XR	Grief/memory/CH (intangible)	XR and tactile	Embodied, multisensory	Qualitative design research	NR	Conference
[53]	Huang and Huang	2025	VR	Heritage landscape (tangible)	HMD and desktop VR	Exploratory navigation	Custom questionnaire, comparative	25 and 12	Conference
[54]	Costa et al.	2024	VR	Woodcutting (ICH)	HMD (Unity VR)	Embodied, interactive storytelling	System description	N/A	Conference
[55]	Kebir et al.	2025	VR	Bardo Palace architecture (tangible)	HMD (VR)	Exploratory navigation	SUS, interviews	NR	Journal
[56]	See et al.	2018	VR	Tomb of a Sultan (tangible)	HMD (VR) and Kinect	Embodied, interactive storytelling	Interviews, observation	NR	Conference
[57]	Tong et al.	2024	VR (360°)	Indigenous storytelling (ICH)	HMD (VR 360°)	Interactive storytelling, adaptive narrative	PQ, interviews, observation	NR	Journal
[58]	Hu	2024	VR and AR	CH (general)	HMD, mobile, and desktop	Exploratory navigation	Custom questionnaire	72	Journal
[59]	Tsita et al.	2023	VR	Contemporary art museum (tangible)	HMD (Oculus Quest)	Exploratory navigation, virtual guide	SUS, SSQ, interviews	NR	Journal
[60]	Yu et al.	2025	VR	Centennial Drama (ICH)	HMD (VR)	Contextual storytelling, adaptive narrative	PQ, interviews, observation	NR	Journal
[61]	Xie	2021	VR	ICH mobile display (general)	Mobile VR	Exploratory navigation	System description (technical)	N/A	Journal
[62]	Cai and Yang	2023	VR	Danzhai Miao Batik (ICH)	HMD (VR)	Embodied, interactive Storytelling	System description	N/A	Conference
[63]	Ji et al.	2021	VR	ICH course (embodied learning)	HMD (VR)	Embodied interaction	Observation, custom questionnaire	20	Conference
[64]	Zhang et al.	2024	VR	Digital cultural tourism (general)	HMD (VR)	Virtual guide, AI communication, exploratory navigation	Custom Likert, interviews	NR	Conference
[65]	Kobeisse	2023	AR and TUI	Historical artefacts (tangible)	Mobile AR, and tangible	TUI, gamification	Custom questionnaire (in situ)	80 and 572	Conference
[66]	Elrawi	2017	MR	Islamic CH (tangible)	MR setup	Interactive storytelling	System description	N/A	Conference
[67]	Andrade	2023	MR (hybrid)	CH discourse (general)	Mixed (hybrid)	Interactive storytelling	Discourse analysis (qualitative)	N/A	Conference
[68]	Monteiro et al.	2023	AR	CH sites (general)	Mobile AR	Co-design, heuristic evaluation	Participatory workshop, expert heuristics	20, 12, and 9	Conference

, providing an overview of each study’s XR technology type, CH context, hardware platform, interaction paradigm, evaluation methods, and sample size.

3.1.2. Risk of Bias in Included Studies

Risk of bias was assessed for all the included studies across the four dimensions defined in Section 2.1.7: (D1) clarity of system description, (D2) appropriateness of evaluation design, (D3) transparency of results reporting, and (D4) potential for novelty bias. The distribution of ratings across all the dimensions is presented in the figure below, and the per-study ratings are provided in

Table 4. Risk of bias assessment per included study.

Ref #	Author(s)	Year	D1: System Description	D2: Evaluation Design	D3: Results Transparency	D4: Novelty Bias	Overall Risk	Notes
[1]	Xu et al.	2025	Low	High	Moderate	High	High	Sample NR; single session; limited stats
[2]	Su	2025	Low	High	Moderate	High	High	Sample NR; qualitative only; novelty likely
[3]	Khalloufi et al.	2023	Low	High	Low	High	High	Sample NR; SUS/SSQ used but N not disclosed
[4]	Li	2025	Low	High	High	High	High	Sample NR; minimal methodological detail
[5]	Mogre et al.	2025	Low	High	High	Moderate	High	No user study; system proposal only
[6]	Huang et al.	2025	Low	Moderate	Low	Moderate	Low	Validated instruments (SUS, GEQ); multi-condition
[7]	Li et al.	2018	Low	Low	Low	Moderate	Low	N = 52 (26 pairs); custom instrument; controlled
[8]	Petrelli and Roberts	2023	Low	Moderate	Moderate	High	Moderate	Qualitative only; NR sample; rich observation
[9]	Li et al.	2022	Low	High	High	Moderate	High	No user study; technical system description
[10]	Olaz et al.	2022	Low	Moderate	Moderate	Moderate	Moderate	Participatory workshop; NR sample
[11]	Gatto et al.	2025	Low	Low	Low	Moderate	Low	N = 30; accessibility evaluation; multimodal
[12]	Neamțu et al.	2024	Low	Low	Low	Low	Low	N = 111 (37/condition); comparative; multi-XR type
[13]	Xu et al.	2024	Low	High	Moderate	High	High	Sample NR; qualitative only; single session
[14]	Sun and Wang	2023	Low	High	High	High	High	Sample NR; minimal evaluation detail
[15]	Russo et al.	2024	Low	High	Moderate	High	High	Sample NR; BIM-VR; no validated instrument
[21]	Fanini et al.	2021	Low	High	High	Low	Moderate	Technical evaluation only; no user study
[22]	Hulusic et al.	2023	Low	Low	Low	Moderate	Low	N = 62 and 25; GEQ and IEQ used; multi-session aspects
[23]	Lucifora et al.	2024	Low	Low	Low	Moderate	Low	N = 30; SUS, PQ, and SSQ; validated instruments
[24]	Rahaman et al.	2023	Low	High	Moderate	High	High	Sample NR; qualitative; 360° novelty effect
[25]	Wang and Li	2025	Low	High	Moderate	High	High	Sample NR; SUS used but N not disclosed
[26]	Yan and Du	2025	Low	High	Low	Moderate	Moderate	SEM analysis; NR sample but quantitative model
[27]	Wang et al.	2025	Low	High	Moderate	High	High	Sample NR; qualitative; CHI workshop paper
[28]	Xhako et al.	2024	Low	Moderate	Low	Moderate	Low	N = 11 experts; expert eval; limited generalizability
[29]	Jangra et al.	2025	Low	Low	Low	Moderate	Low	N = 80; comparative; multiple conditions
[30]	Kong	2024	Low	High	High	Moderate	High	Co-design only; no formal evaluation; NR sample
[31]	Liang	2024	Moderate	High	High	Moderate	High	Doctoral colloquium; design research; no user study
[32]	Bekele	2021	Low	Moderate	Moderate	Moderate	Moderate	N = 11 experts; expert eval; technical focus
[33]	Bekele et al.	2021	Low	High	Moderate	High	High	Sample NR; custom questionnaire; cloud system
[34]	Boray et al.	2025	Low	High	High	Moderate	High	No user study; system description only
[35]	Du et al.	2024	Low	Moderate	Moderate	Moderate	Moderate	N = 6 specialists; expert interview; SIGGRAPH
[36]	Ch’ng et al.	2023	Low	Low	Low	Moderate	Low	N = 30 (10 groups); observation and interviews; social AR
[37]	Prandi et al.	2025	Low	Low	Low	Low	Low	N = 73 and 125; survey, and focus group; large public deployment
[38]	Souropetsis and Kyza	2025	Low	High	Moderate	High	High	Sample NR; custom questionnaire; single session
[39]	Vilar et al.	2025	Low	High	Moderate	Moderate	Moderate	SUS used; NR sample; web-based XR
[40]	Wu	2023	Low	High	High	High	High	Sample NR; minimal evaluation; conference paper
[41]	Zhang	2023	Low	High	High	High	High	Sample NR; custom Likert only; minimal detail
[42]	Serres et al.	2023	Low	High	Moderate	High	High	Sample NR; WebAR; custom questionnaire
[43]	Guimarães et al.	2015	Low	High	High	High	High	Sample NR; 2015 paper; limited eval detail
[44]	Gao et al.	2018	Low	Moderate	Moderate	High	Moderate	N = 10 pilot; small sample; early AR study
[45]	Chen et al.	2024	Low	Moderate	Low	Moderate	Low	N = 16 (8 pairs); comparative VR and SAR; controlled
[46]	Nikolakopoulou and Koutsabasis	2025	Low	High	Moderate	Moderate	Moderate	Participatory workshop; NR sample; co-design
[47]	Nikolakopoulou et al.	2022	Low	High	Moderate	High	High	Sample NR; custom questionnaire; MR and TUI
[48]	Zhang et al.	2023	Low	High	High	Moderate	High	No user study; SIGGRAPH system paper
[49]	Muñoz et al.	2025	Low	Moderate	Low	Moderate	Low	N = 6 and 9; SUS and formative; iterative design
[50]	Schauer and Sieck	2023	Low	High	High	Low	Moderate	Technical system description; no user study
[51]	Xiong et al.	2025	Low	High	High	Moderate	High	No user study; SIGGRAPH system paper
[52]	Di Feola and Rostami	2025	Low	High	Moderate	Moderate	Moderate	Qualitative design research; NR sample
[53]	Huang and Huang	2025	Low	Moderate	Low	Moderate	Low	N = 25 and 12; comparative; custom questionnaire
[54]	Costa et al.	2024	Low	High	High	Moderate	High	No user study; VR system description
[55]	Kebir et al.	2025	Low	High	Moderate	High	High	Sample NR; SUS and interviews; NR sample size
[56]	See et al.	2018	Low	High	Moderate	High	High	Sample NR; qualitative only; early VR study
[57]	Tong et al.	2024	Low	Moderate	Low	High	Moderate	PQ and interviews; NR sample; 360° novelty
[58]	Hu	2024	Low	Low	Low	Moderate	Low	N = 72; multi-platform comparative; good reporting
[59]	Tsita et al.	2023	Low	High	Low	High	High	Sample NR; SUS and SSQ used; single session
[60]	Yu et al.	2025	Low	High	Low	High	High	Sample NR; PQ and interview performed in a single session
[61]	Xie	2021	Low	High	High	Low	Moderate	Technical description; mobile VR; no user study
[62]	Cai and Yang	2023	Low	High	High	Moderate	High	No user study; system description only
[63]	Ji et al.	2021	Low	Low	Moderate	Moderate	Low	N = 20; observation and questionnaire; embodied learning
[64]	Zhang et al.	2024	Low	High	Moderate	High	High	Sample NR; custom Likert; AI guide novelty
[65]	Kobeisse	2023	Low	Low	Low	Low	Low	N = 80 and 572; large in situ; multi-method
[66]	Elrawi	2017	Moderate	High	High	Moderate	High	No user study; 2017 system paper; limited detail
[67]	Andrade	2023	Moderate	High	High	Moderate	High	Discourse analysis only; no empirical user study
[68]	Monteiro et al.	2023	Low	Low	Low	Moderate	Low	N = 20, 12, and 9; participatory and expert heuristics; multi-stage

.

Overall, 17 studies (27%) were rated as low-risk, 13 (21%) as moderate-risk, and 33 (52%) as high-risk. The dominant risk factor across the corpus was D2 (evaluation design): 40 out of 63 studies (63%) received a high-risk rating, reflecting the prevalence of unreported sample sizes, single-session evaluations, and convenience sampling. D4 (novelty bias) was the second most frequent concern, with 25 studies (40%) rated high-risk, given that most evaluations were conducted in first-use, lab-based settings where novelty effects on engagement and immersion cannot be disentangled from genuine system quality. D1 (system description) showed the strongest performance, with 60 studies (95%) rated low-risk, indicating that technical implementations were generally described with sufficient clarity. D3 (results transparency) was mixed: 21 studies (33%) reported raw data or quantitative statistics clearly (low-risk), 23 (37%) provided partial reporting (moderate), and 19 (30%) reported results without sufficient statistical detail (high-risk). Results are graphically presented in Figure 2.

A quick summary of the thematic analysis is presented in Figure 3. In this figure, the large variation of the subjects discussed is evident since all categories display a high degree of variability. In the next sections, each of these thematic dimensions is further discussed.

3.2. Technology

Hardware platforms, software engines, and content-creation methods fundamentally determine the fidelity, interactivity, and accessibility of XR heritage experiences. The corpus reports a range of devices, from consumer smartphones to specialized HMDs and projection systems, and content pipelines spanning manual 3D modeling, photogrammetry, structured-light scanning, and increasingly, AI-assisted workflows. Together, these technological choices shape an application’s visual fidelity, interactivity, performance, and scalability.

Table 5. Key technologies in XR for CH.

	Category	Analysis
Hardware platforms	HMDs	Oculus Quest/Meta Quest 2 [3,6,11,14,22,23,24,25,26,27,28,29,30,31,32] HTC Vive [7,12,13,28,32] HoloLens 1 and 2 [29,32,33,34,35] Nreal Light [9,36]. Google Cardboard [8,30] Sony PSVR [8] Transparent OLEDs (TOLEDs) [36] Pico 4 [27]
	Mobile devices	Smartphones and tablets for AR [1,6,11,12,30,36,37,38,39,40,41,42,43,44,45]
	Projection systems	Projectors for SAR, CAVE, 3D projection mapping systems [9,10,46,47,48,49]
	MR setups	Embodied/tangible setups [50,51,52,53]
Software frameworks	Game engines	Unity [3,6,7,8,9,11,12,14,24,25,28,29,33,37,40,42,45,46,50,54,55,56,57,58,59] Unreal Engine [25,27,54] Meta SDK [52]
	Web	WebGL and WebXR for browser-based experiences [22,41,43] NFC and QR codes [11,15,39] Image recognition for AR [40,43] ApacheCordova [38]
	AR/MR SDKs	ARCore [37,41,43] ARFoundation [6] AR.js [38] Oculus SDK [30] Vuforia [7,12,42,45,51] XR Interaction Toolkit [11,55] Meta XR [28] SceneVR, D’Fusion, J-Monkey, and EON Reality [59]
Content-generation techniques	3D modeling and scanning	Photogrammetry and aerial photogrammetry [1,2,7,11,15,22,25,28,37,53,54,56,60] 3D Laser Scanning [1,2,29] Terrestrial and UAV photogrammetry [15,61] 3D structured light scanning [2,62] Vuforia area targets (LiDAR) [51] 3D modeling software (ZBrush, 3ds Max, Blender, VRoid Studio) [2,3,11,12,13,25,31,48,59,61,63,64,65] 360-degree photography [25,26,57,58] CAD software [12] 3D printing [11,12,50]
	AI methods	Generative AI for artifact restoration and natural communication [5,24,58]
	Asset creation	Adobe Photoshop for textures and Wacom tablets for manual digital restoration [11,57,63]
Supporting infrastructure	Cloud services	Amazon S3, Azure Spatial Anchors, Azure Cosmos DB for multiuser synchronization and content management [28,33]
	Sensors and trackers	GPS and Visual Positioning Service (VPS) for location-based AR [36,38,41] Anti-latency trackers for precision [50] Leap Motion for hand tracking [12,25,27] Kinect for body tracking [57]

catalogs the main platforms, development frameworks, content workflows, and supporting services identified in the reviewed literature.

The following charts present the distribution of works in terms of hardware, software, content generation, and supporting infrastructure (see Figure 4, Figure 5, Figure 6 and Figure 7, respectively).

3.3. Interaction Styles and Paradigms

Interaction design mediates the ways in which users navigate virtual environments, manipulate artifacts, and access narrative content. Across the reviewed projects, there is a clear movement from controller-based and screen-based interfaces toward more natural, embodied, and collaborative paradigms, enabled by improvements in hand tracking, spatial sensing, and networking. Key interaction paradigms identified in the sources include the following: single-user versus multiuser systems that support social collaboration; a variety of input methods such as gaze, gesture, and speech; and different narrative structures, from linear, guided storytelling to nonlinear, user-driven exploration. A particularly notable trend is the move toward embodied and tangible interactions, which use physical movements and objects to control the virtual experience, creating a stronger connection between the user’s body and the digital content. The interaction paradigms employed are presented in

Table 6. Reported user experience outcomes.

Interaction Paradigm	Description and Purpose	Analysis
Exploratory navigation	Single-user navigation in VR/AR.	[2,13,15,25,27,28,39,57,60,65]
Embodied interaction	Using the user’s own body movements, gestures, and physical actions as the primary input.	[8,25,27,47,63]
Tangible user interfaces (TUI)	Employing physical objects, props, or replicas as interfaces to manipulate digital content.	[23,25,29,42,48,50,51,52,66]
Multiuser collaboration	Systems are designed for multiple users to interact simultaneously within a shared virtual or mixed-reality space.	[1,7,25,28,34,37]
Gamification	Integration of game mechanics (e.g., challenges, rewards, scoring) and serious games to enhance motivation, engagement, and learning outcomes in a playful context.	[6,12,39,40,41,42,44,48]
Multimodal interaction	Combining multiple input modalities such as gaze, gesture, speech, and touch to create a more natural and flexible user interaction experience.	[24,25,27,34,57]
Interactive storytelling	Structuring the experience around a narrative that can be linear (guided) or nonlinear, allowing users to influence the story’s progression through their actions and choices.	[1,8,10,14,26,42,45,67,68]
	Virtual guide (3D character) to deliver curated information in a personified and engaging manner.	[41,60,65]
	AI-powered communication to support natural language conversation with virtual characters.	[65]
	Adaptive narrative (personas): personalize storytelling content based on user background and interests.	[42,49,58,61]
	Contextual storytelling: immerse users in a narrative by animating historical scenes.	[14,49,58,61]

.

The chart in Figure 8 presents the distribution of works in terms of interaction.

3.4. User Experience (UX) Dimensions

User experience (UX) is a primary concern in XR heritage work: immersion, presence, engagement, and learning outcomes are often central project goals. A detailed analysis is provided in

Table 7. Reported user experience outcomes.

UX Dimension	Key Findings (Positive and Negative)	Analysis
Immersion and presence	Positive: HMD-based VR and MR systems are consistently reported inducing high levels of immersion and presence, making users feel physically and psychologically “in” the virtual environment. Wearable VR outperforms mobile VR in immersion.	[2,8,13,25,30,58,60,61,65]
Engagement	Positive: interactive and gamified elements significantly increase user engagement, motivation, and time spent with the content. Multiuser systems enhance social engagement.	[4,6,14,39,40,45,49,61,66]
Learning and knowledge retention	Positive: XR applications, particularly those with interactive and game-based elements, are shown to improve learning outcomes, knowledge retention, and understanding of cultural context.	[1,6,13,15,23,28,34,39,41,42,48,56]
Emotional connection and empathy	Positive: immersive storytelling and virtual embodiment can foster empathy and a strong emotional connection to historical events and heritage, enhancing social cohesion.	[24,38,49,53,58,61]
Accessibility and inclusivity	Positive: XR can make inaccessible sites visitable and provide multisensory experiences (e.g., tactile and audio) for users with visual impairments.	[1,11,14,30,36,39,42,43,62]
Usability and cognitive load	Negative: users can experience cognitive load when interfaces are complex or when technology overshadows the cultural content. Challenges with navigation, gestures, and controllers are common.	[3,4,12,15,26,29,34]
Cybersickness and discomfort	Negative: motion sickness (nausea, dizziness, eyestrain) remains a significant barrier, especially in VR systems. Physical discomfort from headsets is also reported.	[3,24,25,30,57,60]

.

User experience outcomes reported by the research works studied are quantitatively summarized in Figure 9.

The literature reports instances where XR increased motivation, understanding, and retention, as well as cases where usability issues, such as confusing controls or cognitive overload, reduced the educational value. Positive outcomes reported include enhanced presence and engagement, particularly for HMD-based systems, and gains in learning and empathy when narrative or embodied interactions were used. Common negative outcomes include cybersickness, interaction complexity, and the possibility that technological novelty distracts from cultural interpretation.

3.5. Evaluation Methodologies

Robust evaluation practices are needed to substantiate claims about XR’s educational and cultural benefits. The corpus shows a mix of qualitative and quantitative approaches that, when combined, can provide richer evidence than alone. Qualitative methods (semi-structured interviews, thematic analysis, participatory workshops) are widely used to capture user perceptions and to include stakeholder perspectives. Quantitative instruments frequently include standardized scales, such as the System Usability Scale (SUS), the Simulator Sickness Questionnaire (SSQ), the Presence Questionnaire (PQ), and the Game Experience Questionnaire (GEQ), alongside custom Likert-scale surveys and performance metrics (task time, accuracy). The usage of these methods in the studied corpus is presented in

Table 8. Common evaluation frameworks and instruments.

Method/Instrument	Analysis
Qualitative methods
Semi-structured interviews and focus groups	[2,6,13,25,28,34,41,49,50,56,57,58,60,61,65]
Observation and thematic analysis	[2,8,13,25,28,50,58,60,61,66]
Co-design and participatory workshops	[10,29,31,33,39,40,47,69]
Quantitative methods and instruments
System Usability Scale (SUS)	[6,24,26,40,50,56,60]
Game Experience Questionnaire (GEQ)	[6,23]
Presence Questionnaire (PQ)	[24,58,61]
Simulator Sickness Questionnaire (SSQ)	[24,25,30,57,60]
Immersion Experience Questionnaire (IEQ)	[23]
Custom questionnaires (Likert scales)	[6,12,14,26,27,29,34,39,42,43,49,65,66]

. The distribution of works in terms of qualitative and quantitative methods is presented in Figure 10 and Figure 11, respectively.

The risk of selective reporting bias across the synthesized corpus is considered to be moderate–high. Publication bias likely favors studies reporting positive usability and engagement outcomes, as negative or null results are less frequently submitted or accepted for publication. The inclusion of conference proceedings alongside journal articles was intended to partially offset this tendency. Within studies, outcome reporting bias is evident in the predominance of self-report measures and the underreporting of negative UX outcomes such as cybersickness, interaction failure, and disengagement. These limitations are reflected in the high-risk D4 (novelty bias) ratings assigned to 40% of studies in the risk of bias assessment (Section 3.1.2).

4. Analytical Synthesis

In the previous chapter, an analysis of the identified thematic categories was conducted, followed by a quantitative analysis of their representation in the works studied. This section goes one step further and provides an analytical synthesis of the reviewed literature. Rather than aggregating quantitative effect sizes, the synthesis adopts an interpretive and comparative approach to identify recurring patterns, design configurations, and cross-cutting relationships that emerge across studies. This approach is consistent with systematic reviews in interdisciplinary fields where methodological heterogeneity and context-specific outcomes preclude formal statistical meta-analysis. The analytical synthesis focuses on the ways in which technological choices, interaction paradigms, and experiential design strategies are combined in practice; additionally, there is a focus on the ways in which these combinations shape user experience, learning outcomes, and cultural interpretation in heritage contexts. By examining convergences and divergences across platforms, application domains, and evaluation settings, this section aims to move beyond isolated case descriptions toward a more integrative understanding of how XR is currently being conceptualized and deployed in CH.

Importantly, this synthesis also foregrounds tensions and trade-offs that recur across literature, including those between immersion and usability, realism and interpretation, technological novelty and cultural meaning. In doing so, it provides a conceptual bridge between the empirical findings summarized in the Results Section and the critical reflection developed in the Discussion Section, offering a structured lens through which the opportunities and limitations of XR in CH can be more systematically interpreted.

The central analytical finding of this synthesis is what we term the Design Coherence Framework for XR Heritage (DCF-XR). Across the 64 reviewed studies, the most consistently successful XR heritage experiences (measured by user engagement, learning outcomes, and cultural authenticity) share a common structural feature: the technology deployed, the interaction paradigm chosen, the experiential and interpretive goals pursued, and the evaluation instruments selected to assess them are mutually reinforcing rather than independently optimized. Coherence across all four dimensions (technology, interaction design, user experience, and evaluation) is not merely a quality assurance concern; instead, it is the structural condition under which XR can fulfill its cultural heritage mission. Conversely, the most frequently cited failure patterns across the corpus (cybersickness, shallow engagement, low cultural meaning, and evaluations that measure the wrong outcomes) correspond not to the weakness of any single dimension in isolation, but to configurations where one or more dimensions are misaligned with the others. This coherence principle is not self-evident from individual case studies; it emerges only through cross-corpus synthesis. The four failure modes that typify these misalignments are examined systematically in Section 4.6, and together they constitute the primary conceptual contribution of this review.

4.1. Technology

In terms of technology the works studied showcase a wide range of options. From these ranges, some recurring patterns can be identified, mainly regarding the combination of technologies to achieve certain effects. The explorations of XR tools together with complementary systems can be categorized as follows: (1) hybrid and collaborative XR; (2) XR integrated with AI, cloud, and computational services; (3) digitization and visualization platforms; (4) mobile and location-aware AR/XR.

4.1.1. Hybrid and Collaborative XR

The below-described combinations focus on systems that merge different reality technologies (VR, AR, MR) or incorporate hardware solutions designed specifically for shared and embodied interactions, aiming to overcome the typically solitary nature of headset experiences:

• VR, SAR, and bidirectional interaction: This grouping enables collaborative exhibitions where VR users interact with content that is simultaneously projected into a public space via SAR, allowing other participants to influence the VR environment and vice versa [46].
• Hybrid VR, AR, tangible user interface (TUI), and multiuser interaction: This is a system combining VR headsets (HTC Vive) and mobile AR apps, allowing co-located users with asymmetrical access to communicate and interact around shared digital heritage objects, mediated by a physical, recognizable object (an AR cube) [7].
• MR, TUI, Substitutional Reality (SR), and hand tracking: This blend, exemplified by the School House Virtual Museum, integrates physical proxy objects (like a desk or buttons) placed in the real world with virtual objects in a VR environment (Oculus Quest 2), enhancing immersion and providing passive haptic feedback via natural hand tracking [23].
• MR, TUI with 6 DoF tracking, and spatial audio: Installations such as LanternXR couple a physical replica of an artifact with handheld, tracked props (a camera and a candlestick) using highly accurate sensors (anti-latency), displaying the interactive scene on large, high-resolution monitors accompanied by spatial audio [50].
• Hybrid MR, tangible controller, and motion capture: This hybrid installation approach promotes intangible CH by coupling a physical, simplified mechanical controller (like a crane model containing an ESP32 microcontroller, rotary encoder, and switch) with a virtually reconstructed environment featuring characters animated using motion capture data [48].
• MR, hand tracking, and spatial mapping/scene understanding: The Falconry Heritage prototype leverages the Microsoft HoloLens 2’s advanced sensor array for environmental tracking and real-time scene understanding, integrating the virtual falcon’s hunting logic with the user’s natural hand gestures [35].

4.1.2. XR Integrated with AI, Cloud, and Computational Services

This trend focuses on using network architecture and intelligent systems to enhance XR functionality, particularly for real-time operation, data handling, and dynamic user response:

• MR, cloud computing, and multimodal interactions: Such applications integrate MR display devices (Microsoft HoloLens) with cloud services (e.g., Azure Spatial Anchors, Amazon Polly, Amazon S3, Azure Cosmos DB) for managing distributed content, ensuring persistence of shared experiences, and processing multimodal input (gaze, gesture, speech) [33,34].
• VR, AI-driven natural language understanding (NLU), knowledge graph (KG), and emotional avatar: A VR system (Oculus Quest 2) utilizes AI to analyze users’ verbal input for emotional and moral values (NLU/KG), dynamically generating a personalized conversational avatar to promote perspective-taking in relation to cultural interpretation [24].
• Generative AI, AR, data science, and 3D assets: A proposed preservation system uses generative AI (like GANs or diffusion models) to digitally restore damaged artifacts, relying on data science techniques to verify historical accuracy, and then leverages AR to visualize these reconstructed 3D models in a real-world context [5].

4.1.3. Digitization and Visualization Platforms

These combinations form the technical bedrock for creating, optimizing, and presenting high-fidelity digital assets:

• 3D digitization and game engines: The standard approach for generating immersive CH assets involves acquiring precise spatial data through methods like laser scanning, drone photogrammetry, or Structure-from-Motion (SfM), followed by processing and rendering within real-time platforms such as Unity or Unreal Engine [6,56].
• Web3D/WebXR, pixel shaders, and progressive refinement: This web-based technique uses GLSL shaders to enable localized, interactive transitions between different time periods within a 3D scene, employing efficient data handling methods like spatial indexing (BVH trees) and progressive streaming [22].
• VR animation, 3D modeling software, and image editing: This combination is used specifically for ICH projects, combining multiple 3D modeling tools to build virtual environments and traditional objects and integrating 2D graphical work (e.g., texture mapping in Adobe Photoshop) for digital restoration and realistic appearance [63].

4.1.4. Mobile and Location-Aware AR/XR

• XR, Visual Positioning Service, GPS, Transparent OLED, and Head-Mounted Display (HMD): This system uses a fusion of global GPS data (for approximate location) and high-accuracy VPS (for precise AR content anchoring) in combination with both HMDs (Nreal Light) and Transparent OLED screens (TOLED) for visual output in a moving vehicle context [36].
• Mobile marker-based AR, tangible artifact, and location-based information: This hybrid mobile intervention uses the smartphone’s internal sensors (GPS, gyroscope, accelerometer) for location-based AR and relies on a physical installation structure that doubles as a recognizable marker (marker-based AR) when precision is needed [38].
• Mobile AR, image recognition, GPS, and game engine: This widely adopted approach uses image recognition technology (via Vuforia) to trigger virtual content (like AR figures or augmented information) anchored to physical objects, with potential augmentation from GPS data for broader localization [45].

4.2. Interaction Styles and Paradigms

Across the reviewed corpus, interaction design emerges as a critical mediator between XR technology and cultural meaning-making. Rather than a uniform interaction model, the literature reveals a set of recurring interaction paradigms, each associated with distinct experiential affordances and limitations. These paradigms can be grouped into four higher-level categories: exploratory navigation, embodied and tangible interaction, social and collaborative interaction, and narrative-driven interaction.

Exploratory navigation remains the most prevalent paradigm, particularly in VR and mobile AR applications that prioritize the spatial exploration of reconstructed sites or artifacts. While this approach supports self-directed learning and personal pacing, its effectiveness is strongly dependent on spatial clarity, locomotion design, and interface simplicity. Studies consistently report that poorly designed navigation mechanics increase cognitive load and cybersickness, undermining interpretive goals.

Embodied and tangible interactions represent a significant shift toward leveraging the user’s body and physical artifacts as primary input channels. These approaches, enabled by hand tracking, physical proxies, and tangible user interfaces, strengthen the coupling between physical action and digital response, reinforcing authenticity and experiential engagement. However, the meta-analysis shows that embodied interaction is most effective when interaction mappings are simple and metaphorically aligned with cultural practices. Overly complex gesture vocabularies or abstract mappings frequently lead to confusion and usability breakdowns.

Social and collaborative interaction constitutes a growing trend, particularly in multiuser VR, AR co-viewing, and hybrid VR–SAR installations. These systems move XR beyond individual experience toward shared interpretation and collective sense-making. The literature suggests that social presence amplifies engagement and discussion but introduces additional design challenges, including synchronization, role asymmetry, and coordination overheads. Successful implementations often employ asymmetrical roles or shared tangible anchors to reduce interaction complexity.

Narrative-driven interaction, including guided storytelling, adaptive narratives, and virtual guides, plays a central role in framing cultural interpretation. Linear narratives support structured learning and reduce cognitive burden, whereas nonlinear or adaptive narratives enable personalization and deeper exploration. The meta-analysis indicates that narrative mechanisms are most effective when integrated with interaction design rather than layered on top of it; disjointed narration and interaction often result in fragmented user experiences.

4.3. User Experience Outcomes

The synthesis of user experience (UX) outcomes across the reviewed studies reveals a consistent pattern: XR technologies can substantially enhance immersion, engagement, and learning in CH contexts, yet these benefits are highly contingent on interaction design quality, content fidelity, and evaluation methodology.

Immersion and presence are most strongly associated with head-mounted VR and MR systems, particularly when combined with embodied interaction and coherent narrative framing. High levels of presence correlate with emotional engagement and empathy, especially in experiences involving historical reconstruction or performative heritage. However, the meta-analysis also shows that increases in visual fidelity or realism do not automatically improve immersion if accompanied by discomfort, interaction friction, or perceptual inconsistencies.

Engagement emerges as a multifaceted outcome influenced by interactivity, gamification, and social participation. Game-like mechanics and challenges reliably increase motivation and time-on-task, particularly for younger audiences. Social XR experiences further amplify engagement by enabling shared discovery and discussion. At the same time, several studies caution that extrinsic engagement mechanisms may overshadow interpretive depth if they are not carefully aligned with curatorial intent.

Learning and knowledge retention are frequently reported as positive outcomes, especially in XR applications that encourage active exploration, manipulation, or narrative participation. The meta-analysis suggests that learning gains are strongest when interaction requires users to make meaningful choices or perform culturally relevant actions, rather than passively consume information. Nevertheless, learning outcomes are often inferred from self-reported measures, limiting the strength of evidence.

Emotional connection and empathy are particularly prominent in narrative-driven and embodied XR experiences. Virtual embodiment, perspective-taking, and affective storytelling can foster empathy toward historical actors or cultural practices. However, these effects are context-dependent and may be sensitive to representational accuracy and cultural framing, reinforcing the importance of authenticity and community involvement.

Negative UX outcomes, including cybersickness, cognitive overload, and frustration, remain widespread. These issues disproportionately affect first-time users and can negate otherwise positive experiential gains. Importantly, the meta-analysis indicates that negative UX effects are rarely isolated technical problems; they often result from misalignments between technology, interaction design, and user expectations.

4.4. Evaluation Methods

Mixed-method approaches enable triangulation of usability and learning outcomes, but heterogeneity in evaluation instruments, small sample sizes, and short-term studies often hinder direct comparisons across projects. The participant numbers and corresponding studies are categorized below according to sample size and purpose:

Small/expert/formative cohorts (N < 20): Studies with these cohorts typically rely on highly specialized individuals for in-depth qualitative feedback or initial usability testing, acknowledging the value of expert insight in developing complex XR systems. Examples of this are described below:

• The formative usability trial for the LanternXR system utilized nine laboratory staff members [50]. The subsequent LanternXR expert evaluation used a focus group of six specialists, including three curators, an art historian, and technical researchers [50].
• An initial pilot study on campus heritage utilized two groups of prospective students and their parents, alongside five current full-time students [45]. The prototype assessment for the campus heritage AR app included 10 current students [45].
• The fundamental evaluation of the collaboratively designed MR application involved a total of 11 experts (curators, archaeologists, and researchers) across two sessions [33].
• A key part of the social AR intervention focusing on the umarell phenomenon involved a focus group of 15 young adults to shape the design process [38].
• The second expert evaluation of the Media Bus XR prototype involved semi-structured interviews with six specialists in tourism, XR, HCI, and digital heritage [36].
• The qualitative analysis on embodied interaction for intangible CH selected 20 students as learners, each experiencing the system for 10 to 15 min [64].
• In the research on AR integration heuristics, initial data collection involved 20 participants, followed by intensive qualitative work with 12 participants for interviews, culminating in a participatory workshop with 9 participants [69].

Moderate user studies (N 20–62): Studies utilizing participants in this range are typical of controlled usability studies and focused empirical research, often employing convenience samples such as students, staff, or a limited number of visitors. Examples of this are described below:

• The social AR study involving co-viewing heritage objects included a total sample of 30 participants (organized into 10 groups with 3 participants each) [37].
• The large-scale tactile and XR accessibility project leveraged a final test campaign involving 30 users (20 male, 10 female) ranging in age from 11 to 72 years [11].
• The usability assessment for the VR system enhancing social cohesion utilized a sample of 30 museum visitors [24].
• The hybrid VR and AR acceptance model analysis involved 52 users (organized into 26 pairs) [7].
• The VR collaborative exhibition system based on Spatial AR included a comparative study with 16 participants (8 pairs) [46].
• The evaluation of the comprehensive virtual museum system for the schoolhouse focused on 62 participants [23]. The evaluation of immersion and educational potential for the virtual schoolhouse included 25 participants (13 students and 12 educators) [23].
• The comparative analysis of Interactive Virtual Museum formats utilized a total sample of 30 school participants and 12 museum visitors, yielding a valid school group sample of 25 participants [54].
• The original user study for the campus heritage AR prototype included 10 current students [45].

Large cohorts/public demonstrations (N > 70): These studies achieved high user numbers through public exhibitions, non-selective data collection over time, or large-scale comparative testing. Examples of this are described below:

• A large-scale study exploring the impact of VR on museum experiences involved a total of 80 participants (architectural students and faculty members), divided equally between mobile VR and wearable VR groups [30].
• The framework evaluation involving tangible AR interfaces utilized data collected from an in situ study featuring 80 visitors (40 female and 40 male) who completed a paper questionnaire during a two-week installation [66]. Additionally, records showed that 572 visitors interacted with the display based on digital survey entries [66].
• Testing of the developed individual VR/AR platform documented a total of 72 observations collected from surveyed users over three days [59].
• The comparative analysis of three XR applications related to the Dacian Bronze Matrix utilized 37 filled questionnaires for each solution, totaling 111 questionnaires [12].
• An intervention designed to promote the umarell cultural phenomenon publicly demonstrated the system to approximately 125 teenagers and young adults and collected data via a survey involving 73 participants [38].

In a nutshell, the evaluation methodologies applied have provided significant results regarding the pros and cons of using XR in the CH context. However, the methods themselves are also considered a current limitation since in order to strengthen evidence, researchers should pursue larger, more diverse cohorts and longitudinal designs that assess retention and behavioral change (see Figure 12).

4.5. Cross-Cutting Result

Across all four dimensions, technology, interaction, user experience, and evaluation, the meta-analysis reveals a common underlying pattern: successful XR heritage applications are those that achieve coherence between technological affordances, interaction paradigms, experiential goals, and assessment methods. Breakdowns typically occur when one dimension dominates the others, such as technologically sophisticated systems with weak interaction design or engaging experiences evaluated with insufficient methodological rigor. This finding underscores the need for holistic design and evaluation frameworks that treat XR heritage experiences as socio-technical systems rather than isolated technological artifacts.

Across the four synthesized thematic domains, contributing studies varied substantially in design type, sample size, XR modality, and CH context. The majority of studies contributing to the technology theme were journal articles reporting implemented systems (n = 48), while the interaction and UX themes drew equally from conference and journal sources. Studies contributing to the evaluation theme were disproportionately small-scale (N < 30), with only eight studies reporting samples exceeding 70 participants. No statistical heterogeneity assessment was conducted given the qualitative nature of the synthesis. Results of sensitivity analyses are not applicable. The overall pattern across syntheses confirms that high-risk-of-bias studies (n = 33, 52%) were distributed across all four thematic domains, indicating that the limitations identified are systemic rather than domain-specific.

4.6. A Typology of Design Failure Modes

The cross-domain synthesis presented in Section 4.5 not only reveals what makes XR heritage applications succeed but also explains why so many fall short of their cultural mission. A close reading of the corpus reveals that design shortcomings are not distributed randomly across the literature; rather, they cluster into four identifiable failure modes that re-appear across otherwise different studies, XR modalities, and CH contexts. These failure modes are best understood through the lens of the four-dimensional Design Coherence Framework for XR Heritage (DCF-XR), which holds that effective XR heritage applications are those in which all four thematic dimensions (technology, interaction design, user experience, and evaluation) are mutually aligned and oriented toward the same interpretive and cultural goal. When one or more dimensions are optimized in isolation, disconnected from the others, a characteristic failure pattern emerges. The four failure modes identified below each correspond to a specific dimensional misalignment within the DCF-XR.

Failure mode 1—technologically rich, interactionally poor: The most prevalent pattern in the corpus is the deployment of high-fidelity XR environments, featuring detailed photogrammetric reconstructions, high-resolution HMD rendering, or sophisticated spatial mapping, coupled with passive or underdeveloped interaction design. In these cases, users are positioned primarily as spectators navigating a visually impressive but experientially shallow environment, typically through basic locomotion controls or pre-scripted waypoints. Studies exhibiting this pattern consistently report high initial curiosity and novelty-driven engagement, but shallow learning outcomes and limited cultural meaning-making. The risk-of-bias analysis (Table 4) further shows that these studies disproportionately rely on first-use, single-session evaluations, where novelty effects mask the absence of deeper interpretive value. Within the DCF-XR, this failure mode reflects a condition in which the technology dimension is highly developed while the interaction design dimension remains underdeveloped; as a consequence, neither the user experience nor evaluation dimensions can be meaningfully engaged with. The pattern suggests a fundamental design inversion: technological investment precedes and overshadows interaction design rather than being guided by the cultural and experiential goals it should serve.

Failure mode 2—narratively strong, evaluatively absent: A second recurring pattern involves experiences with well-developed narrative and interaction design (engaging storytelling, adaptive personas, embodied gestures, and culturally grounded content) that are released without any formal user evaluation. This pattern is particularly prominent among SIGGRAPH and CHI system papers, which prioritize creative or technical innovation over empirical evidence. From the DCF-XR perspective, these works demonstrate that the technology, interaction design, and user experience dimensions can be well-conceived and internally coherent, yet they fail to contribute to cumulative knowledge because the evaluation dimension is entirely absent. Without evidence of whether the intended cultural experience was actually delivered (to whom, under what conditions, and with what outcomes), the contribution remains a demonstration rather than a validated design. The systemic nature of this pattern reflects a structural tension between publication venue cultures: creative and technical venues reward innovation and demonstration, while empirical rigor remains a secondary concern.

Failure mode 3—evaluatively rigorous, culturally thin: A third pattern is, in some ways, the inverse of the second. Several studies in the corpus employ validated evaluation instruments (SUS, SSQ, PQ, and GEQ) and adequately sized samples, but restrict their outcome measures exclusively to technical usability, simulator comfort, or generic engagement. Cultural learning, heritage interpretation, emotional connection to place or practice, and community relevance are not measured at all, leaving the cultural value of the XR intervention entirely unassessed. Within the DCF-XR, this failure mode reveals a misalignment between the evaluation dimension and the user experience and cultural goals it ought to serve: the evaluation apparatus is methodologically sound in isolation but incoherent with the interpretive purpose of CH applications. This finding carries an important implication for the field: evaluation rigor is necessary but not sufficient. The choice of what to evaluate is as consequential as how rigorously it is conducted, and the persistent absence of CH-specific validated evaluation instruments (noted in Section 3.5) directly enables this failure mode to persist.

Failure mode 4—participatory in design, solitary in use: A fourth pattern is visible across a subset of studies that employ co-design, participatory workshops, or community consultation during the development phase, but ultimately deliver single-user, non-social XR experiences that do not preserve the communal and relational dimensions of the heritage practices they represent. This is particularly problematic for intangible cultural heritage (ICH), where heritage meaning is inherently embedded in collective practice, social transmission, and shared performance. When the participatory ethos of the design process is not carried forward into the interaction architecture of the final system, the application may be technically and culturally authentic in its content, yet it is socially decontextualized in its form. Within the DCF-XR, this failure mode reflects a disconnection between the interaction design dimension, which was shaped by community input, and the user experience dimension, which ultimately unfolds as an individual rather than collective encounter. Addressing this pattern requires carrying community-centered values not only into the design process but into the operational structure of the experience itself, for example through multiuser affordances, shared narrative roles, or community-editable content layers.

Synthesis: The DCF-XR is presented as a design diagnostic.

Table 9. Failure mode typology mapped to the DCF-XR dimensions.

Failure Mode	Technology	Interaction Design	User Experience	Evaluation
1. Technologically rich, interactionally poor	Strong	Weak	Shallow	Novelty-only
2. Narratively strong, evaluatively absent	Strong	Strong	Intended	Absent
3. Evaluatively rigorous, culturally thin	Present	Present	Unmeasured	Technically present, culturally misaligned
4. Participatory in design, solitary in use	Present	Co-designed	Individualized	Partial

summarizes the four failure modes in terms of which DCF-XR dimensions are present and which are absent or misaligned in each pattern.

The typology presented above is offered not as a criticism of individual studies, many of which make valuable contributions within their own scope, but as a structural account of the field’s most persistent and cross-cutting shortcomings. Crucially, no individual study can address all four dimensions simultaneously without a deliberate, integrative design commitment from the outset. The DCF-XR provides practitioners and researchers with a practical diagnostic lens: before a system is deployed, each of the four dimensions should be explicitly addressed, and their mutual coherence should be treated as the primary design criterion. This integrative, multi-stakeholder, longitudinally evaluated design practice is precisely what this review consistently identifies as the most underrepresented approach in the current XR heritage literature.

5. Discussion

The overall certainty of evidence synthesized in this review is considered to be low–moderate. While findings are consistent in direction XR technologies generally enhance immersion, engagement, and learning in CH contexts the strength of this evidence is limited by the methodological characteristics of the contributing studies: the majority rely on small convenience samples, single-session evaluations, non-validated instruments, and first-use novelty conditions. These factors, documented in the risk of bias assessment (Section 3.1.2), reduce confidence in the generalizability and durability of reported outcomes. Findings should therefore be interpreted as indicative of promising trends rather than established effects.

The studied sources demonstrate a clear transition from early proof-of-concept XR deployments, that we have witnessed in the past, toward more mature and complex user-centered systems that emphasize social presence, embodiment, and multimodal interaction. This evolution reflects a tendency to align XR technologies with the interpretive, educational, and experiential goals of modern museums and CH institutions. In this context, modern XR applications seem to transition from being mere technological showcases to becoming a part of broader curatorial strategies; here, they are employed in richer means to support contextualization, participation, and experiential learning.

At the same time, this review reveals that, even if technology has achieved an increased maturity level, its application cannot always be automatically translated to cultural or educational effectiveness. Across the corpus, progress in immersion, engagement, and learning are frequently accompanied by persistent design, methodological, and ethical challenges. These findings highlight the need to not only determine what XR technologies can do but also identify why they are deployed in heritage contexts.

The findings of this review can be situated within the broader CH scholarship that frames heritage as an interpretive, socially constructed process rather than a fixed transmission of the past. From this perspective, XR systems function not merely as visualization technologies but as mediating instruments that shape the ways in which heritage is selected, narrated, and experienced. Distinctions commonly drawn in heritage studies such as between living and archived heritage, or between authoritative and participatory forms of interpretation are directly implicated in XR-based applications. Highly curated, high-fidelity reconstructions may privilege singular historical narratives, while participatory and co-designed XR experiences offer opportunities to surface multiple voices and evolving meanings and thus proven of extreme significance for opening access and debate of heritage subjects. Issues of authenticity in XR should therefore be understood not only in terms of visual realism or technical accuracy, but also in relation to interpretive transparency, cultural ownership, and the legitimacy of representation. Relating XR design choices within these heritage-related theoretical concerns helps clarify why community participation, narrative framing, and ethical co-creation are not optional enhancements, but central to culturally responsible XR deployment. These heritage-related theoretical considerations provide a lens through which the limitations and implications discussed below can be more critically interpreted.

The Design Coherence Framework for XR Heritage (DCF-XR) proposed in Section 4 has implications that extend beyond the XR-CH domain. The framework resonates with established theoretical positions in interaction design, particularly the concept of experience-centered design advanced by McCarthy and Wright [70]; this holds that the quality of an interactive experience cannot be reduced to usability or technical performance but must be evaluated in terms of the meaning it enables users to construct. It equally connects with foundational work in heritage studies, notably Smith’s critique of the “authorized heritage discourse” [71], which argues that the meaning of heritage is not inherent in artefacts or sites but is actively constructed through interpretive practice, social engagement, and community ownership. Taken together, these theoretical anchors illuminate a risk that the corpus documents empirically: XR systems that prioritize technical spectacle over interpretive scaffolding risk reproducing a passive, object-centered heritage discourse in an immersive register, delivering presence without meaning. The findings of this review suggest that the field is only partially moving toward participatory, meaning-centered XR design. This transition is hampered as much by the culture of evaluation as by technology itself. As long as the dominant evaluation instruments measure comfort and usability rather than cultural interpretation and community relevance, the evaluation dimension of the DCF-XR will remain misaligned with the other three, and the systemic failure modes identified in Section 4.6 will persist.

5.1. Persistent Limitations

In the here-studied works, both significant progresses and recurring limitations are prominent. In summary, the main limitations that continue to constrain the broader impact and sustainability of XR in CH can be highlighted as follows: (1) technological and resource barriers; (2) usability and user experience challenges; (3) perceptual and immersion gaps; (4) issues of authenticity and representation; (5) accessibility and inclusivity gaps; and (6) limitations in evaluation rigor. Each of these are individually discussed below.

5.1.1. Technological Barriers

Cost, availability, and performance constraints remain substantial obstacles, particularly for small- and medium-sized institutions, constraining the wide applicability of the developed solutions. High-end HMDs, tracking systems, and spatial computing infrastructures impose financial and logistical burdens, while latency, low frame rates, and tracking inaccuracies can undermine presence and user comfort. Overcoming these obstacles can enhance the visibility, acceptability and impact of small- and medium-sized museums, making them more competitive and experientially richer in relation to their larger and less versatile counterparts.

Content creation presents an equally significant challenge. High-fidelity digitization pipelines based on photogrammetry or LiDAR require extensive post-processing to achieve real-time performance, often involving mesh decimation and texture baking that may introduce visual artifacts and reduce perceived authenticity. This gives emphasis to the need to further develop and apply more cost-effective AI-based solutions; plain RGB data can enhance digitization accuracy, minimizing digitization costs. For intangible and performative heritage, additional difficulties arise in capturing the temporal, embodied, and contextual dimensions, demanding rich metadata, careful editing, and adaptive presentation strategies to avoid decontextualization.

5.1.2. Usability and User Experience Challenges

Of all the challenges discussed in this research work, usability challenges are the most critical: they can result in disruptions to immersion and presence, destroying the experience for users. Among all persistent user experience challenges, the following, reported in the here-studied works, are categorized as high-risk challenges:

• Cybersickness, non-intuitive interaction techniques, and steep learning curves remain major barriers to adoption, particularly for first-time or casual users. Cybersickness continues to limit session duration and suitability for general audiences, especially in locomotion-heavy VR experiences.
• Interface intuitiveness is not guaranteed by so-called “natural” interaction techniques; several studies indicate that familiar inputs such as touchscreens or controllers may outperform gesture-based interfaces for novice users in museum settings.
• Interaction complexity often necessitates explicit onboarding, increasing cognitive load and potentially detracting from interpretive goals. Discoverable controls, clear feedback, and progressive disclosure are therefore critical design strategies.
• Closely related is the issue of cognitive and perceptual load. Excessive density of digital overlays, particularly in AR, has been shown to increase task-completion times, error rates, and perceived effort.

Designers and curators must balance informational richness with interpretive clarity to ensure that XR enhances, rather than competes with, the heritage asset itself. In this regard, several studies caution against technological spectacle; here, novelty and visual impact can overshadow cultural meaning, turning XR into a barrier to understanding rather than a facilitator.

5.1.3. Perceptual and Immersion Gaps

Even when usability issues are mitigated, immersion can be compromised by inconsistencies in visual fidelity and behavior. Flat or poorly shaded replicas, unrealistic object dynamics, and limited environmental responsiveness disrupt the sense of presence and social plausibility. In experiences involving virtual humans, near-photorealistic representations may evoke discomfort associated with the uncanny valley, negatively affecting emotional engagement and credibility.

5.1.4. Authenticity and Representation

Representing living traditions and culturally sensitive practices raises questions of authorship, interpretation, and ownership. Without the meaningful involvement of source communities, XR applications risk simplifying, aestheticizing, or misrepresenting heritage narratives. Participatory and co-design approaches are therefore essential for ethical reasons and for ensuring cultural accuracy and long-term relevance.

5.1.5. Accessibility and Inclusivity Gaps

Although XR has the potential to broaden access to otherwise inaccessible sites and artifacts, many applications remain insufficiently inclusive. The absence of alternative modalities such as tactile interfaces, audio-first interaction, spatial sound, or NFC-based content access can exclude users with sensory or motor impairments. Integrating universal design principles from the outset remains an underdeveloped but critical area of practice. Inclusivity gaps are considered the most underdeveloped part of the studied works and this is partially due to the immaturity of the scientific community as a whole, which limits our ability to address these aspects within these emerging technologies. Universal design principles can be applied to address this issue not just by researching new interaction technologies but by adapting and selecting the most appropriate ones based on user characteristics and interaction contexts.

5.1.6. Evaluation Rigor

Evaluation rigor is strongly connected to the immaturity of technologies in their ability to understand and address the full range of usability barriers that are still existent in contemporary works. This is partially due to the shallow approach followed on evaluating these developments, which relies on small convenience samples and short-term, single-session evaluations. Such evaluations amplify novelty effects and are limiting external validity mainly because smaller, well-picked samples do not correspond to the wide variability of the real-life users of such systems. Museums and CH institutions are among the most challenging locations for deploying technologies due to their extremely wide user base. This inevitably results in the deployment and preservation of old-school approaches that provide a limited working set of functions (e.g., audio guides). The lack of standardized and domain-specific evaluation frameworks further complicates comparison across studies and weakens cumulative knowledge-building. The authors of this research work consider this to be the most important barrier that is currently inhibiting the further exploitation of XR technologies in such contexts.

5.1.7. Limitations of the Review Process

Beyond the limitations of the included evidence, the review process itself is subject to several constraints. First, the search was restricted to English-language publications, which may have excluded relevant work published in Chinese, Spanish, Arabic, or other languages; this is a particular concern given the global nature of CH research. Second, while conference proceedings from major venues (SIGGRAPH, CHI, IMX, VRCAI) were manually reviewed, the coverage of grey literature and non-indexed proceedings remains incomplete. Third, formal inter-rater reliability (e.g., Cohen’s κ) was not calculated during screening or data extraction, which limits the replicability of the selection process. Fourth, the review was not pre-registered in a prospective register, which increases the risk of post hoc analytical decisions. Finally, the risk-of-bias appraisal tool used was custom-developed for this review rather than an externally validated instrument, which may reduce its comparability with other systematic reviews in the field.

5.2. Implications for Future Research and Practice

We make the following suggestion based on our discussion of the findings of this research work: future research should temporarily put aside technical innovation and instead prioritize longitudinal and in situ evaluations of existing technologies; the goal should be to assess immediate usability and engagement, learning retention, meaning-making, and behavioral impact over time. It is important to develop standardized evaluation metrics which are tailored to CH, and which balance usability, experiential quality, and cultural interpretation; such tools would substantially strengthen evidence-based practices and, in the long run, could be beneficial for the entire domain of designing XR for CH applications. It is, of course, equally important to adopt ethically grounded co-design frameworks that position heritage communities as active contributors in transforming CH experience, rather than these communities being the passive subjects of the resulting technology.

For practitioners, including curators, heritage managers, and experience designers, the findings underscore the importance of clearly articulated, interpretive goals that subordinate technological spectacle to cultural meaning. Participatory design, scalable and cost-effective content pipelines, and built-in accessibility should be treated as foundational requirements rather than optional enhancements; they should be given top priority to sustain cultural values over technical excellence, ensuring that cultural values are well-served by this era of immersivity.

6. Conclusions

This systematic review synthesized evidence from 64 peer-reviewed studies on the application of XR technologies in CH by organizing findings across four interrelated dimensions: technology, interaction design, user experience, and evaluation. The analysis demonstrates that XR holds significant potential to support immersive presentation, deeper engagement, and enriched learning for both tangible and intangible heritage. Recent work increasingly emphasizes embodiment, social presence, and multimodal interaction, reflecting a maturation of the field beyond early experimental deployments.

At the same time, persistent challenges related to cost, usability, authenticity, accessibility, and limited longitudinal evaluation continue to constrain impact and scalability. The review reveals that technological advancement alone is insufficient; meaningful cultural outcomes depend on thoughtful interaction design, rigorous evaluation, and ethically informed, participatory practices. The Design Coherence Framework for XR Heritage (DCF-XR) advanced by this review provides practitioners and researchers with a practical diagnostic lens: interventions that optimize only one dimension in isolation (however technically sophisticated they are) are unlikely to yield the culturally meaningful outcomes that constitute the ultimate purpose of XR in heritage.

By providing a cross-XR, thematically structured synthesis of recent empirical work, this review highlights both methodological and ethical gaps while identifying promising directions for future research and practice. Addressing these gaps will require larger and more diverse user studies, standardized and culturally sensitive evaluation frameworks, and closer collaboration between technologists, curators, and heritage communities. In doing so, XR can evolve from a compelling visualization tool into a sustainable, inclusive, and culturally grounded medium for heritage interpretation and preservation. Achieving this requires not merely better technology or more rigorous evaluation in isolation, but the coherent alignment of all four dimensions identified by the DCF-XR: technology that serves interaction, interaction that serves experience, and evaluation that measures what genuinely matters for cultural heritage.