1. Introduction

As the global population continues to grow and human activities increasingly encroach upon natural habitats, the effective management of natural resources has emerged as a critical priority. Protected areas, including national parks, forest parks, and other nature reserves, provide essential environmental services that significantly enhance human well-being. In the context of rising health challenges and escalating healthcare costs, public health authorities are increasingly turning to preventive strategies, recognizing the potential of natural environments as valuable resources for promoting health and preventing disease [1–3]. The therapeutic benefits of exposure to forest environments have garnered increasing attention in recent years [4]. The natural environment is now widely acknowledged to play a crucial role in both physical and mental health, with direct interaction with nature being essential for improving the overall quality of life [5]. The University of Washington’s “Green Cities: Health” initiative underscores the positive impact of nature on human health, synthesizing evidence from nearly 4,000 peer-reviewed studies that highlight the connection between nature and health. Similarly, Japan Medical University has established the Forest Medicine Research Society, which regularly publishes research on the link between forests and human health, aiming to raise public awareness of the substantial health benefits associated with forest environments.

Akgüller et al. (2025) [6] further confirmed that natural disasters, as extreme natural events, can significantly reshape the dynamics of economic sectors—their study on the Kahramanmaraş Earthquake showed that different economic sectors exhibit divergent resilience: some face increased volatility while others gain stability. This implies that natural environments not only provide direct health value but also interact with human economic systems through various pathways, underscoring the need to explore the multi-dimensional roles of natural spaces like forest parks. Batrancea et al. (2025) [7] revealed that “robustness to external shocks” can be systematically achieved through multi-dimensional optimization—by integrating uncertainty quantification and adaptive preferences, the framework mitigates crisis impacts while maintaining performance. This insight is highly relevant to forest park research: just as financial systems need robustness against market shocks, natural spaces like forest parks also require considering multi-dimensional demands (health/leisure/economic) to enhance resilience amid environmental or social changes. Batrancea et al. (2023) [8] confirms that wellbeing-related infrastructure—including natural environmental facilities like air quality and green spaces—is inherently interdependent with economic growth, as it enhances citizens’ wellbeing while driving regional economic vitality and mitigating disparities.

In an era where health needs are increasingly prominent, restoring physical and mental well-being has emerged as a primary motivation for travel [9]. Forest parks, designated as areas for travel, sightseeing, relaxation, and engagement in scientific and cultural activities, provide a unique setting within the natural forest environment. They offer aesthetic appeal, educational opportunities, and recreational benefits, all achieved through scientific conservation and thoughtful development. Forest parks are not only significant for their ecological and economic value but also for the rich psychological and physiological benefits they provide, extending beyond their recreational and educational functions. As crucial ecological resources and public goods, the value of forest parks is largely realized through ecotourism. To attract visitors, the development and design of forest parks must address current visitor needs while also adapting to evolving demands. This dual approach will maximize the parks’ functional value, enhance visitor experiences, and improve their competitiveness, thereby ensuring sustainable development.As interest in the restorative potential of forest parks continues to grow, scholarly attention has increasingly converged on this area of research. In 1983, Kaplan and colleagues first introduced the concept of a “Restorative Environment,” defining it as a setting that aids individuals in recovering from mental fatigue and stress-induced negative emotions. Kaplan further elaborated on this concept, describing restorative environments as spaces that facilitate the recovery from mental exhaustion and stress-related adverse emotional states. Environmental restoration refers to the degree to which an environment supports this recovery process, with higher levels of support being more conducive to initiating a restorative response [10]. Restorative environmental Perception (REP) emphasizes the subjective evaluation of an environment’s capacity to enable effective human recovery. The restorative effects of forest environments have been extensively documented, with visitors’ perceptions of these environments directly influencing their tourism satisfaction and, subsequently, their loyalty as visitors. In 1983, Ulrich introduced the Stress Reduction Theory, highlighting the benefits of interacting with natural environments. According to Ulrich, exposure to nature can rapidly enhance mood, with positive emotions fostering motivation, physiological balance, behavioral impulses, and adaptive responses, ultimately reducing both psychological and physiological stress [11]. Ulrich’s research suggests that the process of stress recovery involves not only healing from excessive psychological and physiological stress but also from under-stimulation or low levels of stress. It also includes replenishing the energy expended during the psychological or physiological responses to stress. Additionally, American psychologist Erich Fromm introduced the term biophilia in 1980, referring to the innate human tendency to be drawn to living things [12].

The notion of environmental preference, a cornerstone of environmental psychology, has garnered substantial research attention over the past five decades, resulting in a robust and well-structured theoretical foundation. Synthesizing both domestic and international definitions, environmental preference can be conceptualized as encompassing two primary dimensions: environmental cognition and environmental evaluation. It reflects the attitudes individuals hold toward specific environments, which subsequently shape their behavioral choices. The development of this preference begins with the initial observation of the environment, progresses through cognitive assessment and evaluation, and culminates in the formation of a preference [13]. Here, “preference” denotes the emotional response individuals exhibit toward various environmental stimuli, representing the ultimate evaluation of their affinity for a particular environment based on the dynamic interplay between the individual and their surroundings [14]. These preferences are forged through direct psychological responses, such as emotions, past experiences, and evolutionary principles, which can either attract or repel individuals from certain environments [15,16]. Environmental preference is inherently dynamic, influenced by a range of factors including past experiences, environmental familiarity, life stage, gender, education, and cultural background, all of which contribute to its variability over time [17].

Staats & Hartig (2004) [18] delve into the connection between environmental preferences and psychological needs, positing that restorative needs act as a catalyst for preference evaluations. They propose that individuals’ anticipatory judgments of an environment’s restorative capacities significantly shape their environmental preferences. For instance, natural environments are frequently favored due to their perceived ability to restore mental well-being. Variations in environmental preferences are particularly pronounced among individuals experiencing specific restorative needs, which can influence their preferences for either urban or natural settings. Furthermore, the degree to which an environment delivers on its restorative promise can refine these preferences. However, some studies, such as that by Roberts et al.(2021) [19], confirm that environments which are preferred can, in turn, become restorative. For example, museums can function as restorative spaces for regular visitors by facilitating self-regulation and stress reduction.Numerous studies have demonstrated that environmental preferences have a positive impact on perceptions of restorativeness, with higher preferences correlating with stronger restorative experiences [20]. Berg et al. (2003) [21] identified a positive relationship between individuals’ preference for a particular environment and its capacity to alleviate stress or mental fatigue, suggesting that those under stress or experiencing fatigue are more likely to benefit from the environment’s restorative qualities. Conversely, restorative perception has also been shown to influence environmental preference. Pals et al. (2009) [22], using the restorative environmental perception scale, found that dimensions such as charm, distance, and congruence are positively associated with environmental preference and can predict both preference and overall pleasantness. Liu et al. (2018) [23] constructed a path model examining the relationships among environmental preference, restorative perception, and health benefit assessment, using Fuzhou National Forest Park as their research site. Their findings indicated that the landscape environmental preference of visitors significantly enhances their restorative assessment.

The concept of place attachment, which originated in behavioral geography, has been widely studied globally and has garnered considerable attention in China since its introduction by scholars such as Huang Xiang in 2006. It is typically defined as the emotional bond between individuals and specific places, grounded in emotions, perceptions, and the passage of time [24]. Scannell & Gifford (2010) [25] reviewed numerous definitions and proposed a three-dimensional “person-cognitive process-place” framework to synthesize the concept of place attachment. In this framework, the person dimension reflects the perceived significance of the place to the individual; the cognitive process dimension encompasses the emotional, cognitive, and behavioral aspects of attachment; and the place dimension highlights the characteristics of the place itself, including the specificity of spatial elements and the importance of social or physical components. Place attachment enhances the psychological restorative effects of the environment, meaning that individuals’ emotional connections to aesthetically pleasing environments can aid in the recovery of physiological, psychological, and social capabilities that may have been depleted during adaptation to external stressors. Two primary theories explain the psychological restorative function of place attachment. The first is the Attention Restoration Theory, which emphasizes the environment’s role in restoring cognitive functioning. The second is the Psychological Evolution Theory, also known as the Stress Reduction Theory, which highlights the environment’s capacity to alleviate stress [26]. Place dependence influences all dimensions of restorative environmental perception, while place identity partially mediates these relationships. However, place identity affects dimensions of restorative environmental perceptions other than physical remoteness [27].

Despite the acknowledged benefits of forest therapy, a critical “black box” remains in understanding the cognitive-emotional processing that occurs between natural environments and human psychological restoration. Most studies assume a direct effect, yet environmental psychology suggests that human responses to landscape are mediated by evaluation (Preference) and emotional bonding (Attachment). Does a visitor need to like the forest for it to be restorative? Must they feel attached to it? The “person-cognitive process-place” framework suggests these are distinct steps. Furthermore, traditional linear regression models (like SEM) assume these relationships are uniform for all tourists. They fail to capture equifinality—the idea that different visitors might achieve restoration through different “recipes” of perception and emotion.

To address these gaps, this study integrates PLS-SEM and fsQCA to: (1) determine the mediating roles of Environmental Preference and Place Attachment; (2) test the serial mediation pathway; and (3) identify the configurational paths that lead to high restorative perception. This dual approach allows us to validate general mechanisms while respecting the complexity of individual tourist experiences. Notably, Akgüller et al. (2025) [28] proposed the fractional transfer entropy (FTE) framework, which captures complex interactive relationships by adjusting memory parameters and constructing directed weighted networks—an approach that resonates with the configurational thinking of fsQCA. Both methodologies emphasize moving beyond linear relationships to reveal multi-dimensional, context-dependent interaction patterns, providing a cross-disciplinary reference for analyzing the complex links between natural environments, human perceptions, and behavioral outcomes.

2. Literature review and hypotheses develop-ment

2.9. Chain mediation of environmental preference and place attachment

Preferred environments are more likely to be perceived as restorative. Studies have confirmed a link between place attachment and preference: restorative experiences enhance environmental preference, which in turn promotes place attachment [77]. Liu et al. (2018) [23] constructed an urban park restorative evaluation psychological model, empirically verifying the mechanism of environmental preference and place attachment on restorative perception via Fuzhou urban parks, confirming place attachment’s mediating role and proposing the sequence: “environmental preference → place dependence → place identity → restorative perception.” Natural environments easily trigger environmental preference; stronger preference correlates with higher restorative perception, and preference positively affects place attachment, which further influences restorative perception. Thus, natural environment perception may indirectly affect restorative perception through the chain mediation of environmental preference and place attachment. However, existing studies have gaps: few explicitly examine the chain mediating role of environmental preference and place attachment between natural environment perception and the five dimensions of environmental restorative perception, and the internal sequential mechanism of this chain mediation remains understudied—gaps addressed by the following hypotheses:

H29: Environmental preference and place attachment mediate between natural environment perception and compatibility.

H30: Environmental preference and place attachment mediate between natural environment perception and extent.

H31: Environmental preference and place attachment mediate between natural environment perception and mentally away.

H32: Environmental preference and place attachment mediate between natural environment perception and physically away.

H33: Environmental preference and place attachment mediate between natural environment perception and fascination.

Based on the research hypotheses, the conceptual model of this study was constructed as shown in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Conceptual model.

https://doi.org/10.1371/journal.pone.0343515.g001

3. Materials and methods

3.1. Study site

In this study, Zhangjiajie National Forest Park (Fig 2), which aligns closely with the study’s objectives, was selected as the research site. An empirical study was conducted by collecting relevant data from tourists through field surveys. The reasons for choosing this site are as follows: (1) Zhangjiajie National Forest Park, located in the Wulingyuan mountain range of Zhangjiajie City, covers an area of 130 square kilometers and is the first national forest park in China. In December 1992, the park was designated as a UNESCO World Nat-ural Heritage site due to its unique quartz sandstone Great Peak Forest. The park boasts abundant forest resources, including 517 species of woody plants. It is a rare tourist destination, offering opportunities for relaxation and immersion in nature, thanks to its breathtaking scenery and diverse flora and fauna in the high mountains and gorges. (2) The park attracts numerous domestic and international tourists due to its distinctive natural beauty. In 2023, the cumulative number of visitors to Zhangjiajie National Forest Park reached 7,012,000, with a peak of 4,872,000 ticket purchases at one time. This represents a year-on-year increase of 233.4% and a 13.4% increase compared to 2019. Given the high visitor volume, this site was chosen for the questionnaire research to gather a larger sample of tourist data, thereby enhancing the credibility of the study’s findings.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Zhangjiajie National Forest Park (Author’s own photos).

https://doi.org/10.1371/journal.pone.0343515.g002

3.2. Measurement items

The natural environment perception scale comprised seven items, the environmental preference scale contained four items, and the place attachment scale consisted of seven items adapted from Williams and Vaske, covering the dimensions of place dependence and place identity. The restorative environmental perception scale included 18 items. All scales employed a Likert 5-point format, with responses ranging from “strongly disagree” (1) to “strongly agree” (5). A higher score indicates a stronger perception of natural environment, environ-mental preference, place attachment, and restorative environment perception. The sources of scale indicators are detailed in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Measurement items and sources.

https://doi.org/10.1371/journal.pone.0343515.t001

3.3. Data collection

In this study, data was collected by randomly distributing paper questionnaires, and tourists participated in the survey and answered the questionnaires independently. The study used only anonymous, non-sensitive questionnaires and posed no more than minimal risk; data collection was carried out in the everyday flow of visitors. Because requesting written signatures would have created a record of identity and increased participants’ psychological burden, consent was obtained orally. Before handing out the questionnaire, the interviewer explained the study purpose, procedures, anonymity safeguards, and the right to withdraw at any time to each potential participant. After explicit verbal consent was given, the interviewer ticked a box on the front page of the questionnaire to document consent, and a member of the research team co-signed as a witness. The Academic Committee approved the use of verbal informed consent on the grounds that the study involved low risk and that obtaining written consent was imprac-tical. The questionnaires were distributed from June 22, 2024 to June 25, 2024. 490 questionnaires were distributed, 460 were recovered, and 432 were valid. The recovery rate was 94%. Table 2 exhibits the basic information of the respondents.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographic profile of respondents.

https://doi.org/10.1371/journal.pone.0343515.t002

In terms of gender, among the 432 tourists in this survey, there is a small difference between the number of male and female tourists, which is in line with the requirements of this questionnaire; in terms of age distribution, the largest proportion in this survey is 18–24 years old, accounting for 39.58%. People aged 25–34 and 35–44 come in second, with 25.69% and 15.28% respectively. In terms of education, the proportion of college degree and bachelor’s degree is the highest, at 48.61%, indicating that the population in this survey is generally better educated; in terms of jobs, the proportion of students and unit staff/workers is higher, at 41.67% and 13.89% respectively; in terms of annual personal income, people with less than 24,000 yuan predominate, followed by 24,000–60,000 and 60,000–120,000, while the proportion of other incomes is relatively balanced, which indicates that the economic incomes of the surveyed population are lower than the average level, which is mainly due to the fact that the surveyed popula-tion is mainly composed of students and unit workers. Regarding the number of visiting times, tourists who came to Zhangjiajie National Forest Park for the first time were in the majority, accounting for 67.36% of all tourists, followed by those who came for the second time and the fifth time and above, accounting for 14.12% and 10.88% respectively

4. Results

4.2. Reliability and validity tests

The total reliability of the questionnaire was tested using SPSS 21.0 to obtain its Cronbach’s Alpha value, if this value is higher than 0.8, it indicates high reliability, if this value is between 0.7–0.8, it indicates good reliability, if this value is between 0.6–0.7, it indicates acceptable reliability and if this value is less than 0.6, it indicates poor reliability. The analysis shows that Cronbach’s Alpha value = 0.940, which indicates that the questionnaire is highly reliable. The validity of the questionnaire scale should be tested in four aspects: content validity, criterion validity, construct validity and conjoint validity [90]. The criterion validity coefficient is commonly used to evaluate criterion validity, but it is not typically reported in tests because it is difficult to calculate and use [91]. The construction of the scale in this paper is mainly based on the existing research results at home and abroad, interviews with tourists and three rounds of expert opinion solicitation, and the indicators are screened through scientific methods, so the scale has high content validity. The degree and ability of the questionnaire scale’s content to measure theoretical abstract concepts is referred to as construct validity. The usual practice is to use the variance contribution of the first principal component of each latent variable to determine the variance. A general requirement of more than 40% is acceptable [91]. The test of convergent and discriminant validity of a questionnaire scale is known as conjoint validity. Convergent validity is tested by analyzing the standardized loading coefficients of each measurement item of the latent variable and extracting the average variance (AVE) [92]. The closer the standardized factor loadings are to 1, the better the effect is In this study, the standardized factor loadings of 0.6 were used as the criterion for the exclusion of indicators in the validation factor analysis stage. When the AVE value of each latent variable is greater than 0.5 and the standardized factor loading is greater than 0.6, the measurement question items can explain most of the variance variance of each latent variable, indicating that the scale has good convergent validity. Discriminant validity was determined by comparing the square root of the mean variance extracted from the latent variables with the absolute value of the correlation coefficients between the variables, and if the former was greater than the latter, it indicated that the variables had good discriminant validity [93]. The AVE and CR scores of the model are used to measure the convergent validity of the pairs of variables within a factor, and an AVE higher than 0.4 or a CR higher than 0.7 usually indicates high convergent validity.

Table 3 presents the descriptive statistics for the measurement items, including mean, standard deviation (SD), skewness, and kurtosis. The mean values for the constructs generally ranged from 3.60 to 4.56, indicating a positive reception of the natural environment and restorative qualities by the respondents. The standard deviations ranged between 0.63 and 1.09, reflecting a reasonable level of variability in the responses.The skewness values ranged from −1.51 to −0.25, and kurtosis values were largely within the acceptable range of −1 to +3. Although PLS-SEM is a non-parametric method and does not strictly require normal distribution, these results suggest that the data does not deviate severely from normality, ensuring the robustness of the parameter estimates.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Descriptive statistics and normality test.

https://doi.org/10.1371/journal.pone.0343515.t003

Furthermore, lateral collinearity was assessed to ensure that common method bias or high correlations between predictors did not distort the results. As indicated in Table 3, the Variance Inflation Factor (VIF) values for all items ranged from 1.76 to 2.71. All values are well below the conservative threshold of 3.0 (and the lenient threshold of 5.0), confirming that multicollinearity is not a significant issue in this study. This validates the distinctiveness of the constructs and supports the structural model’s reliability.

According to Table 4, a total of eight factors corresponding to the combined reliability CR values are greater than 0.7, which can indicate that the questionnaire scales have high convergent validity, indicating that the extraction of the measures within the factors is more excellent.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Confirmatory factor analysis.

https://doi.org/10.1371/journal.pone.0343515.t004

As can be seen from Table 5, the bolded values in the graph are AVE square root values, which are used to measure the discriminant validity of the scale, if the AVE square root value of one of the factors is greater than the correlation coefficient between it and the other factors, then it means that this factor has a more excellent discriminant validity. For example, the AVE square root value of place attachment in the table is 0.677, which is greater than the absolute maximum value of the correlation coefficient between the factors, 0.563, which means that it has good discriminant validity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Pearson correlation and the square root of AVE.

https://doi.org/10.1371/journal.pone.0343515.t005

4.3. Structural equation modeling test

Based on the established fit criteria, an SRMR < 0.05 indicates an ideal model fit, whereas d_ULS < 0.95 and d_G < 0.95 signify good model fit [94]. The empirical results show that the model achieves an SRMR of 0.033, a d_ULS of 0.719, and a d_G of 0.422, all of which satisfy the recommended thresholds, thus confirming that the structural equation model demonstrates a satisfactory fit to the data.

Meanwhile, the relationships among Natural Environment Perception (NEP), Environmental Preference (EP), Place Attachment (PA), and Restorative Environmental Perception (REP) were further examined via structural-equation modeling; the results are reported in Table 6 and illustrated in Fig 3. Of the 18 hypothesized paths, 5 were non-significant – namely, Natural Environment Perception (NEP) → Place Attachment (PA), Natural Environment Perception (NEP)→Compatibility (C), Natural Environment Perception (NEP) →Extent (E), Natural Environment Perception (NEP) → Physically away (PAY), and Environmental Preference (EP) →Physically away (PAY) – leading to the rejection of hypotheses H2, H4, H5, H7, and H12. The remaining 13 hypotheses received empirical support.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Results of model output.

https://doi.org/10.1371/journal.pone.0343515.t006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. PLS-SEM output results.

https://doi.org/10.1371/journal.pone.0343515.g003

4.4. Mediation effect test

As can be seen from Table 7, the mediating roles of Place Attachment (PA) between Natural Environment Perception (NEP) and Compatibility (C), Natural Environment Perception (NEP) and Extent (E), Natural Environment Perception (NEP) and Mentally away (MA), and Natural Environment Perception (NEP) and Fascination (F) are not significant, so Hypothesis H24, H25, H26 and H28 are invalid. Furthermore, the mediating role of Environmental Preference (EP) in the relationship between Natural Environment Perception (NEP) and Physically away (PAY) was not significant; thus, Hypothesis H22 is rejected. The remaining ten hypotheses concerning mediation effects are empirically supported.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. Mediation effect test results.

https://doi.org/10.1371/journal.pone.0343515.t007

4.5. Multiple‐path analysis based on fsQCA

4.5.1. Variable selection and data calibration.

The PLS-SEM results indicate that Natural Environment Perception, Environmental Preference, and Place Attachment exert significant effects on all five dimensions of Restorative Environmental Perception. Yet it remains unclear whether these three antecedents operate configurationally and which specific combinations of conditions are sufficient for high Restorative Environmental Perception. We therefore retain these factors as the causal conditions for the subsequent fsQCA, aiming to uncover their joint effects from a configurational perspective. Prior to the fsQCA, the survey data must be transformed into fuzzy sets with membership scores ranging from 0 to 1 to enhance interpretability. Because fsQCA requires unidirectional variables, multi-item constructs were first reduced to a single score by averaging their respective indicators. Following Ragin’s guidelines, three qualitative anchors were specified: the 5th percentile, the 95th percentile, and the 50th percentile [95]. Table 8 presents the fuzzy-set calibration anchors for all conditions.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Calibration anchors for all constructs.

https://doi.org/10.1371/journal.pone.0343515.t008

4.5.2. Necessity analysis of single conditions.

Prior to examining configurational solutions, we conducted a necessity analysis with fsQCA 3.0 to determine whether any individual antecedent condition is required for the outcome Restorative Environmental Perception. Following Fiss [96], a condition is deemed necessary when its consistency score exceeds 0.9. The results, reported in Table 9, show that all three single conditions—Natural Environment Perception, Environmental Preference, and Place Attachment—surpass this threshold. Consequently, each condition constitutes a necessary prerequisite for high Restorative Environmental Perception and must be retained in the subsequent sufficiency analysis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Necessity analysis of antecedent conditions for restorative environmental perception.

https://doi.org/10.1371/journal.pone.0343515.t009

4.5.3. Configurational analysis: Sufficiency and solution types.

Given the medium-N sample of 432 valid questionnaires, we adhered to Schneider and Wagemann’s [97] guideline that at least 75% of all empirically possible cases should be retained. Inspection of the truth-table distribution led to a frequency threshold of 7; the results for each outcome are reported in Table 10. fsQCA 3.0 generated three solution types—complex, parsimonious, and intermediate. Following current best practice, we report the intermediate solutions and use the parsimonious solutions to identify core conditions. All individual configurations and the overall solutions exceed the 0.75 consistency benchmark recommended for sufficiency analyses (Ragin, 2008).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. Configurational paths of antecedents leading to restorative environmental perception.

https://doi.org/10.1371/journal.pone.0343515.t010

For Compatibility, two equifinal configurations emerge, with an overall consistency of 0.821 and coverage of 0.865, indicating high explanatory power. Extent also exhibits two sufficient pathways (overall consistency = 0.851; coverage = 0.751). Likewise, Mentally away is produced by two configurations (consistency = 0.810; coverage = 0.863). Willingness to physically away yields a single sufficient configuration (consistency = 0.865; coverage = 0.590). Finally, fascination is covered by one configuration (consistency = 0.782; coverage = 0.883). In sum, all models demonstrate satisfactory fit and substantive explanatory capacity.

Configuration paths with Compatibility as the output variable: Configuration 1 has high environmental preference as its core condition. This configuration path indicates that when an individual’s environmental preference is high, their compatibility value also increases. Configuration 2 has high natural environment perception and high environmental preference as its core conditions. This configuration path indicates that when an individual’s natural environment perception and environmental preference are both high, their compatibility value also increases.

Configuration paths with extent as the output variable: Configuration 1 has natural environment perception as its core condition. This type of configuration path indicates that when an individual’s natural environment perception is high, their extent will also be high. Configuration 2 has high natural environment perception and high place attachment as its core conditions. This type of configuration path indicates that when an individual’s natural environment perception and place attachment are both high, their Extent will also be high.

For configuration paths with Mentally away as the output variable: Configuration 1 has natural environment perception as its core condition. This type of configuration path indicates that when an individual’s the natural environment perception is high, their mental away is also high. Configuration 2 has high place attachment as its core condition. This type of configuration path indicates that when an individual’s place attachment is high, their Mentally away is also high.

For configuration paths with physically away as the output variable, the core conditions are natural environment perception, environmental preference, and place attachment. This configuration path indicates that when an individual has high natural environment perception, environmental preference, and place attachment, their Physically away will also be high.

For configuration paths with fascination as the output variable: Configuration 1 has natural environment perception as its core condition. This configuration path indicates that when an individual’s natural environment perception is high, their fascination will also be high. Configuration 2 has high environmental preference as its core condition. This configuration path indicates that when an individual’s environmental preference is high, their Fascination will also be high.

5. Discussion

The ecological benefits and economic value of forest parks are extremely high. Forest parks are also rich in psychological and physiological resources, in addition to their recreational and educational functions. It is crucial to pay attention to the impact of the natural environment of forest parks on human psychology. Structural equation modeling is utilized in this study to explore the relationship between the natural environment and restorative environmental perception, and to verify the mechanism and path of the natural environment’s influence on restorative environmental perception. The results showed that 22 out of the 33 pre-proposed path relationships were valid. The results are being discussed and analyzed in the following.

The findings unequivocally identify Environmental Preference as a cognitive ‘gatekeeper’ in the restorative process. Unlike the Stress Reduction Theory which emphasizes automatic physiological responses, our results suggest that for the Zhangjiajie tourist, restoration is cognitively mediated. The sensory input of the natural environment perception must first pass through an evaluative filter—’Do I like this?’—before it can trigger the deeper dimensions of fascination and being away. If this aesthetic evaluation is negative, the restorative chain is effectively blocked, regardless of the objective quality of the nature.

The natural environment perception exerts a significant and positive influence on individuals’ restorative perception. Humans possess an innate connection with nature that not only supplies material necessities for survival but also harbours vital psychological resources indispensable for modern life and essential to maintaining physical and mental health [98]. Accumulating evidence indicates that natural environments are valued not merely for their intrinsic worth but also for their recreational functions [99] and, more importantly, as a critical resource for promoting psychological and physiological well-being [100–103]. Even brief encounters with nature or passive observation of surrounding natural scenery can effectively alleviate mental fatigue. The present findings corroborate the restorative capacity of natural environments: natural environment perception significantly enhances both the “Mentally away” and “fascination” dimensions of restorative experience, whereas its effects on compatibility, extent, and Physically away remain non-significant.

Natural environments inherently provide the quality of “being away,” enabling individuals to detach from daily stressors and fatigue and thereby replenish their psychological resources. Attention Restoration Theory (ART) posits that this sense of separation is essential for psychological recovery [104]. The captivating and novel features of natural settings capture individuals’ involuntary attention, fostering deep immersion and a sense of fascination that not only facilitates restoration but also exerts a significant positive effect on well-being [104]. Compatibility refers to the congruence between environmental affordances and individuals’ needs. Although natural environments are generally perceived as highly compatible, situational factors—such as temperature or humidity falling outside an individual’s comfort range—may attenuate this effect [105]. Extent denotes the perceived scope or richness of the restorative environment; while natural settings typically offer high restorative potential, the perceived intensity may sometimes be insufficient to elicit a significant restorative response when individuals’engagement with the environment is low. “Physically away” describes objective distance from everyday surroundings. Although physical separation can contribute to restoration, its impact is contingent upon concurrent psychological detachment; if the mind remains preoccupied, the restorative benefits of physical relocation are markedly diminished [106].

The results reveal that environmental preference fully mediates the relationship between natural-environment perception and restorative perceptions. Although natural environment perception exerts direct and significant effects only on the Mentally away and Fascination dimensions of restorative perception, the mediating role of environmental preference amplifies and extends its influence to four of the five restorative dimensions—namely Compatibility, Extent, Fascination, and Mentally away—while excluding Physically away. Environmental preference thus functions as an automatic appraisal mechanism through which individuals rapidly determine whether a given setting is worthy of approach or avoidance, thereby translating initial perceptions of nature into broader restorative experiences [11,107,108].Individuals’ preference for natural environments is shaped by their perception of these settings’ superior restorative potential. For environmental preference to function as a mediator (or moderator), three conditions must be satisfied simultaneously: (1) a significant association between the type of environment and preference; (2) a significant association between the type of environment and perceived restorative potential; and (3) a statistically significant association between restorative potential and preference when the influence of environment type on preference is held constant. Only when all three conditions are met can environmental preference effectively exert its mediating influence [109,110]. Contrary to these theoretical claims, our results indicate that place attachment does not significantly mediate the relationship between natural-environment perception and restorative perception. This finding diverges from prior assertions that the restorative quality of highly cherished settings stems from their capacity to offer a secure and comfortable context, thereby enabling self-regulation, stress reduction, and the redirection of attention toward problem-solving and self-reflection—processes that ultimately help individuals organize their thoughts and emotions [111]. This non-significant effect may be attributable to the multidimensional nature of place attachment. Its two core dimensions—place dependence and place identity—can exert divergent influences on perceived restoration, while natural-environment perception and restorative perception themselves retain a degree of conceptual and empirical independence. Certain environmental qualities may therefore impinge directly on restorative outcomes without necessitating the involvement of place attachment as an intermediary. Additionally, contextual characteristics, individual differences, and limitations inherent to research design and measurement instruments may further constrain the mediating capacity of place attachment. Future studies should disentangle the distinct roles of place-dependence and place-identity across varied environmental settings and individual profiles, thereby yielding a more nuanced understanding of this phenomenon [27,112]. Future research should therefore delve into the distinct mechanisms underlying the dimensions of place attachment and examine their mediating effects across diverse environmental contexts and individual profiles, thereby providing a more comprehensive understanding of this phenomenon. The present study further reveals that environmental preference and place attachment jointly operate as a significant serial mediator between natural-environment perception and restorative perception. This sequential effect likely arises because environmental preference functions as an initial bridge, intensifying individuals’ emotional bonds with a specific setting and thereby fostering the formation of place attachment [113]. Once established, place attachment deepens psychological investment in the environment, motivating individuals to actively seek restorative experiences within that locale. Such heightened engagement facilitates the recognition of restorative qualities, thereby elevating overall restorative perceptions [114]. Moreover, a serial mediation pathway can accumulate the effects of multiple mediators, amplifying the influence of natural environment perception on restorative perception [115]. Consequently, the significant serial mediation of environmental preference and place attachment is attributable to the former’s role in strengthening emotional ties and the latter’s role in deepening psychological commitment, jointly enhancing restorative experiences and, ultimately, restorative perceptions.

This study finds that place attachment positively impacts certain dimensions of the restorative environmental perception, specifically compatibility and degree, aligning with some prior research. Yang (2018) [77] highlighted a strong correlation between place attachment to one’s origin and restorative perception, noting that stronger attachment enhances restorative experiences in familiar environments. Chi & Su (2012) [76] showed that attachment to an environment affects the restorative impact of natural scenes on emotional and attentional states, with only attached environments providing restorative benefits. Conversely, low attachment can lead to negative emotional activation, hindering psychological recovery. Feng & Cui (2015) [116] found that high-quality urban environments foster restorative experiences, thereby strengthening place attachment, with physical environment quality being a key predictor. Place attachment aids self-regulation by reducing cognitive load and enhancing positive emotions, supporting goal attainment [117–119]. Favorite places, which are often sites of strong attachment, facilitate self-reflection and stress relief, crucial for self-regulation [25]. Wang & Zheng (2015) [120] observed that emotional restoration in green spaces was more significant for individuals from grassland origins, suggesting that origin environments influence restorative perceptions. Liu et al. (2017) [121] and Peng (2017) [122] explored the interplay between place attachment and restorative perception in recreational and community park settings, revealing that place attachment and environmental characteristics jointly influence restorative effects. However, this study’s findings differ from Xi et al. (2021) [27], who reported broader impacts of place attachment on restorative perception in resort settings, possibly due to differences in study sites. Future research could compare natural and man-made environments to further elucidate these dynamics.

According to Liu et al. (2019) [68], environmental preference has a significant and positive impact on four dimensions of restorative perception except physically away, and environmental preference has a significant and positive impact on place attachment. Individuals’ preferred place environments can promote the emergence of a sense of attachment to the environment and the development of individual emotions, thus triggering an individual’s emotional connection to the environment, generating a sense of place identity, and further contributing to the individual’s acquisition of a restorative experience. The environmental preference has a significant and positive impact on place attachment; the more favorable the place environment is, the more attachment is created, which is consistent with Zhu (2020)’s findings [123]. The study also confirmed that environmental preferences play a partial mediating role between perceptions of the natural environment and perceptions of the restorative environment. Perceptions of the natural environment influenced the degree dimension of the restorative environmental perception through the chain mediation of environmental preferences and place attachment. This is in accordance with the findings of Faraji et al. (2024) [124].

The PLS-SEM findings largely converge with those of the fsQCA. Environmental preference exerts a significant mediating effect and consistently appears as a core condition across alternative configurational paths in the fsQCA, whereas the mediating role of place attachment is nonsignificant. Natural environment perception demonstrates a pronounced positive effect on both mental away and fascination in the PLS-SEM and is likewise identified as a core condition in the fsQCA solutions. Complementing the PLS-SEM results, the fsQCA not only corroborates the symmetric relationships uncovered but also offers a more holistic and nuanced understanding of the antecedents of the restorative environmental perception through configurational analysis.

Contrary to Hypothesis H24-H28, PLS-SEM results showed that place attachment alone did not significantly mediate the relationship between natural environment perception and restorative environmental perception. This finding diverges from studies emphasizing the restorative role of “favorite places”. However, the fsQCA results provide a nuanced explanation: place attachment appears as a peripheral condition in high-restoration configurations.

This suggests that for the typical tourist in Zhangjiajie (often a first-time visitor), restoration is primarily driven by the immediate sensory impact of the natural environment perception. The formation of place attachment typically requires repeated interaction and time, which may not be present for all tourists. Thus, while place attachment can enhance restoration (as shown in the serial mediation path), it is not a necessary condition. High restoration can occur purely through strong nature perception and preference, even in the absence of deep attachment. This distinction between “core” (environmental preference) and “peripheral” (place attachment) factors is a key theoretical contribution of this study, refining our understanding of how restoration occurs in tourism contexts compared to residential contexts.

A notable limitation of this study is the seasonality of data collection. The survey was conducted in June, a period characterized by lush vegetation and vibrant natural scenery in Zhangjiajie. According to Pratiwi et al. (2019) [125], landscape imagery and restorative potential can vary significantly across seasons. Winter landscapes, for instance, might evoke different levels of fascination or compatibility due to reduced greenery. Therefore, our findings primarily reflect the ‘summer mechanism’ of restoration. Future studies should adopt a longitudinal approach to examine how seasonal changes in natural environment perception moderate the restorative process.

6. Conclusion

In this study, we constructed a model of the relationship between natural environment perception, environmental preference, place attachment, and restorative environmental perception, and proposed 33 research hypotheses. At the same time, Zhangjiajie National Forest Park was selected as a case study, and the data were obtained through a questionnaire survey and analyzed by Smart PLS 4.0 and fsQCA 3.0. The following conclusions were drawn: (1) Two dimensions of restorative environmental perception (mentally away and fascination) are significantly influenced by natural environment perception. The more tourists perceive the natural environment subjectively, the greater their mentally away and fascination. (2) Environmental preference is significantly influenced by natural environment perception. The more tourists perceive the natural environment subjectively, the more they tend to prefer it. (3) The relationship be-tween natural environment perception and restorative environmental perception was significantly mediated by environmental preference. However, the mediating effect of place attachment alone was not significant. In contrast, the chain mediation effect of environmental preference and place attachment was significant. Additionally, fsQCA was employed to complement the symmetric findings of PLS-SEM. The configurational analysis confirmed that environmental preference consistently functioned as a core condition across multiple pathways leading to high levels of restorative perception outcomes. Natural environment perception also emerged as a core condition in nearly all high-outcome configurations, while place attachment played a more conditional and peripheral role. These findings reinforce the mediating dominance of environmental preference and provide a more nuanced, asymmetric understanding of how combinations of conditions influence tourists’ restorative experiences.

Supporting information

Acknowledgments

We express our gratitude to the respondents and investigators for their participation in collecting data for this study.