How Restorative Design in Aquatic Center Enhances User Learning Engagement: The Critical Role of Attention Restoration: An Environmental Psychology Approach with Implications for Sports Buildings

Abstract

With the increasing depth of research on built environments, theories of restorative environments and concepts of biophilic design have garnered widespread attention in the field of architecture. Based on Attention Restoration Theory (ART) and Flow Theory, this study systematically investigates how the architectural environment of aquatic centers influences users’ learning engagement in sports through psychological mechanisms. Analysis of cross-sectional data from 865 users revealed that all four core dimensions of restorative environments (being away, extent, fascination, and compatibility) significantly positively affect users’ learning engagement in aquatic centers. Psychological flow was found to mediate the relationship between these restorative dimensions and learning engagement. Building on previous research, this study constructs a theoretical framework of “restorative design–flow experience–learning behavior”, integrating the architectural features of aquatic centers with users’ psychological experiences. This approach addresses the gap in existing research where architectural elements and user psychological experiences have been studied in isolation, providing a theoretical basis for optimizing user experience through environmental interventions in sports architecture. The findings extend the application of environmental psychology in sports architecture and offer practical guidance for designing aquatic environments that promote learning engagement.

1. Introduction

Contemporary built-environment design is shifting from function to experience [1]. In sports architecture, cases from the London Aquatics Centre to the Sydney International Aquatic Centre show that environmental quality directly affects users’ athletic experience and training outcomes [2,3]. Aquatic centers serve elite training and public fitness. They combine natural elements with architectural systems. This semi-natural, semi-artificial character gives them distinct potential for psychological regulation compared with indoor venues that lack nature and with purely natural settings that lack professional facilities [4,5]. Water offers visual fluidity, specular reflections, and tactile cues. Long-span roofs bring abundant daylight. Water sounds can mask ambient noise. Together, these factors create a multisensory, immersive setting [6,7,8]. Yet most studies prioritize physical performance optimization [9], such as water treatment and energy use [10,11], while giving less attention to users’ psychological states.

The scientific development of recreational sports places higher demands on the physical environment [12]. Even professional athletes often experience attentional lapses and mental fatigue during prolonged, high-intensity training [13]. These issues are more pronounced among general exercisers who lack formal attention training. Urban populations also face heavy work stress and attention depletion, which undermines performance and exercise adherence [14]. Although coaching and behavioral interventions are common solutions, the role of the physical environment remains under-addressed [15]. For typical exercisers, environmental quality strongly shapes exercise experience and willingness to persist. Aquatic activities, due to their immersive qualities and coordination demands, require heightened concentration [16]. Therefore, identifying architectural strategies that support engagement among everyday users is of practical importance for improving the service quality of public sports facilities [17,18]. In concentration-intensive aquatic disciplines such as swimming and diving, whether the environment supports psychological recovery and focus has become a recent international research focus [19]. With increasing integration of sports and health care, optimizing psychological states through architectural design has emerged as a frontier topic at the intersection of sport science and architecture [20].

Grounded in Attention Restoration Theory (ART) and Flow Theory, this study uses ART to explain how aquatic centers mitigate mental fatigue and enhance attention through restorative environments. Flow Theory is used to examine how environmental conditions foster deep immersion and sustained engagement during exercise. Using a sample of 865 aquatic-center users in China, we conduct an empirical study and develop a model linking restorative design to exercise engagement. The study offers a new perspective for environmental psychology within sports architecture. It proposes a pathway to optimize performance through spatial design and provides evidence for planning and design in both elite sport and public fitness. From the user’s standpoint, we also validate the benefits of restorative design in large public buildings, offering insights that support sustainable architectural development.

2. Literature Review

2.1. Restorative Potential of Aquatic Centers

Aquatic centers, which combine recreation, fitness, and skill training, have expanded rapidly within urban waterfronts and the sports–leisure industry [21]. Compared with conventional indoor venues, they draw on natural or engineered water bodies and integrate landscape with built facilities to offer both challenge and leisure value [22]. Their core value goes beyond standard programmatic functions. By fusing water’s natural qualities with man-made systems, they form a distinctive semi-natural, semi-artificial environment. This makes aquatic centers an ideal setting for interdisciplinary inquiry connecting architectural environmental design with sport psychology.

Water’s properties align well with Attention Restoration Theory as a means to alleviate mental fatigue, while aquatic activities provide favorable conditions to study flow. From an architectural and environmental perspective, water elements readily evoke “soft fascination”, and spatial configurations can foster “being-away” and “extent” [23]. From sport and education perspectives, learning aquatic skills demands deep involvement and a balance between challenge and skill—key precursors to flow [7]. Thus, aquatic centers possess restorative potential at the environmental level and naturally align with flow conditions at the activity level, making them well-suited to study both restorative and flow experiences.

2.2. Attention Restoration Theory and Perceived Restorative Design

Attention Restoration Theory (ART) was formally proposed by Stephen and Rachel Kaplan in 1989 [24]. It explains why contact with natural environments can restore mental fatigue and improve attention and cognition. The theory starts from a key premise: modern urban life heavily consumes a resource called directed attention—a top-down, effortful form of attention that enables us to ignore distractions, suppress intrusive thoughts, and persist with dull or difficult tasks [25]. ART posits that certain features in natural settings attract attention effortlessly, allowing the neural systems responsible for directed attention to rest and recover; this process is termed attention restoration [26]. As such, ART is a central framework for understanding how environments help replenish attentional resources [27]. Prolonged demands on directed attention lead to fatigue. Environments with specific attributes can shift attention toward involuntary attention and thereby restore cognitive capacity. ART highlights four attributes: being-away, extent, fascination, and compatibility. Together, they support relaxation and the return of focused attention [28]. Although ART was developed with natural settings in mind, later studies show that these restorative dimensions are not limited to forests or lakes. They can be elicited in artificial or semi-artificial environments and have positive effects across education, work, health, and tourism contexts [29,30]. Empirical work supports these claims. Herzog et al. found that settings high in restorative components increase environmental preference and perceived restorative potential; being-away and compatibility were the strongest predictors [31]. In educational contexts, Bellini et al. reported that being-away, compatibility, and extent are positively associated with immersion in learning. Extent did not directly affect flow but influenced it indirectly via intrinsic motivation; fascination enhanced immersion indirectly via extrinsic motivation [32]. Research on informal learning spaces aligns with this view: interviews with university students indicate that quiet individual study areas, views of nature, orderly layouts, and diverse facilities promote engagement—even indoors—when the four ART attributes are present [33]. Measurement tools have also advanced. The Perceived Restorativeness for Activities Scale (PRAS) extends ART assessment to activity-based settings, including sport and skill learning. Validation in university sport courses shows that the four dimensions significantly affect attention restoration and learning experience [34]. In sum, evidence converges that ART enhances engagement and performance in learning contexts through attentional recovery, motivational gains, and affective regulation.

Perceived Restorative Design refers to users’ subjective appraisal of environmental features that support mental and physical recovery. Evidence shows that sports and leisure settings with natural elements, landscape diversity, well-organized layouts, and facilities aligned with activity needs can enhance psychological restoration and participation [9,35]. However, most empirical work has focused on nature, urban parks, and healthcare facilities. Research on restorative design perception in aquatic centers and its effects on learning-type activities remains limited [36,37]. Aquatic centers combine the potential restorative power of waterscapes with the high engagement of sport. They provide a distinctive context to test the applicability of Attention Restoration Theory in educational and training settings [38]. Therefore, this study examines the relationship between perceived restorative design and learning engagement in aquatic centers.

2.3. Flow Theory

Flow Theory was developed by Mihaly Csikszentmihalyi in the 1970s. Drawing on studies of artists, athletes, and musicians, he identified a distinctive psychological state in which individuals become fully absorbed in an activity, sustain intense concentration, and experience strong satisfaction and accomplishment [39,40]. The theory holds that when environmental features and task conditions jointly meet several core elements, people are more likely to enter flow. Flow is associated with stronger motivation, longer participation, and better learning and performance outcomes [41]. Key elements include a balance between challenge and skill, clear goals, immediate feedback, deep concentration on the task, a sense of control, altered time perception, and a temporary reduction in self-consciousness [42]. In aquatic centers, well-designed restorative environments can support several of these conditions, such as the balance between challenge and skill, focused attention, and changes in time perception, thereby facilitating flow [43]. Learning and training activities in aquatic centers often satisfy the prerequisites for flow. Participants must cope with water dynamics, technical execution, and physical demands. They also need rapid feedback and continuous adjustment of posture and balance. Supported by the aquatic setting and surrounding environmental cues, attention tends to narrow, time is easily forgotten, and deep focus emerges.

2.4. Learning Engagement

Learning engagement refers to the behavioral, emotional, and cognitive involvement that individuals display during learning. It concerns not only presence in class but also being mentally and emotionally invested. It is widely used to predict motivation, learning outcomes, achievement, and the intention to persist [44].

Following Fredricks and colleagues, learning engagement has three dimensions. Behavioral engagement captures attendance, attention, participation in activities, and rule compliance. Emotional engagement reflects feelings toward content, instructors, peers, and the school context, such as interest, belonging, and identification. Cognitive engagement concerns effort, use of deep strategies, and willingness to take on challenging tasks [45].

Learning engagement is a core construct for assessing the depth and quality of learners’ participation. This study focuses on engagement during aquatic skill acquisition. Drawing on Schaufeli and colleagues, we define learning engagement in aquatic centers as a positive, fulfilling state characterized by vigor, absorption, and dedication while users acquire aquatic skills [46].

3. Research Hypotheses

3.1. Being-Away, Extent, Fascination, and Compatibility and Learning Engagement in Aquatic Centers

Being-away is considered a core precondition of restorative environments. It emphasizes relief from daily routines and psychological pressure [47]. Early studies focused on the value of detachment in natural settings. More recent work has deepened its meaning and broadened contexts. Laumann and colleagues made a foundational contribution by operationalizing being-away into two subdimensions: physical escape and psychological escape. Their experiments showed that even without changing the physical setting, psychological detachment can promote attention restoration. This highlights the psychological essence of being-away and supports its application beyond wilderness contexts. A limitation is the laboratory setting, which may not capture the complexity of real environments [48]. Subsequent research tested being-away in specific contexts. In higher education, dormitories with natural views provided detachment that correlated with better graduate student mental health. This finding suggests cumulative benefits that extend beyond short-term restoration [49]. Work on library spaces shows that “being away from social pressure” is key to focused study. This qualitative finding underscores the role of social detachment, not only environmental detachment, in promoting engagement, especially in highly social campus settings [33].

Most prior studies emphasize static and passive restoration. Our study shifts focus to a dynamic, physically engaging context. In aquatic centers, detachment is achieved not only by looking but also by doing. Users dive, move water, and engage the whole body. This form of action-oriented participation can powerfully disengage routine thought patterns. Tourism research supports this idea. Feelings of escape can couple with activities and dynamically enhance satisfaction [50]. We therefore examine whether and how being-away, achieved through bodily practice in aquatic activities, functions as a key antecedent that directly affects learners’ engagement.

H1: 

Being-away is positively associated with learning engagement in aquatic centers.

Extent highlights environmental richness and coherence that allow users to feel immersed in a self-contained and explorable world [51]. It concerns not only physical scale but also structure and order. Its core lies in reducing unpredictable distractions to lower cognitive load and support sustained attentional recovery. Herzog and colleagues refined the construct by distinguishing coherence and scope. They showed that both jointly enable deep immersion [31]. A high-extent environment is harmonious and interconnected, which guides users to engage more deeply. A limitation is that much of this work relies on appraisals of natural scenes and gives limited direction for design in built settings. Recent studies translate extent into practice. Research on rural tourism destinations shows that comprehensiveness and coherence increase immersion and participation [52]. Natural landscapes, cultural elements, and infrastructure co-create the sense of extent at the destination scale. This system’s view informs analysis of complex venues such as aquatic centers. The most directly relevant evidence comes from controlled experiments on lighting in aquatic centers. Uniform, glare-free lighting improves visual comfort and performance among elite swimmers [53]. This offers a concrete and measurable interpretation of extent from an environmental engineering perspective. A visually coherent light field helps build a sense of wholeness, prevents distraction from uneven brightness and glare, reduces cognitive load, and supports focus on technique. This evidence addresses earlier abstraction and guides measurement in this study.

Accordingly, we operationalize extent as the coherence and wholeness of spatial layout, visual landscapes, and functional zoning in aquatic centers. A well-designed, seamless environment should reduce disturbance and offer clear paths for exploration, which supports sustained focus and immersion in learning.

H2: 

Extent is positively associated with learning engagement in aquatic centers.

Fascination is a core attribute of restorative environments. It refers to effortless attention capture and maintenance, which supports the recovery of directed attention [54]. In aquatic settings, its expression relates to but also differs from traditional nature contexts. Evidence spans natural and built environments. Classic work shows that “blue space” scenes containing water receive higher preference and restorativeness ratings than “green space” scenes with vegetation alone, underscoring the unique appeal of water [55]. Affective priming studies further indicate an interaction between fascination and emotion, where high-fascination nature scenes yield stronger restorative appraisals [56]. In aquatic centers, design and activity features that sustain gentle attention, such as surface reflections, moving splashes, and natural color palettes, can heighten interest and willingness to participate, which supports learning engagement. Research also points to the dynamic qualities of water as a key driver of fascination across natural and engineered settings [57]. Environments rich in fascination can therefore aid attentional recovery and optimize the exercise experience itself by lowering perceived effort and fostering positive affect. In aquatic activities, this pathway can translate into deeper focus and greater engagement.

H3: 

Fascination is positively associated with learning engagement in aquatic centers.

Compatibility refers to the fit between the environment and users’ goals, activity needs, and skill levels [58]. In aquatic centers, alignment among curricula, facility layout, teaching methods, and participants’ abilities directly shapes the learning experience [59]. For beginners, shallow-water zones, handrails, and a gradual teaching pace can reduce psychological pressure and increase engagement [60]. For lifeguard training, professional equipment and realistic simulations can enhance immersion and skill acquisition [61]. Evidence from urban landscape studies shows that compatibility is a strong predictor of environmental preference and perceived restorativeness, sometimes exceeding fascination and being-away [62]. In learning settings, compatibility is often identified as a core driver of engagement. When facilities and programming align with learners’ goals in aquatic centers, they can lower barriers to participation, strengthen motivation, and support sustained involvement.

H4: 

Compatibility is positively associated with learning engagement in aquatic centers.

3.2. Mediating Role of Psychological Flow

Psychological flow is an optimal experience marked by full involvement, focused attention, intrinsic enjoyment, fusion of action and awareness, altered time perception, and a temporary loss of self-consciousness. It also features a balance between challenge and skill [39]. A key frontier in flow research examines antecedents, especially how environmental design creates conditions for flow. We argue that restorative features in aquatic centers promote flow, which in turn drives deeper learning engagement.

First, restorative environments provide the cognitive and affective prerequisites for flow. Xie and colleagues propose a bridge between environmental restorativeness and flow. Restorative settings reduce stress and distraction and create a mental space that supports undivided attention on the task, a primary condition for flow [63]. A limitation of this model is its focus on broad natural contexts without detailing how specific design attributes differentially influence flow. Second, different restorative dimensions promote flow through distinct pathways. Being-away reduces internal interference by enabling psychological relief from routine demands, which clears the way for immersion. Extent creates coherent and structured settings, such as clear visual wayfinding and well-defined functional zones, which lowers cognitive load for environmental navigation [64]. This effortless understanding frees cognitive resources for skill acquisition and supports sustained flow. Fascination captures involuntary attention through the dynamic appeal of waterscapes, such as reflections and motion. This gentle attentional pull reduces the effort needed to maintain focus and eases entry into flow. Compatibility aligns environmental challenges with users’ skill levels and goals. Strong support for activity goals enhances control and self-efficacy, which are essential for achieving the challenge–skill balance and for intrinsic enjoyment. Third, flow functions as a mediator that activates deep learning engagement. When users enter flow, they experience intrinsic pleasure, altered time perception, and intense concentration. These experiences reinforce learning behaviors. We therefore posit that restorative environments in aquatic centers indirectly foster high engagement by creating psychological and physical conditions that enable entry into and maintenance of flow. Based on this logic, we propose (The proposed model appears in Figure 1):

H5: 

Flow mediates the effect of being-away on learning engagement.

H6: 

Flow mediates the effect of extent on learning engagement.

H7: 

Flow mediates the effect of fascination on learning engagement.

H8: 

Flow mediates the effect of compatibility on learning engagement.

4. Research Design

4.1. Participants

This study surveyed regular users of aquatic centers in China, specifically individuals who engage in recreational or instructional activities and are neither professional nor elite athletes. To ensure alignment with the research scope, we excluded professional athletes (e.g., national or provincial team members and registered athletes), coaches, and venue staff. The sample included public users who actively used the facilities and participated in aquatic activities. It covered variation in gender, age, and frequency of use to enhance diversity and external validity.

Data were collected in eight cities with geographic and economic diversity: Beijing, Shanghai, Guangzhou, Shenzhen, Hangzhou, Chengdu, Wuhan, and Qingdao. These sites represent eastern coastal, southern coastal, central, and western regions. The study was approved by the Ludong University Ethics Committee (IRBLDU20250429). All respondents were informed about the study purpose, anonymity, and voluntary participation. Written informed consent was obtained before participation. Demographic information is reported in

Table 1. Descriptive analysis of basic information.

Variables	Option	Frequency	Proportion (%)
Gender	Male	445	51.45
	Female	420	48.55
Age	18–35	535	61.85
	36–55	246	28.44
	56 and above	84	9.71
Household registration	City	330	38.15
	Suburbs	232	26.82
	Rural	303	35.03
Educational background	Bachelor’s degree or above	525	60.69
	Below Bachelor’s Degree	340	39.31
Participants in this region	Eastern Coastal	210	24.27
	Southern Coastal	257	29.71
	Central	209	24.16
	Western	189	21.84
Number of times per month	1–2 times	244	28.21
	3–4 times	356	41.16
	5 times and above	265	30.64

.

4.2. Questionnaire Distribution

This study employed a convenience sampling method, with data collection conducted from 1 May to 2 July 2025. Surveys were administered offline across eight cities in China, including Beijing, Shanghai, and Guangzhou. The questionnaire distribution followed an on-site, random interception approach with immediate retrieval upon completion. After obtaining permission from facility management at each site, the research team set up survey stations at exits and lounges during class dismissals and peak-use periods. Trained surveyors distributed paper questionnaires to users who met the inclusion criteria. The first page included screening items to exclude professional athletes and non-target groups. After a brief oral and written introduction, respondents completed the questionnaire independently without disturbance. On-site surveyors conducted an initial check for completeness and logic and prompted respondents to fill in missing key items. All questionnaires were collected on site, then rechecked. Forms with obvious inconsistencies or severe missing data (more than 30 percent) were removed to ensure data quality. In total, 899 questionnaires were distributed and 865 were returned. The response rate was 96.2 percent, providing a sound basis for subsequent analyses.

4.3. Variable Measurement

All variables in this study were measured using well-established scales developed by prominent scholars, adopting a five-point Likert scale ranging from 1 (“strongly disagree”) to 5 (“strongly agree”). Except for the Chinese version of the Perceived Restorativeness Scale, the remaining instruments were translated and back-translated to ensure conceptual equivalence and ease of understanding for respondents. The specific measurements are as follows:

For Perception of Restorative Design, this study adopted the Chinese version of the Perceived Restorativeness Scale (PRS) revised by Li et al. to measure the restorative design effects of aquatic sports centers [65]. The scale is based on Attention Restoration Theory (ART) and follows the structure of the original PRS-11 [66], consisting of 11 items under four dimensions: Being-Away (3 items), Extent (3 items), Fascination (3 items), and Compatibility (2 items). The scale has been validated with Chinese samples, demonstrating strong reliability and validity, and is suitable for direct application in assessing users’ perceptions of restoration. While the original version used reverse scoring, the present study adjusted it to positive scoring for ease of understanding and consistency, with internal consistency verified during testing.

For Flow, this study employed the Psychological Flow Scale (PFS) developed by Norsworthy et al. [67], which comprises three dimensions: absorption, effortless control, and intrinsic reward, each measured by three items.

For Learning Engagement, this study drew upon the Utrecht Work Engagement Scale–Student (UWES-S) originally developed by Schaufeli et al. [68], later validated cross-culturally and adapted into Chinese by Fang et al., as well as the engagement scale developed by Chen et al. and the Sport Engagement Scale by Guillén and Martínez-Alvarado [69,70]. We adapted the learning engagement scale by incorporating aquatic skill-learning elements. For example, the original item “I feel happy when I am fully engaged in learning” was revised to “I feel happy when I am fully engaged in aquatic skill learning”. Other items were modified only to reflect the study context. The final instrument measured three dimensions: Vitality, Dedication, and Concentration. Spector notes that five- to seven-point options are most common and effective, and that maintaining a consistent response scale within a study is important to avoid confusion [71]. The original instruments for our variables used different scales. Cross-cultural research also suggests that respondents in East Asian contexts tend to avoid extreme options, and that a five-point scale can mitigate this tendency and yield better distributions and psychometric properties [72]. Therefore, all variables in this study used a five-point Likert scale. Three experts with doctoral degrees and full professorships reviewed the adapted items. To ensure clarity and effectiveness, we conducted a pilot test with 15 participants after expert review. The pilot showed acceptable reliability and validity. These pilot cases were excluded from the final sample.

4.4. Research Procedures

This study employs a quantitative research approach, utilizing covariance-based structural equation modeling (CB-SEM) to ensure the scientific rigor, validity, and reliability of the research findings. Data were initially collected through standardized questionnaires and subsequently cleaned and preprocessed using SPSS 26.0, with invalid responses excluded and missing data appropriately handled. The analytical process followed these sequential steps: First, common method bias was examined to eliminate potential data bias. Subsequently, reliability and validity analyses were conducted to verify the measurement instruments’ dependability and effectiveness. Building upon this foundation, correlation analysis was performed to explore basic relationships among variables. Finally, structural equation modeling was constructed to test theoretical hypotheses, complemented by bootstrap testing for mediating effects. All analyses underwent rigorous statistical testing with a significance level of p < 0.05, ensuring the scientific credibility of the research conclusions.

5. Result

5.1. Common Method Bias Test

In this study, Harman one-way test was used for common method bias test. As the results in

Table 2. Total Variance Explained.

Total Variance Explained
Component	Initial Eigenvalues			Extraction Sums of Squared Loadings			Rotation Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	8.806	29.352	29.352	8.806	29.352	29.352	2.417	8.058	8.058
2	2.663	8.876	38.228	2.663	8.876	38.228	2.397	7.991	16.048
3	1.862	6.206	44.434	1.862	6.206	44.434	2.396	7.986	24.034
4	1.632	5.440	49.874	1.632	5.440	49.874	2.385	7.951	31.985
5	1.589	5.297	55.171	1.589	5.297	55.171	2.377	7.925	39.910
6	1.426	4.755	59.926	1.426	4.755	59.926	2.288	7.627	47.537
7	1.360	4.532	64.458	1.360	4.532	64.458	2.191	7.302	54.839
8	1.269	4.229	68.687	1.269	4.229	68.687	2.161	7.203	62.043
9	1.205	4.016	72.702	1.205	4.016	72.702	2.160	7.199	69.242
10	1.048	3.493	76.196	1.048	3.493	76.196	2.086	6.954	76.196

Extraction Method: Principal Component Analysis.

show, there are 10 factors with eigenvalues greater than 1, the total explained variance is 76.196%, and the variance explained by the first principal factor is 8.058%, which is less than the critical criterion of 40%, so there is no serious common method bias problem in this study.

5.2. Correlation Analysis

Table 3. Correlation analysis table.

Variable	Mean	Standard Deviation	1	2	3	4	5	6
PA	3.686	0.757	1
PE	3.310	0.805	0.212 ***	1
PF	3.607	0.782	0.226 ***	0.154 ***	1
PC	3.523	0.812	0.278 ***	0.254 ***	0.287 ***	1
PSF	3.581	0.687	0.335 ***	0.361 ***	0.342 ***	0.364 ***	1
LE	3.475	0.713	0.376 ***	0.390 ***	0.325 ***	0.394 ***	0.440 ***	1

Note: *** p < 0.001.

demonstrates the mean, standard deviation and two-by-two Pearson correlation coefficients for each variable. The results of the analysis show that there is a significant positive correlation between all variables (p < 0.001). Among them, the correlation coefficients between the dependent variable LE and the respective variables ranged from 0.325 to 0.440, indicating that all these predictor variables were positively associated with the outcome variables with moderate strength. The correlation coefficients of the mediator variable PSF with the respective predictor variables ranged from 0.335 to 0.364, whereas the correlation between PSF and LE reached 0.440, the highest value of all the two-by-two relationships, a result which tentatively supports the soundness of the subsequent mediation effects analysis. The correlation coefficients between the predictor variables remained within the range of 0.154 to 0.287, showing that the independent variables were correlated but did not suffer from serious multicollinearity problems. All correlation coefficients reached a significance level of p < 0.001, indicating that these associations are highly statistically significant. For visualization, the correlation results of this study are visualized in this paper, see Figure 2.

5.3. Reliability Test

Table 4. Reliability analysis table.

Item	PA	PE	PF	PC	PSF_A	PSF_B	PSF_C	LE_A	LE_B	LE_C	Total
Reliability Value	0.758	0.791	0.784	0.788	0.882	0.848	0.876	0.875	0.878	0.883	0.915

shows that the Cronbach’s α coefficients of all variables are higher than the acceptable standard of 0.7 (PA = 0.758, PE = 0.791, PF = 0.784, PC = 0.788), and the reliability coefficients of the PSF and LE dimensions are between 0.848 and 0.883, and the reliability of the total scale reaches 0.915, which indicates that the measurement tool has good internal consistency.

The reliability value of PA is 0.758, PE is 0.791, PF is 0.784, PC is 0.788, PSF_A is 0.882, PSF_B is 0.848, PSF_C is 0.876, LE_A is 0.875, LE_B is 0.878, and LE_C has a reliability value of 0.883, and the reliability value of the total scale is 0.915. In summary, the reliability values of the questionnaire’s dimensions are all above 0.7, indicating that the questionnaire has a good reliability value, and it can be continued to carry out the subsequent factor analysis.

5.4. Validity Test

Exploratory factor analysis was used in this study to test the structural validity of the questionnaire. As shown in

Table 5. KMO and Bartlett’s test.

Test	Value
KMO value	0.887
Approximate chi-square	13,478.607
Degrees of Freedom	435
Significance	0.000

, the KMO value was 0.887 (>0.8) and Bartlett’s test of sphericity was significant (χ² = 13,478.607, p < 0.001), indicating that the data were suitable for factor analysis. Factors with eigenvalues greater than 1 were extracted using principal component analysis and rotated by the variance maximization method to obtain a clear factor structure. As shown in

Table 6. Table of Rotated Component Matrix.

Rotated Component Matrix a
	Component
	1	2	3	4	5	6	7	8	9	10
PA1	0.055	0.134	0.059	0.037	0.081	0.130	0.084	0.065	0.095	0.785
PA2	0.065	0.086	0.090	0.022	0.140	0.021	0.045	0.072	0.058	0.804
PA3	0.109	0.076	0.062	0.145	0.053	0.078	0.047	0.121	0.051	0.771
PE1	0.086	0.076	0.078	0.095	0.119	0.115	0.807	0.055	0.089	0.056
PE2	0.083	0.118	0.070	0.081	0.073	0.074	0.800	0.104	−0.010	0.064
PE3	0.077	0.120	0.078	0.112	0.048	0.107	0.807	0.070	0.026	0.057
PF1	0.057	0.019	0.111	0.056	0.050	0.101	0.018	0.094	0.795	0.045
PF2	0.077	0.068	0.061	0.086	0.113	0.045	0.065	0.075	0.821	0.081
PF3	0.126	0.115	0.078	0.093	0.057	0.057	0.016	0.112	0.798	0.077
PC1	0.106	0.158	0.075	0.071	0.087	0.064	0.091	0.779	0.142	0.119
PC2	0.062	0.080	0.098	0.111	0.111	0.093	0.053	0.787	0.070	0.068
PC3	0.084	0.053	0.055	0.101	0.060	0.109	0.093	0.822	0.087	0.083
PSF1	0.169	0.081	0.840	0.041	0.113	0.188	0.096	0.104	0.075	0.058
PSF2	0.202	0.064	0.815	0.048	0.102	0.179	0.089	0.107	0.129	0.112
PSF3	0.146	0.115	0.829	0.110	0.106	0.208	0.080	0.051	0.102	0.085
PSF4	0.147	0.108	0.198	0.066	0.124	0.793	0.113	0.126	0.069	0.093
PSF5	0.167	0.036	0.197	0.096	0.068	0.793	0.111	0.094	0.096	0.127
PSF6	0.194	0.083	0.173	0.093	0.039	0.823	0.121	0.085	0.077	0.048
PSF7	0.826	0.086	0.173	0.087	0.098	0.186	0.090	0.111	0.129	0.096
PSF8	0.849	0.061	0.155	0.053	0.058	0.178	0.087	0.095	0.088	0.056
PSF9	0.832	0.062	0.169	0.084	0.084	0.135	0.104	0.074	0.086	0.114
LE1	0.065	0.183	0.074	0.794	0.242	0.092	0.108	0.102	0.137	0.095
LE2	0.096	0.166	0.069	0.836	0.203	0.085	0.081	0.117	0.081	0.063
LE3	0.069	0.175	0.056	0.822	0.162	0.085	0.155	0.110	0.071	0.080
LE4	0.074	0.207	0.099	0.158	0.811	0.080	0.130	0.119	0.094	0.097
LE5	0.098	0.173	0.111	0.217	0.826	0.052	0.069	0.114	0.114	0.124
LE6	0.081	0.155	0.124	0.250	0.800	0.108	0.091	0.071	0.064	0.120
LE7	0.048	0.844	0.078	0.172	0.121	0.111	0.128	0.095	0.084	0.101
LE8	0.093	0.809	0.101	0.177	0.234	0.069	0.096	0.137	0.085	0.130
LE9	0.080	0.816	0.087	0.180	0.183	0.051	0.156	0.102	0.071	0.132

Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. ^a Rotation converged in 7 iterations.

, the loadings of each measurement question item on the corresponding factor were all greater than 0.7, and there was no significant cross-loading phenomenon, confirming that the questionnaire had good structural validity.

5.5. Validated Factor Analysis

Based on the multiple factors explored in the aforementioned Exploratory Factor Analysis (EFA), a Validated Factor Analysis was conducted in this study to further confirm whether these factors conformed to the validity structure. Confirmatory Factor Analysis (CFA), as a measurement model in Structural Equation Model (SEM), is mainly used to measure the relationship between an observed sample size or indicator and the underlying variables or factors, and usually includes both discriminant and convergent validity analyses (See

Table 7. Distinguished Validity Analysis.

Dimension	PA	PE	PF	PC	PSF_A	PSF_B	PSF_C	LE_A	LE_B	LE_C
PA	0.716
PE	0.273	0.751
PF	0.294	0.197	0.741
PC	0.361	0.322	0.367	0.745
PSF_A	0.323	0.325	0.338	0.335	0.845
PSF_B	0.342	0.394	0.308	0.374	0.578	0.807
PSF_C	0.327	0.332	0.343	0.351	0.514	0.531	0.845
LE_A	0.321	0.381	0.334	0.387	0.300	0.344	0.311	0.838
LE_B	0.393	0.348	0.332	0.374	0.386	0.336	0.330	0.605	0.842
LE_C	0.406	0.404	0.308	0.396	0.341	0.329	0.305	0.544	0.558	0.847

Distinguished Validity Analysis Table and

Table 8. Convergent validity analysis table.

Item	PA	PE	PF	PF	PSF_A	PSF_B	PSF_C	LE_A	LE_B	LE_C
CR	0.759	0.794	0.785	0.789	0.882	0.849	0.881	0.876	0.879	0.884
AVE	0.512	0.564	0.549	0.555	0.714	0.652	0.715	0.703	0.709	0.718

Aggregate Validity Analysis Table for details).

Distinguishing validity is used to measure the accuracy and validity of a measurement tool in evaluating a particular concept, and is crucial in the field of research and evaluation to ensure the reliability and accuracy of data and conclusions. In Table 6, Distinguishing Validity Analysis, the diagonal values are the square root of AVE of each dimension, which are 0.716, 0.751, 0.741, 0.745, 0.845, 0.807, 0.845, 0.838, 0.842, 0.847, respectively, which are greater than the other values in the same columns, which indicates that the questionnaire has good discriminant validity and provides a reliable basis for the subsequent hypothesized research basis.

Convergent validity is used to assess the overall validity of multiple questions (or variables) in a questionnaire, which can help the researcher to judge the degree of internal consistency and interrelatedness of related questions. The commonly used indicators are CR (Constructive Reliability) and AVE (Average Variance Extraction), in which the CR value usually needs to be greater than 0.6, and when the AVE value is greater than or equal to 0.5, it indicates that the measurement tool has good internal consistency and high validity. Table 7 shows that the CR and AVE values of each dimension of the questionnaire are above 0.5, indicating that the questionnaire has good convergent validity, which not only ensures the reliability and validity of the measurement tool, but also enhances the comparability of the data and the ability to support decision-making, which helps to produce accurate, consistent and credible research results.

5.6. Structural Equation Modeling Analysis

We conducted structural equation modeling in Amos 26.0. Model fit was evaluated using commonly reported indices, including GFI, AGFI, RMR, RMSEA, NFI, IFI, CFI, and the Tucker–Lewis Index TLI. The structural model is shown in Figure 3, with overall fit statistics in

Table 9. Results of model fit analysis.

Fit Index	Range of Fit	Results	Rating Results
GFI	>0.9	0.947	Accept
AGFI	>0.9	0.931	Accept
RMR	<0.05	0.021	Accept
RMSEA	<0.05	0.031	Accept
NFI	>0.9	0.952	Accept
IFI	>0.9	0.977	Accept
CFI	>0.9	0.977	Accept
TLI	>0.9	0.972	Accept

, path estimates in

Table 10. Table of path coefficients.

Path	Estimate	S.E.	C.R.	P
PSF←PA	0.218	0.042	4.726	***
PSF←PE	0.301	0.038	6.78	***
PSF←PF	0.243	0.036	5.426	***
PSF←PC	0.214	0.036	4.54	***
LE←PSF	0.227	0.058	3.849	***
LE←PA	0.214	0.041	4.69	***
LE←PE	0.245	0.038	5.423	***
LE←PF	0.147	0.035	3.359	***
LE←PC	0.191	0.035	4.156	***

Note: *** p < 0.001.

, and mediation results in

Table 11. Table of mediation test results.

Path	Mediation Effect	Bootstrap 95% CI	SE	Direct Effect	Total Effect	Mediation Proportion	Significant?
PA→PSF→LE	0.049	[0.023, 0.076]	0.012	0.214	0.263	0.188	Fit
PE→PSF→LE	0.068	[0.039, 0.098]	0.014	0.245	0.313	0.218	Fit
PF→PSF→LE	0.055	[0.030, 0.080]	0.012	0.147	0.202	0.273	Fit
PC→PSF→LE	0.049	[0.022, 0.075]	0.012	0.191	0.240	0.203	Fit

. As reported in Table 8, all fit indices meet conventional criteria for good fit GFI = 0.947, AGFI = 0.931, RMR = 0.021, RMSEA = 0.031, NFI = 0.952, IFI = 0.977, CFI = 0.977, TLI = 0.972. These values indicate that the specified model provides an adequate representation of the data. Table 10 summarizes the structural paths. PA, PE, PF, and PC each show significant positive effects on PSF p < 0.001, with the largest coefficient for PE 0.301, followed by PF 0.243, PA 0.218, and PC 0.214. PSF has a significant positive effect on LE 0.227, p < 0.001. In addition, PA, PE, PF, and PC each retain significant direct effects on LE p < 0.001, with PE 0.245 being the largest, followed by PA 0.214, PC 0.191, and PF 0.147. This pattern indicates that the four restorative dimensions are associated with higher levels of PSF and LE, and that PSF transmits part of their impact on LE.

5.7. Mediation Test

We used a bootstrap approach with 5000 resamples to estimate indirect effects and their confidence intervals. Mediation is considered present when the 95% bootstrap confidence interval does not include zero. As shown in Table 10, all four indirect paths from PA, PE, PF, and PC to LE through PSF are significant, indicating partial mediation in every case. See Table 11,the estimated indirect effects are 0.049 for PA, 0.068 for PE, 0.055 for PF, and 0.049 for PC, with corresponding 95% intervals that exclude zero. Direct effects remain positive and statistically meaningful, so PSF transmits part, but not all, of the influence of restorative design on engagement.

Beyond significance, the effect sizes are informative in two ways. First, in absolute terms, PE shows the largest indirect effect through PSF, followed by PF, while PA and PC have comparable indirect magnitudes. Second, the proportion mediated indicates how much of each total effect operates through PSF. The shares are 18.8 percent for PA, 21.8 percent for PE, 27.3 percent for PF, and 20.3 percent for PC. Thus, although PE has the largest indirect effect in absolute size, PF shows the strongest reliance on PSF in relative terms because over a quarter of its total effect on LE is carried by PSF. Taken together, the pattern supports a partial-mediation account.

6. Discussion

6.1. Direct Effect

The results show that being away, extent, fascination, and compatibility each have a distinct and significant pathway to learning engagement in aquatic centers. Rather than acting as undifferentiated “pleasantness”, these dimensions operate through four mechanisms: psychological detachment (being away), system-level immersion via spatial coherence and continuity (extent), effortless attentional capture (fascination), and task–environment fit that calibrates the challenge–skill balance (compatibility). Together, these mechanisms elevate vitality, focus, and persistence, indicating that restorative design can be a direct lever for improving engagement in performance-oriented venues. Importantly, demonstrating these effects within sports architecture extends Attention Restoration Theory (ART) beyond leisure and classroom contexts and clarifies which architectural attributes translate into measurable learning behaviors in aquatic facilities.

6.2. Mediation Effect

Psychological flow mediates the link between restorative design and learning engagement by turning environmental perception into sustained participation. In aquatic centers, it is the overall configuration of space and program that lightens residual cognitive load, steadies attention, and keeps tasks aligned with users’ abilities. In such settings, learners are more likely to reach a state where action and awareness come together, time feels less pressing, and distractions recede; this is the experience that carries the influence of restorative design into observable engagement. Our results indicate that when the environment jointly supports detachment from daily demands, coherent immersion, effortless attentional pull, and a good task–ability match, the likelihood and depth of flow increase, and engagement follows. The evidence therefore points to a simple sequence: restorative qualities create the conditions, flow provides the mechanism, and together they explain focused and persistent participation in aquatic centers.

6.3. Research Comparison

Prior work converges on a coherent picture in which restorative qualities support attention and engagement, and flow sustains persistence. Field evidence reported by Pasanen and colleagues shows that exposure to restorative settings improves mood and sustained attention, consistent with the core claims of Attention Restoration Theory [73]. Jaggard’s findings point to the importance of effortless fascination when benefits are not constrained by deliberate control of attention [74]. Xie and collaborators connect restorative context with flow and well-being, indicating a plausible route from environmental qualities to motivated action [63]. Building on this trajectory, studies of learning in smart environments by Gao and colleagues emphasize immersion and participation, while recent sport-focused theorizing by Farrokh and coauthors foregrounds how context conditions shape the emergence of flow [75,76].

The conclusions of this study are consistent with prior research in several respects. A large body of literature has confirmed that restorative environments can promote learning and engagement through attention restoration and emotional regulation. For example, Pasanen et al. demonstrated through field experiments that walking in natural environments significantly improves mood and sustained attention, thereby supporting the generalizability of Attention Restoration Theory. Similarly, Jaggard found that when individuals are required to deliberately engage their attention, the benefits of restorative environments diminish, highlighting the role of effortless fascination in restoration processes. In the study of flow, Xie et al. emphasized that restorative environments reduce psychological stress and thereby enhance flow experiences, which in turn improve well-being and a sense of identity. These findings align with the present study, all indicating that environmental design exerts a positive influence on learning and concentration.

6.4. Theoretical Contributions and Practical Implications

This study advances Attention Restoration Theory and Flow Theory in sports architecture by showing that restorative design in aquatic centers shapes learning engagement both directly and through flow. Our integrated model makes the sequence clear. The restorative attributes identified by Attention Restoration Theory make flow more likely rather than serving as an endpoint, and flow in turn is the immediate psychological driver of sustained engagement. By demonstrating this progression from restorative perception to flow to engagement within aquatic facilities, we extend the reach of Attention Restoration Theory beyond general restorative settings and give Flow Theory concrete, architecture-based antecedents grounded in how venues are actually used. In doing so, we refine how environmental attributes translate into behavior in sports architecture and offer cumulative evidence that brings the two theories together within a single explanatory framework.

For the architectural design community and practicing architects, the findings point to clear moves that turn restorative potential into sustained engagement through flow in aquatic centers. Start the arrival with a measured sequence that slows users before the deck, add a short decompression vestibule with filtered views to the pool, calm the acoustic character at entries, and keep instructional zones at a remove from noisy recreation so learners can set aside everyday demands. To express extent, maintain long, continuous sightlines along the pool length, simplify the deck edge to reduce visual noise, align structural bays and lighting to read as a steady rhythm, and use transparent partitions or carefully controlled mirrored elements to deepen perceived space while keeping glare under control. To elicit fascination, bring in stable daylight with responsive shading, frame composed views of water movement rather than busy backdrops, layer gentle water and air sounds instead of sharp echoes, and select natural textures and a restrained palette that hold attention without effort. To achieve compatibility, tune lane widths and depths and equipment to ability levels, schedule cohorts so program difficulty matches users, provide clear goals and feedback through pace clocks and progress boards at eye level, and place instructors and circulation so guidance is easy to follow. Applied together, these decisions allow architects to deliver aquatic centers that sustain engagement and support learning at a high standard.

7. Conclusions

This study examined how restorative design in aquatic sports centers relates to learning engagement and clarified the role of psychological flow in that relationship. The results show that the four restorative dimensions—being away, extent, fascination, and compatibility—are each positively associated with engagement in this venue type, and that flow transmits their influence from environmental perception to sustained participation. These findings bring Attention Restoration Theory and Flow Theory together in a single tested sequence within sports architecture, extending both frameworks to an applied context where learning and training routinely occur. Beyond establishing the links, the study shows that engagement is highest when restorative qualities operate as a coherent whole that enables detachment, maintains immersion, invites effortless attention, and keeps tasks well matched to users. This integrated account offers a practical direction for architects and operators: strengthen these qualities through design and programming so that aquatic centers more reliably elicit flow and, in turn, support focused and durable learning engagement.

8. Limitations and Future Directions

Despite its theoretical and practical contributions, this study has several limitations that warrant attention. First, the cross-sectional design limits our ability to establish causal inferences or capture dynamic psychological processes. Future research should employ longitudinal designs or experimental interventions to examine how restorative features influence engagement and flow over time. Second, the exclusive reliance on self-reported measures may introduce common method bias. Although statistical checks such as the Harman single-factor test indicated this was not a major concern, future studies would benefit from incorporating objective metrics such as behavioral observations or physiological indicators of concentration. Third, we did not explicitly measure or control individual-level sociodemographic covariates, including socioeconomic status and prior aquatic experience, which may confound the observed relationships. Future work should collect these covariates, adopt stratified sampling, and model them using multivariate or multilevel approaches to obtain less biased estimates. Fourth, the sample was drawn exclusively from water sport centers in China. Cultural factors, including collective mindset and attitudes toward nature, may affect generalizability. Cross-cultural replications in diverse regional contexts are needed to verify the universality of the proposed model. Finally, while this study focused on four core restorative dimensions, other influential environmental factors such as acoustic quality and lighting conditions were not considered, and subsequent outcomes such as skill acquisition performance were beyond the scope of this paper. Extending the model to include these variables would enhance its practical relevance.