The Impact of Natural and Cultural Landscape Quality on Attachment to Place and the Intention to Recommend Tourism in a UNESCO World Heritage City

Abstract

Cultural landscapes in World Heritage cities are attracting a growing global tourist population. Given the limitations of self-report methods in capturing tourists’ immediate and deep perceptions, and the lack of comprehensive investigation into the cultural types and naturalness of landscapes, this study aims to investigate how cultural landscape types influence tourists’ recommendation intention through the mediating roles of place attachment and perceived restorativeness while examining the moderating effect of landscape naturalness. Integrating Place Attachment Theory (PAT), Attention Restoration Theory (ART), and the Associative–Propositional Evaluation Model (APE), three studies were conducted using behavioral and neurophysiological approaches. Study 1, a scenario-based experiment, revealed that high-culture landscapes enhance recommendation intention via place attachment, with the effect of perceived restorativeness being stronger under low naturalness conditions. Study 2, an event-related potential (ERP) experiment, showed that landscapes with low culture and low naturalness elicit stronger emotional responses, as indicated by heightened P2 and LPP amplitudes. Study 3 demonstrated the efficacy of a Decision Tree model in classifying landscape naturalness based on EEG features. This study deepens the understanding of the complexity of tourist experiences in cultural heritage sites, provides new evidence for the application of Place Attachment Theory in tourism contexts, and offers scientific foundations and practical implications for optimizing landscape design in heritage sites, enhancing tourist experiences, and exploring brain–computer interface applications in the tourism field.

1. Introduction

World Heritage cities, by virtue of their unique Outstanding Universal Value (OUV), are attracting an ever-increasing global tourist population. These cities not only house rich historical sites but also embody unique intangible cultural heritage (ICH), such as traditional crafts, festive events, and folk performances, which significantly stimulate tourist engagement [1]. Cultural tourism, as a profound experiential modality, has become an integral component of the global tourism industry. Particularly in World Heritage cities, tourists gain unique novelty and cultural identity through immersive experiences with ICH [1]. This trend not only corroborates the central status and continually rising importance of cultural tourism in the global market [2] but also presents new demands and opportunities for the sustainable development of heritage sites. In an increasingly competitive tourism market, tourists’ recommendation intention, which refers to their propensity to recommend a site to others, has emerged as a crucial indicator for gauging the success of the tourism experience and influencing destination attractiveness [3]. Consequently, an in-depth exploration of factors influencing tourists’ recommendation intention in World Heritage cities, particularly elements related to the ICH experience, holds significant theoretical and practical implications for enhancing heritage site management and promotional strategies.

Current research on cultural tourism has elucidated its importance and impact from multiple perspectives. On one hand, scholars, from macroeconomic and socio-developmental viewpoints, have investigated the contributions of cultural tourism to destination economic benefits, employment opportunities, and social advantages [4]. Furthermore, a substantial body of research has concentrated on the micro-psychological aspects of cultural tourists, including cultural motivation [5], authenticity perception [6], cross-cultural experiences [7], and post-tour satisfaction and behavioral intentions [8]. On the other hand, studies have also focused on cultural tourism promotion strategies [9], analyzing the role of brand image construction, digital marketing, and intelligent technologies in enhancing destination promotion [10]. However, while these studies offer valuable insights into the macro-development and promotion of cultural tourism, there remains scope for enhancement in focusing on tourists’ immediate and profound perceptions of the attributes of tourism attractions themselves, such as the specific cultural symbols or environmental characteristics they embody. Specifically, even though some studies have attempted to explore tourists’ feelings about attractions, data collection predominantly relies on traditional questionnaire surveys and interview methods [11]. Although such self-report methods can capture tourists’ subjective attitudes and evaluations, they are potentially susceptible to interference from factors such as social desirability bias [12] and recall bias [13]. This susceptibility makes it challenging to comprehensively and objectively reveal tourists’ more covert and immediate physiological arousal and neurocognitive responses in specific scenarios, thereby lacking an objective assessment of tourists’ authentic experiences at scenic sites from physiological or neuroscientific perspectives. Electroencephalography (EEG) methods are useful tools for tourism marketing research [14]. Moreover, previous research, when classifying tourist attractions, has tended to categorize them based on their extrinsic functions (e.g., recreational, cultural) [15], with less attention paid to the differential impact of inherent cultural attributes and landscape naturalness on tourist experience and subsequent behavior.

Previous studies on tourism experience, in terms of methodology and perspective, have largely focused on capturing tourists’ macro-level feelings and explicit attitudes through traditional means such as questionnaires and interviews [11]. While these studies have provided an important basis for understanding overall tourist satisfaction and behavioral intentions, they primarily reflect evaluations filtered through cognitive processing and conscious screening. These methods mainly focus on tourists’ propositional attitudes expressed at the conscious level, i.e., views and evaluations directly reflected through language or text. However, the influential Associative–Propositional Evaluation (APE) model in social psychology posits that individual attitudes are not monolithic but comprise two relatively independent components [16]. This model distinguishes between “propositional attitudes”, based on logical inference and conscious judgment, and “associative attitudes”, formed by automatically activated, rapid, and often unconscious affective associations with an object. In light of this, current tourism user research primarily employs self-report methods to measure tourist experiences and evaluations, which, in essence, reveals more about their “propositional attitudes”. Therefore, “associative attitudes”, which can reflect tourists’ more immediate, spontaneous, and even subconscious reactions when encountering tourism attractions, have received insufficient attention and exploration in existing research. Prior studies on tourism experience have often overlooked the comprehensive analysis of this dual attitude, failing to integrate propositional and associative attitudes, leading to an incomplete understanding of the authentic tourist experience. To more comprehensively and deeply investigate the complexity of the tourism experience, this study integrates the dual perspectives of “propositional attitudes” and “associative attitudes”, exploring how they jointly influence tourists’ overall perception and subsequent behavior.

The primary purpose of this study is to understand how cultural landscape characteristics in World Heritage cities influence tourists’ willingness to recommend destinations to others and to identify the underlying psychological and neurophysiological mechanisms driving these behavioral intentions. Grounded in Place Attachment Theory (PAT), Attention Restoration Theory (ART), and the APE model, this study, using the World Heritage city of Quanzhou as a case, systematically investigates how different cultural landscape types influence tourists’ recommendation intention. It explores the dual mediating pathways of place attachment and perceived restorativeness while also examining the moderating role of landscape naturalness (high/low). To address these objectives, a multi-method research design is employed, integrating scenario experiments for self-reported data with electroencephalography (EEG) to capture neurophysiological responses; machine learning techniques will further be utilized to analyze EEG data concerning landscape naturalness. Study 1 utilizes scenario experiments to test the dual pathways through which cultural landscape types influence recommendation intention via place attachment and perceived restorativeness while examining the moderating role of landscape naturalness. Study 2 employs EEG technology to analyze tourists’ neurophysiological responses, focusing on P2 [17] and Late Positive Potential (LPP) components [14]. Study 3 develops machine learning algorithms (SVM, DT, GBDT) to classify landscape naturalness based on EEG signals. The principal contributions of this study are anticipated to be as follows: theoretically, a novel application and extension of Place Attachment Theory to elucidate tourist behavioral decisions shaped by cultural landscape types in World Heritage settings; and methodologically, the pioneering integration of self-reported explicit attitudes with neurophysiologically derived (EEG) implicit responses. This dual-perspective approach, complemented by machine learning analysis of EEG data, offers a more comprehensive and nuanced understanding of tourist engagement with cultural landscapes. The findings are expected to offer novel insights for sustainable heritage tourism management and the development of cultural heritage sites.

2. Literature Review and Research Hypotheses

2.1. Research Related to Cultural Tourism

Cultural tourism, as a form of tourism centered on experiencing and appreciating the cultural heritage, arts, lifestyles, and traditions of specific regions, has emerged as one of the most dynamic and growth potential sectors within the global tourism industry [18]. On one hand, research into the economic and social impacts of cultural tourism constitutes a significant pillar of this field. A substantial body of empirical research has confirmed the contributions of cultural tourism to destination economic benefits, employment opportunities, and social advantages [4]. Numerous scholars have dedicated efforts from economic, sociological, and managerial perspectives to assess the contribution of cultural tourism to the economic growth of destinations, for instance, its positive roles in job creation [19], international tourism growth [20], the promotion of infrastructure modernization, and the stimulation of related industries’ development [21]. Concurrently, the socio-cultural effects of cultural tourism have garnered considerable attention, including its complex influences on the revitalization and protection of cultural heritage [22], the promotion of cross-cultural understanding and identity construction [23], and the enhancement in community cohesion and residents’ cultural pride. These studies fully affirm the immense value of cultural tourism as an engine for regional economic development and an important medium for socio-cultural exchange, providing a solid theoretical basis and macro-level guidance for governmental cultural tourism development policies, industry strategic planning, and the practical work of cultural heritage management institutions. Tourism is not merely about people visiting places; it is a fundamental process through which culture itself travels, interacts, and transforms [24]. Another important branch has gradually shifted towards exploring individual cultural tourist behavior and experiences at the micro-level, such as tourists’ cultural motivation [5], authenticity perception [6], cross-cultural experiences [7], and satisfaction and behavioral intentions [8]. For example, tourists’ perception of authenticity affects the quality of the tourism experience through multiple paths. Perceived authenticity can not only trigger existential authenticity and promote the formation of memory and the improvement in subjective well-being [25], but can also significantly predict replay intentions through the experiences of constructive authenticity and existential authenticity [26]. However, the concept of authenticity is both practical and controversial in tourism theory and experience. The concept of authenticity is both a contested, complex construct in tourism research and a recurring theme in tourists’ descriptions of place and cultural experiences; by revealing how authenticity is used in theory and tourism activities, this tension can be partially resolved [27]. Tourist satisfaction is closely related to their chosen mode of travel; active travel modes (such as self-driving, hiking, etc.) often lead to higher satisfaction, while users of public transport typically report lower satisfaction with their trips [28]. On the other hand, research has also focused on the promotion strategies of cultural tourism [9], analyzing the role of brand image construction, digital marketing, and intelligent technologies in enhancing destination promotion [10]. Brand image construction is a vital part of cultural tourism promotion. Through narrative design, the unique cultural characteristics of a destination can be effectively conveyed to potential tourists. Stories have significant application value in tourism experience design and destination promotion. The tourism experience and destination image can be enhanced by telling stories [29]. Destination image is very important for increasing tourist satisfaction and recommendation intention [30]. The effectiveness of electronic word-of-mouth (eWOM) in influencing tourists’ travel decisions is mainly driven by its content richness and the perceived ease of use of the word-of-mouth platform [31]. However, the rapid development of cultural tourism has also brought about the challenge of overtourism. Overtourism refers to the phenomenon that when the number of tourists exceeds the carrying capacity of the destination, it has a negative impact on the local environment, cultural heritage, and the quality of life of residents. For historical cities, excessive tourism may lead to physical damage to cultural heritage, commercialization of local culture, rising living costs for residents, and weakening of community identity [32]. From a policy perspective, the Chinese government has in recent years promoted policies integrating culture and tourism to foster the development of the tourism industry. These integration policies have significantly improved tourism performance, and in regions with higher levels of economic development or urbanization, the positive impact of integration policies on domestic tourism has been more pronounced [33].

Despite the significant progress achieved by the aforementioned studies at their respective levels, which has laid an important foundation for the development of cultural tourism studies, a deeper examination reveals several undeniable limitations in the existing literature regarding the comprehensive disclosure of the complex psychological mechanisms and decision-making processes of cultural tourists. Firstly, from the perspective of research emphasis, existing research outcomes still exhibit, to a certain extent, a bias towards macroeconomic and socio-developmental effects. Conversely, there is a relative insufficiency of attention paid to the immediate, contextualized, deep psychological responses and dynamic emotional experiences generated by individual tourists when facing specific cultural symbols and landscapes within particular cultural settings. Tourism experience is the comprehensive and subjective experience obtained by tourists through multiple sensory perceptions, including the reactions of tourists at the cognitive, emotional, and behavioral levels during the tourism process [34]. The senses of tourists play a key role in the tourism experience [35]. A profound understanding of the tourist’s inner experiential world is crucial for the essential cognition of tourism phenomena; however, this micro-psychological perspective still requires further systematization and deepening in specific cultural landscape perception research. More critically, even in studies focusing on micro-tourist experiences, there are significant methodological bottlenecks, namely an over-reliance on traditional self-report methods. Such methods are undoubtedly effective in capturing explicit, proposition-based attitudes formed after tourists’ careful deliberation and logical reasoning. These attitudes reflect beliefs, judgments, and evaluations that tourists can clearly articulate [16]. However, human perception, decision-making, and behavior are not entirely governed by conscious, rational thought. A core deficiency in current research approaches is their substantial neglect or failure to effectively measure the implicit, association-based attitudes that may arise spontaneously, more immediately, and even subconsciously when tourists encounter cultural landscapes [16]. Such associative attitudes often stem from rapid, automated emotional reactions and cognitive associations formed from an individual’s past experiences, emotional memories, and cultural schemata. This omission in research orientation directly results in a potentially partial, or even biased, understanding of the full picture of tourists’ authentic experiences and the driving mechanisms of their behavioral intentions. For instance, tourists might report experience evaluations in questionnaires that are inconsistent with their actual implicit feelings due to social desirability effects, recall bias, or cognitive dissonance. Although a few forward-looking studies have begun to introduce psychophysiological methods [15], research that systematically integrates these two types of psychological processes (propositional and associative attitudes) and deeply investigates their interaction in cultural landscape perception and evaluation, as well as their impact on subsequent behavioral intentions, remains exceedingly scarce in the field of cultural tourism. This severely limits our comprehensive insight into the complete cognitive and emotional interaction processes of tourists in complex cultural tourism contexts.

2.2. Research Related to Place Attachment and Place Attachment Theory

Place Attachment Theory is a theory that explains the emotional, cognitive, and behavioral connections between individuals and specific places [36]. The theory emphasizes that place attachment is not only an emotional experience but also involves functional dependence and cultural significance. In recent years, it has been increasingly applied in tourism research to analyze the emotional connection between tourists and destinations [37].

In the realm of tourism research, place attachment is widely used to explain the relationship between tourist behavior and destination loyalty. Place-oriented concepts (cognitive and affective destination image and place attachment) are better predictors of destination loyalty than person-oriented concepts (cultural distance, social distance, and emotional solidarity) [38]. Destination attractiveness can significantly enhance tourists’ subjective well-being and place attachment, which in turn further promote destination loyalty [39]. Sop et al. (2023) demonstrated its positive impact on tourists’ place attachment and word-of-mouth communication [40]. These studies provide important perspectives for understanding the deep connections between tourists and destinations. Tourists’ positive and negative emotions influence their recommendation intention through place attachment [41]. Place attachment plays a significant role in predicting residents’ support for tourism development; high place attachment often leads to oppositional behavior, yet residents who have positive emotions towards tourism-induced place changes are less likely to exhibit oppositional behavior [42].

This study posits that cultural landscape type indirectly influences tourists’ recommendation intention by shaping their place attachment. Specifically, cultural landscape type, as a carrier of the destination’s physical and cultural characteristics, has an important impact on the formation of tourist place attachment. Different cultural landscapes carry unique historical memories, cultural meanings, and aesthetic values, which can evoke emotional resonance in tourists [43]. In particular, high-culture landscapes, which typically contain rich historical and cultural elements, artistic value, and aesthetic quality, can provide deeper cultural experiences. Cultural heritage with profound cultural connotations can enhance tourists’ place identity [44]. Place attachment is closely related to the attractiveness of landscapes with natural characteristics and those containing historically significant elements [45]. High-culture landscapes indirectly influence tourists’ recommendation intention by affecting their place attachment. Li et al. (2023) pointed out that specific landscape cultural features influence tourists’ place attachment by providing unique experiences, which in turn affect their subsequent behaviors [46]. High-culture landscapes offer tourists unique aesthetic and cultural experiences; these experiences are transformed into place attachment through cognitive and emotional processing. Place attachment influences tourists’ recommendation intention [41]. It indirectly affects individuals’ word-of-mouth communication [40]. Place attachment plays a key mediating role in this process, transforming external stimuli (cultural landscape type) into internal motivation (recommendation intention). Based on the above theoretical deduction, this study proposes the following hypothesis:

H1: 

Place attachment mediates the relationship between cultural landscape type and recommendation intention.

2.3. The Moderating Role of Landscape Naturalness and the Mediating Role of Perceived Restorativeness

Landscape naturalness generally refers to the proportion and characteristics of natural elements within a landscape, including vegetation cover, water body distribution, and topographical features, relative to the degree of human intervention. Naturalness is considered a key factor influencing human environmental perception and psychological responses [47] and is defined as the degree to which a landscape approaches its perceived natural state [48]. Attention Restoration Theory (ART) posits that natural environments have a positive impact on mental health, aiding in the restoration of attention and the alleviation of mental fatigue [49]. Existing research indicates that landscape naturalness is not only an important representation of ecosystem function but also a potential driver influencing human emotions, cognition, and behavior [50]. In the field of tourism research, the impact of landscape naturalness on tourist experience and behavioral intentions has received widespread attention. Numerous studies have confirmed the importance of natural landscapes as an effective factor in shaping tourists’ cognitive image [51]. Tourists develop a sense of place attachment during nature-based tourism, thereby enhancing their loyalty to nature tourism [52]. Tourism performances that integrate natural scenic beauty with cultural resources can enhance environmental restorativeness, thereby increasing destination loyalty [53]. However, previous research has predominantly focused on the direct effects of naturalness, and its interaction effects with other landscape attributes, such as culture, still require further in-depth exploration. Landscapes with high naturalness are often associated with relaxation, restoration, and positive emotions, possessing greater restorative potential, whereas low-naturalness landscapes, such as artificial environments, may induce feelings of tension or anxiety [54]. Urban green spaces with higher naturalness promote stronger place identity and enhance well-being [55]. Experiences in high-naturalness environments tend to produce greater restorative effects [56].

Perceived restorativeness, derived from Attention Restoration Theory (ART) [57], refers to an individual’s perception of the potential of a specific environment to offer psychological recovery, i.e., the environment’s capacity to help individuals recover from stress and mental fatigue and replenish their cognitive resources [58]. Existing research indicates that tourism activities have become a new means of releasing stress and facilitating self-recovery, with perceived restorativeness playing a significant role in the tourism experience [59]. For instance, experimental research by Fisher et al. (2021) found that high-naturalness landscapes significantly enhanced individuals’ levels of perceived restorativeness compared to artificial environments [54]. At the same perceived level of naturalness, different landscape types exert varying influences on restorativeness. Natural forest landscapes, artificial forest landscapes, and settlement landscapes demonstrated the most significant restorative effects among the studied natural, semi-natural, and artificial landscapes [47].

High-culture landscapes, rich in historical and cultural elements, can offer a more profound cultural experience. According to Attention Restoration Theory, historical elements, narrative qualities, and symbolic meanings within cultural landscapes, such as historic environments, can attract tourists’ involuntary attention, allowing them to break free from routine thought patterns [60]. Concurrently, immersion in a unique cultural atmosphere can also provide an experience of “being away”. When tourists perceive the cultural connotations of a landscape to be “compatible” with their personal interests and exploratory needs and can derive rich experiences from it (“extent”), their perceived restorativeness is enhanced. Built environments that promote cultural activities or integrate architectural and natural elements may possess restorative potential [58].

Landscape naturalness refers to the proportion and characteristics of natural elements within a landscape, such as vegetation cover, water body distribution, and topographical features, relative to the degree of human intervention [48]. Landscapes with high naturalness, characterized by a predominance of natural elements over human intervention, are often associated with relaxation, restoration, and positive emotions, possessing greater restorative potential [54]. While such environments may be perceived as aesthetically pleasing, this study focuses on the objective presence of natural features rather than subjective interpretations of beauty. High-naturalness landscapes are dominated by natural elements with minimal human alteration, while low-naturalness landscapes feature significant human modification. Naturalness is a critical factor influencing human environmental perception and psychological responses, often correlating with ecosystem attributes like biodiversity, though this study primarily focuses on visible natural features as perceived by tourists [50]. Cultural landscape types categorize landscapes shaped by human culture and history, varying in their cultural significance and symbolic depth. High-culture landscapes are rich in historical and cultural elements, offering profound experiences, whereas low-culture landscapes have minimal cultural significance or narrative depth. These distinctions inform the interaction between cultural and natural attributes in shaping tourists’ perceived restorativeness.

Landscape naturalness may moderate the indirect effect of cultural landscape type on destination recommendation intention through perceived restorativeness. High-naturalness environments inherently possess strong restorative potential [56]. When cultural landscapes are embedded in high-naturalness environments, natural elements can further enhance the “fascination”, “being away”, and “extent” of the cultural experience, thereby amplifying the perceived restorativeness of the cultural landscape. Existing research has found that high naturalness combined with high culture can enhance environmental restorativeness, thereby increasing destination loyalty [53]. When tourists experience a high level of perceived restorativeness at a destination, they are more likely to develop positive emotional connections, form a favorable impression of the place, and consequently be more willing to recommend it to others. Therefore, this study infers that cultural landscape type can enhance tourists’ perceived restorativeness, which in turn strengthens their destination recommendation intention. Furthermore, landscape naturalness, as a moderating variable, may influence the intensity of perceived restorativeness. We contend that when a landscape possesses both rich cultural connotations (high-culture attributes) and abundant natural elements (high naturalness), tourists can derive cultural attraction and natural healing from both dimensions, leading to a stronger perceived restorativeness effect and thus a higher likelihood of recommending it to others. For example, when visiting an ancient monastery surrounded by pristine forest, tourists are not only attracted by its historical and religious culture but can also find tranquility and relaxation from the surrounding natural scenery; this dual experience will significantly enhance their perceived restorativeness, making them more willing to recommend the site to others. Accordingly, the following hypothesis is proposed:

H2: 

Landscape naturalness moderates the mediating role of perceived restorativeness in the relationship between cultural landscape type and recommendation intention. Specifically, under conditions of high naturalness, the positive impact of cultural landscape type on perceived restorativeness is stronger; in landscapes with both high culture and high naturalness, tourists’ perceived restorativeness levels are highest, consequently leading to a significant increase in recommendation intention.

2.4. Application of EEG Technology in Tourism

Event-related potentials (ERPs) refer to changes in brain electrical potentials recorded from the scalp that are time-locked to specific internal or external events (such as the presentation of visual or auditory stimuli, or the execution of cognitive tasks), capable of revealing cognitive and emotional responses [61]. The primary advantage of ERPs lies in their excellent temporal resolution (millisecond level), which allows for the precise capture of the dynamic time course of cognitive processes, revealing different stages of information processing. Compared to behavioral methods that rely on subjective reports, ERPs can provide more direct and objective evidence of neural activity, helping to uncover implicit cognitive and emotional responses that are difficult to perceive through introspection or are susceptible to social desirability effects [62].

The introduction of EEG technology has provided tourism researchers with a window to “observe” tourists’ brain activity, thereby enabling a deeper understanding of tourism phenomena. In recent years, it has garnered increasing attention in tourism research, particularly in the domains of tourist emotion, cognition, and decision-making behavior. EEG technology, by recording scalp electrical activity, can monitor tourists’ neural responses to tourism stimuli in real time, offering objective data that transcend subjective reports for understanding the tourist experience [15]. In the hotel and tourism sectors, researchers utilize EEG/ERP techniques to assess the attractiveness and emotional arousal of tourism destination advertisements (e.g., images, videos) [63]. For instance, EEG studies have demonstrated that tourism advertisements can significantly enhance emotion-related brain activity (e.g., P2 and LPP components), validating ERPs as a robust tool for gauging emotional impact in this domain [14]. This neuroscientific approach extends to various media, confirming that destination marketing content embedded in films can also effectively influence viewers’ emotional responses [64]. Further nuances have been uncovered, such as how children’s positive emotional responses to high-valence tourism advertisements can vary during viewing and be moderated by factors like gender and age [65]. Collectively, these findings offer valuable tools for optimizing advertisement design and enhancing the precision of marketing communication.

Beyond direct advertising, EEG technology also sheds light on tourists’ cognitive and emotional states in relation to specific tourism elements and environments. Research has shown, for example, that the type of tourist attraction can significantly influence consumers’ price perception and subsequent purchasing decisions, offering a neuroscientific basis for developing differentiated pricing strategies [15]. Furthermore, EEG has been used to explore the impact of the physical environment, revealing that decreased landscape naturalness correlates with reduced neural markers of restoration (e.g., alpha and beta brainwave activity), thereby diminishing perceived restorativeness [47].

In the study of landscape preferences, Ding et al. (2022) confirmed that electroencephalography (EEG) can effectively make up for the limitations of eye movement tracking and reveal the differences in neural activity that cannot be reflected by eye movement data when different social groups observe plants [66]. Using immersive virtual reality with EEG and heart rate variability measurements, Shi et al. (2025) demonstrated that viewing coastal landscapes significantly aids stress recovery, with coastal trails showing the most potent restorative effects [67]. Wang et al. (2021) found through electroencephalography (EEG) technology and questionnaire research that seasonal landscapes have a significant impact on stress relief. Summer landscapes are superior to winter ones in terms of psychological and physiological recovery effects, and EEG can be used as a new objective tool for landscape assessment and planning [68]. By applying a machine learning classifier to EEG signals, landscape types can be recognized with high accuracy and the distinct neural contributions of color and structure in perception can be successfully isolated [69]. By modeling the data of the electroencephalography (EEG) responses of landscapes, Ren et al. (2024) revealed the advantages of free-form landscapes in inducing physiological comfort and quantified the linear association between specific landscape features and brain alpha wave activity [70].

In ERP research, P2 and the Late Positive Potential (LPP) are two important components closely related to cognitive and emotional processing. P2 is a positive wave appearing approximately 150–250 milliseconds after stimulus presentation, typically associated with early attention allocation and stimulus categorization, reflecting the individual’s initial processing of stimulus salience [61]. P2 is considered to reflect early, pre-attentive, or automated attention allocation processes, related to the rapid assessment of stimulus physical characteristics (such as intensity, novelty) [14]. In emotion research, the amplitude of P2 can be modulated by emotional stimuli; typically, stimuli with higher emotional valence (pleasant or unpleasant) and arousal induce a larger P2 amplitude, suggesting that early attentional resources are preferentially allocated to salient stimuli of biological or sociological significance [71]. The LPP is a sustained positive wave appearing 400–800 milliseconds after stimulus presentation, associated with emotional evaluation and motivational intensity, and is typically most prominent in centro-posterior scalp regions [72]. The LPP is widely regarded as a reliable neural indicator of sustained attentional processing and elaborative processing of emotional stimuli. Its amplitude is closely related to the emotional arousal of the stimulus; stimuli with high arousal, whether positive (e.g., exciting adventure images) or negative (e.g., unpleasant scenes), typically elicit a larger LPP amplitude compared to neutral stimuli [73]. The combined analysis of these two components offers an important perspective for understanding the cognitive–emotional interactions of tourists in different tourism contexts.

Although the application of EEG technology in tourism research is still in its developmental stage, the potential it demonstrates is substantial. It can not only provide physiological corroboration for subjective reports but also reveal underlying preferences and responses that tourists themselves may not be aware of, thereby offering a more scientific basis for tourism product design, service enhancement, and destination management.

3. Research Design

This study focuses on Quanzhou, a historic city located in Fujian Province on the southeastern coast of China, as the primary context for examining the influence of cultural landscape types and landscape naturalness on tourists’ recommendation intentions. The city is renowned for its diverse cultural landscapes, which include ancient temples, historic port structures, traditional residential architecture, and religious sites representing a blend of Chinese, Islamic, and other cultural influences dating back over a millennium. These cultural landscapes were instrumental in Quanzhou’s designation as a UNESCO World Heritage Site in 2021 under the title “Quanzhou: Emporium of the World in Song-Yuan China”, recognizing its pivotal role in maritime trade and cultural exchange during the Song and Yuan dynasties [74]. The city’s unique combination of high-culture landscapes, such as the Kaiyuan Temple and the Luoyang Bridge, alongside areas of varying naturalness, from coastal and mountainous scenery to urbanized zones, makes it an ideal setting to explore the interplay between cultural and natural attributes in shaping tourist experiences. Furthermore, Quanzhou attracts millions of visitors annually. Quanzhou has a rich cultural history, integrating various elements such as religious culture, the culture of the Maritime Silk Road, and Minnan culture [74]. Selecting Quanzhou for this study thus provides a rich and representative case to investigate how cultural heritage and natural environments influence place attachment, perceived restorativeness, and recommendation intentions among tourists.

To systematically investigate the influence mechanism of different cultural landscape types on tourists’ recommendation intention and reveal the moderating role of landscape naturalness, this study employs a multi-method, phased research design integrating traditional self-report measures with cutting-edge neurophysiological techniques. This design responds to the dual structural theory of attitudes proposed by the APE model [16], simultaneously examining tourists’ “propositional attitudes” (formed through conscious cognitive processing) and “associative attitudes” (immediate, spontaneous, and subconscious reactions toward tourism attractions). Study 1 adopts a scenario-based experiment to capture propositional attitudes, systematically examining the impact pathways of cultural landscape type (high culture vs. low culture) and landscape naturalness (high naturalness vs. low naturalness) on place attachment, perceived restorativeness, and recommendation intention, while exploring the moderating role of landscape naturalness. Aligned with the APE model, Study 1 focuses on propositional attitudes, explicit evaluations reflected through conscious logical reasoning and verbal expression. Study 2 utilizes event-related potential (ERP) technology to objectively measure tourists’ brainwave activity when viewing images combining different cultural landscape types and naturalness levels. This method focuses on revealing associative attitudes, which represent rapid, typically unconscious affective associations and cognitive processing automatically activated by stimuli in early stages. With millisecond-level temporal resolution, ERP technology tracks real-time neuroelectrophysiological responses to specific stimuli, providing objective indicators for understanding early attentional allocation (e.g., P2 component) and subsequent sustained attention with affective arousal (e.g., Late Positive Potential, LPP) during initial exposure to attractions. This approach addresses limitations of traditional self-reports in capturing immediate, implicit reactions. Integrating results from Study 1 (scenario experiment) and Study 2 (ERP experiment), this study comprehensively elucidates the effects of cultural landscape type and naturalness on tourist experience and behavioral intentions across propositional and associative attitude dimensions. This mixed-methods approach is theoretically grounded in the methodological triangulation framework, which advocates combining multiple research methods to enhance validity and provide a more comprehensive understanding of complex phenomena [75]. Previous studies in tourism research have successfully demonstrated the value of integrating neurophysiological measures with traditional behavioral methods to capture both explicit and implicit tourist responses [15,47,68]. The framework is illustrated in Figure 1. Finally, Study 3 builds upon Study 2 to validate the robustness and applied potential of the identified neural patterns. By constructing machine learning models to classify landscape naturalness using the EEG data from Study 2, we not only computationally confirm that the neural responses elicited by different landscapes possess distinct and unique patterns, but also provide a methodological exploration for the future development of brain–computer interface (BCI)-based automated landscape assessment or personalized tourism recommendation systems. The integration of these three studies is justified as they collectively contribute to a cohesive conceptual and methodological framework, where Studies 1 and 2 explore the psychological and neurophysiological mechanisms of tourist behavior under the APE model, while Study 3 extends this understanding into practical applications by leveraging EEG data for landscape classification and technological innovation in tourism research.

4. Study 1

4.1. Participants

A 2 (cultural landscape type: high culture vs. low culture) × 2 (landscape naturalness: high naturalness vs. low naturalness) between-subjects design was employed. Participants (N = 300) were recruited via Credamo, an online survey platform, with monetary compensation. Attention-check questions ensured data quality; 11 participants were excluded due to failed checks or incomplete responses, yielding 289 participants (96.3% validity). The sample comprised 172 males (59.5%) and 117 females (40.5%), predominantly aged 21–30 (n = 190, 65.7%). Most held bachelor’s degrees (n = 209, 72.3%).

4.2. Experimental Procedure and Materials

This study employs a 2 (cultural landscape type: high culture vs. low culture) × 2 (landscape naturalness: high naturalness vs. low naturalness) between-subjects experimental design to investigate the effects of different landscape characteristics on tourists’ place attachment, perceived restorativeness, and recommendation intention. Participants were randomly assigned to one of four experimental conditions: high culture + high naturalness group, high culture + low naturalness group, low culture + high naturalness group, and low culture + low naturalness group (see Figure 2).

The experimental procedure is as follows: Participants accessed the study via the Credamo online survey platform. Before starting the questionnaire, the system randomly assigned each participant to an experimental condition. Each participant viewed landscape images and detailed textual descriptions corresponding to their assigned condition:

High culture–high naturalness condition: Example: “Luojia Temple”. The image depicts a seaside temple structure (see Figure 2A). The text emphasizes “red walls and yellow tiles appearing particularly solemn and serene against the azure sea and sky” and “listening to the melodious Buddhist chants accompanied by the sound of ocean waves, experiencing profound Buddhist cultural heritage while taking in the magnificent coastal scenery, thereby embodying the harmonious integration of human faith and marine nature”. This aims to highlight the combination of deep cultural connotations and a beautiful natural environment.

High culture–low naturalness condition: Example: “Guandi Temple”. The image shows an intricately carved temple with painted eaves (see Figure 2B). The text emphasizes its status as “one of Quanzhou’s thriving folk belief centers”, carrying “rich historical narratives and devout public faith”, and notes that it “stands quietly amidst urban streets, surrounded by ordinary city life scenes”. This aims to highlight a strong cultural atmosphere with relatively weaker natural elements.

Low culture–high naturalness condition: Example: “Binhai Park Beach”. The image displays an open sandy beach and seascape (see Figure 2C). The text emphasizes “golden sands stretching into the distance, waves gently lapping the shore, sea breeze bringing refreshing air”, its role as “an excellent spot for leisure walks, ocean access, and sunshine enjoyment”, and clarifies that “this location primarily showcases pure coastal natural scenery with minimal artificial historical or cultural imprints”. This aims to emphasize the purity of natural scenery with minimal cultural elements.

Low culture–low naturalness condition: Example: “Modern architecture of Taihe Plaza”. The image shows a modern commercial complex (see Figure 2D). The text emphasizes its “novel and fashionable architectural design, glass curtain walls gleaming in sunlight”, “interior housing numerous modern brand stores and entertainment facilities”, and notes that “the plaza is surrounded by orderly planned urban roads and newly constructed buildings, presenting a vibrant and convenient modern cityscape”. This aims to depict a modern urban landscape lacking significant cultural heritage or natural scenery.

Before viewing the landscape images and descriptions, participants read priming instructions designed to enhance situational immersion and ensure an authentic experience of the landscape scenario from a tourist perspective. After sufficient immersion in the assigned experimental scenario, participants completed the measurement questionnaire. All items use a 7-point scale (1 = Strongly Disagree, 7 = Strongly Agree) to ensure measurement precision and discrimination. All scales are based on established instruments from the existing literature, adapted appropriately to this study’s context to ensure reliability and validity. First, participants answered questions assessing their perception of cultural landscape type and naturalness (e.g., “This attraction belongs to: Low culture vs. High culture” and “This attraction belongs to: Low naturalness vs. High naturalness”) to verify the effectiveness of the experimental manipulations. Subsequently, place attachment towards Quanzhou as a World Heritage city was measured after participants assumed the tourist role (e.g., “After visiting this attraction as a ‘tourist’, Quanzhou, this World Heritage City, is very meaningful to me”—5 items total). The place attachment items integrate research by Williams et al. (1992), which posits that place attachment comprises place dependence and place identity [76]. Next, perceived restorativeness was measured after assuming the tourist role (e.g., “After visiting this attraction as a ‘tourist’, I believe this attraction helps me forget daily responsibilities and feel relaxed” and “After visiting this attraction as a ‘tourist’, I believe this is a fascinating place that keeps my curiosity alive and prevents boredom”—4 items total). The items are adapted from Negrín et al. (2017) [77]. Finally, recommendation intention for Quanzhou as a tourism destination was measured (e.g., “You are willing to recommend Quanzhou, this World Heritage City, to your relatives and friends” and “When someone asks you for tourism city suggestions, you will recommend Quanzhou, this World Heritage City, to them”—3 items total),the items are adapted from Stanovčić et al. (2021) [78].

After completing the main questionnaire, participants’ demographic information (gender, age, education level, etc.) was collected. The entire experimental process strictly controlled the landscape images and textual descriptions presented to participants, ensuring each participant was only exposed to their randomly assigned experimental condition. Through these procedures, this study aims to investigate the effects of cultural landscape type and landscape naturalness on recommendation intention, and further test the mediating roles of place attachment and perceived restorativeness, as well as the moderating role of landscape naturalness.

4.3. Experimental Results

The results of independent samples t-tests showed that the mean score for the low culture group was 4.4366 (SD = 2.00165), while the high culture group scored 6.1224 (SD = 0.77546). The difference between low culture and high culture scores was significant, t(287) = −9.500, p < 0.001. For naturalness, the low naturalness group mean was 2.8741 (SD = 1.71516), and the high naturalness group mean was 5.9521 (SD = 0.95652). The difference between low naturalness and high naturalness scores was significant, t(287) = −18.891, p < 0.001. Therefore, the experimental manipulations were successful.

A 2 × 2 ANOVA on place attachment revealed a significant main effect of landscape naturalness (F(1, 285) = 4.444, p = 0.036, ηp² = 0.015) and a significant main effect of cultural landscape type (F(1, 285) = 30.926, p < 0.001, ηp² = 0.098). The interaction effect between landscape naturalness and cultural landscape type was not significant (F(1, 285) = 2.370, p = 0.125, ηp² = 0.008). Post hoc comparisons indicated that place attachment was significantly higher for landscapes with high naturalness (M = 5.667, SE = 0.071) compared to those with low naturalness (M = 5.453, SE = 0.072, p = 0.036). Place attachment was also significantly higher for high-culture landscapes (M = 5.842, SE = 0.071) than for low-culture landscapes (M = 5.278, SE = 0.072, p < 0.001). For perceived restorativeness, a 2 × 2 ANOVA showed a significant main effect of landscape naturalness (F(1, 285) = 20.575, p < 0.001, ηp² = 0.067), but no significant main effect of cultural landscape type (F(1, 285) = 0.912, p = 0.340, ηp² = 0.003). The interaction effect between landscape naturalness and cultural landscape type was significant (F(1, 285) = 7.680, p = 0.006, ηp² = 0.026). Post hoc comparisons revealed significantly higher perceived restorativeness for high-naturalness landscapes (M = 5.887, SE = 0.072) compared to low-naturalness landscapes (M = 5.425, SE = 0.072, p < 0.001). Simple effect analysis showed that under low culture conditions, landscape naturalness had a significant effect on perceived restorativeness: high naturalness (M = 5.979, SE = 0.101) was significantly higher than low naturalness (M = 5.236, SE = 0.104, p < 0.001). Under high culture conditions, landscape naturalness had no significant effect on perceived restorativeness (p = 0.209). Under low naturalness conditions, cultural landscape type had a significant effect: high culture (M = 5.615, SE = 0.101) showed significantly higher perceived restorativeness than low culture (M = 5.236, SE = 0.104, p = 0.009). Under high naturalness conditions, cultural landscape type had no significant effect (p = 0.143). A 2 × 2 ANOVA on recommendation intention indicated significant main effects for both landscape naturalness (F(1, 285) = 3.998, p = 0.046, ηp² = 0.014) and cultural landscape type (F(1, 285) = 8.260, p = 0.004, ηp² = 0.028). The interaction effect was not significant (F(1, 285) = 1.463, p = 0.227, ηp² = 0.005). Post hoc comparisons showed significantly higher recommendation intention for high-naturalness landscapes (M = 5.833, SE = 0.075) versus low-naturalness landscapes (M = 5.619, SE = 0.076, p = 0.046), and for high-culture landscapes (M = 5.880, SE = 0.075) versus low-culture landscapes (M = 5.573, SE = 0.076, p = 0.004).

Using the PROCESS macro in SPSS (Model 4) [79] with 289 samples and 5000 bootstrap resamples (95% confidence interval), this study tested the effect of cultural landscape type (low culture = 1, high culture = 2) on recommendation intention and the mediating role of place attachment. The key findings were as follows: cultural landscape type had a significant positive effect on place attachment (β = 0.5591, p < 0.001, 95% CI = [0.3577, 0.7605]), indicating that high-culture landscapes significantly enhance place attachment. The direct effect of cultural landscape type on recommendation intention was negative and significant (β = −0.1623, p = 0.0201, 95% CI = [−0.2988, −0.0257]), suggesting that high-culture landscapes may directly reduce recommendation intention. Place attachment had a significant positive effect on recommendation intention (β = 0.8310, p < 0.001, 95% CI = [0.7560, 0.9060]), indicating that enhanced place attachment significantly increases recommendation intention. The indirect effect of cultural landscape type on recommendation intention via place attachment was significant and positive (indirect effect = 0.4646, BootSE = 0.0951, 95% BootCI = [0.2864, 0.6646]). The total effect of cultural landscape type on recommendation intention was positive and significant (total effect = 0.3024, p = 0.0053, 95% CI = [0.0906, 0.5141]). In summary, high-culture landscapes indirectly promote recommendation intention by enhancing place attachment. Despite the negative direct effect, the total effect remains positive, demonstrating the crucial mediating role of place attachment (see

Table 1. The relationship between cultural landscape type, place attachment, and recommendation intention.

Path	β	p	Boot 95% CI
Cultural Landscape Type → Place Attachment	0.5591	<0.001	[0.3577, 0.7605]
Place Attachment → Recommendation Intention	0.831	<0.001	[0.7560, 0.9060]
Cultural Landscape Type → Recommendation Intention (Total Effect)	0.3024	0.0053	[0.0906, 0.5141]
Cultural Landscape Type → Recommendation Intention (Direct Effect)	−0.1623	0.0201	[−0.2988, −0.0257]
Cultural Landscape Type → Place Attachment → Recommendation Intention (Indirect Effect)	0.4646	-	[0.2864, 0.6646]

).

In this study, using the PROCESS tool in SPSS 27.0.1 software and the bootstrap method, the impact of cultural landscape type (low culture = 1, high culture = 2) on recommendation intention, along with the mediating role of perceived restorativeness and the moderating role of landscape naturalness, was examined. Analysis of Model 7 with 289 samples and 5000 bootstrap resamples (95% confidence interval) revealed the following findings: Cultural landscape type had a significant positive effect on perceived restorativeness (β = 0.9436, p = 0.0038, 95% CI = [0.3079, 1.5794]). This indicates that high-culture landscapes significantly enhance users’ perceived restorativeness. Landscape naturalness also had a significant positive effect on perceived restorativeness (β = 1.3082, p = 0.0001, 95% CI = [0.6712, 1.9453]). Furthermore, the interaction effect between cultural landscape type and landscape naturalness on perceived restorativeness was significantly negative (β = −0.5643, p = 0.0059, 95% CI = [−0.9651, −0.1635]), indicating that landscape naturalness exerts a negative moderating role on the relationship between cultural landscape type and perceived restorativeness. The direct effect of cultural landscape type on recommendation intention was significantly positive (β = 0.2415, p = 0.0020, 95% CI = [0.0894, 0.3936]). Perceived restorativeness had a significant positive effect on recommendation intention (β = 0.7067, p < 0.0001, 95% CI = [0.6223, 0.7911]), indicating that enhanced perceived restorativeness significantly increases users’ recommendation intention. Further analysis of the conditional indirect effects revealed that the indirect effect of cultural landscape type on recommendation intention through perceived restorativeness was moderated by landscape naturalness: At low naturalness, the indirect effect was positive and significant (indirect effect = 0.2681, BootSE = 0.1294, 95% BootCI = [0.0292, 0.5328]). At high naturalness, the indirect effect was negative and non-significant (indirect effect = −0.1307, BootSE = 0.0798, 95% BootCI = [−0.2891, 0.0298]). The index of moderated mediation indicates a significant difference (index = −0.3988, BootSE = 0.1557, 95% BootCI = [−0.7175, −0.1089]). In summary, this study demonstrates that high-culture landscapes can indirectly promote users’ recommendation intention through perceived restorativeness (see

Table 2. The relationship between cultural landscape type, landscape naturalness, perceived restorativeness, and recommendation intention.

Path	β	p	Bootstrap 95% CI
Cultural Landscape Type → Perceived Restorativeness (Main Effect)	0.9436	0.0038	[0.3079, 1.5794]
Landscape Naturalness → Perceived Restorativeness	1.3082	0.0001	[0.6712, 1.9453]
Cultural Landscape Type × Landscape Naturalness → Perceived Restorativeness (Interaction)	−0.5643	0.0059	[−0.9651, −0.1635]
Cultural Landscape Type → Perceived Restorativeness (Low Naturalness)	0.3794	0.0093	[0.0944, 0.6643]
Cultural Landscape Type → Perceived Restorativeness (High Naturalness)	−0.1849	0.1976	[−0.4668, 0.0969]
Cultural Landscape Type → Recommendation Intention (Direct Effect)	0.2415	0.0020	[0.0894, 0.3936]
Perceived Restorativeness → Recommendation Intention	0.7067	<0.0001	[0.6223, 0.7911]
Indirect Effect (Low Naturalness)	0.2681	-	[0.0292, 0.5328]
Indirect Effect (High Naturalness)	−0.1307	-	[−0.2891, 0.0298]
Index of Moderated Mediation	−0.3988	-	[−0.7175, −0.1089]

). However, this indirect effect is moderated by landscape naturalness, with landscapes characterized by low naturalness more strongly enhancing this positive effect.

5. Study 2

5.1. Participants

This study recruited 28 healthy adults (14 males, 14 females), comprising local residents of Quanzhou or students studying in Quanzhou as EEG experiment participants. Age ranged from 20 to 28 years (M = 23.46, SD = 2.15). All participants were right-handed, had normal or corrected-to-normal vision, reported no history of neurological or psychiatric disorders, and had not taken any medication affecting the central nervous system. Participants included local residents and undergraduate and graduate students from diverse academic backgrounds. None had prior experience with EEG experiments and were unaware of the experimental purpose. To ensure data quality, all participants were required to maintain adequate sleep 24 h before the experiment and abstain from alcohol, caffeine, and similar substances. Written informed consent was obtained from each participant, and compensation of CNY 70 was provided. The study received approval from the local ethics committee, with all procedures complying with the Helsinki Declaration.

5.2. Experimental Stimuli

Professionally evaluated landscape images served as experimental stimuli. A stimulus library of 40 high-definition landscape images was constructed, covering four conditions. All images were sourced from official tourism websites, professional photography databases, and promotional materials, depicting authentic landscapes. To ensure validity and representativeness, five tourism experts evaluated and approved the images. All images were standardized for brightness, contrast, and size (1000 × 1000 pixels) to control confounding variables.

5.3. Experimental Procedure

The experimental process of this study strictly followed the mature research paradigms of emotion and cognitive neuroscience, especially the passive viewing and evaluation task widely adopted in ERP research [15,80,81], to ensure the scientific nature of the program, the reliability of the data, and the validity of the conclusion. The stimulation process of the experiment was presented using E-Prime 2.0 Software (Psychology Software tools, Pittsburgh, PA, USA). The entire experimental process consists of 160 tests, corresponding to 40 tests for each given condition. The experimental participants sat comfortably in a dimly lit and sound-reduced electrically shielded room, 100 cm away from the computer screen. Each participant had five practice sessions before the formal experiment to familiarize themselves with the task process. The formal experiment was divided into four groups, with 40 pictures presented in each group. There was a short break between the groups to reduce the fatigue effect. The specific process of each trial is shown in Figure 3. Firstly, a “+” fixation point was presented in the center of the screen for 500 ms to guide the participants to concentrate. Then, a blank screen with a random duration of 300–500 ms was presented. Next, a landscape picture was presented in the center of the screen for 3000 ms. At the same time, a question was presented below the picture: “Based on the picture of this scenic area, how likely is it to recommend this place as a tourist destination to others?” Participants had to rate their likeliness on a scale of 1 to 7 (1 = highly not recommended, 7 = highly recommended). After scoring, a blank screen with a random duration of 300–500 ms was displayed, and then participants proceeded to the next attempt (See Figure 3). To ensure data quality, the system automatically recorded electroencephalography (EEG) marks when the image stimulation was presented for subsequent EEG data analysis. The presentation order of the pictures was balanced among the subjects to eliminate the influence of sequence effect. Before the experiment began, all participants signed written informed consent forms and received detailed guidance to ensure a full understanding of the requirements of the experimental tasks. The entire experimental process (including the wearing of electrode caps, practice and formal experiments) took approximately 90 min.

5.4. Data Acquisition and Analysis

EEG data acquisition was performed using the Neuroscan system (Curry 8, Neuroscan Inc., Charlotte, NC, USA). A 64-channel electrode cap was employed, with electrodes arranged according to the international 10–20 system. Signals were amplified using SynAmps2 amplifiers. Key acquisition parameters were set as follows: sampling rate at 1000 Hz, and electrode impedance maintained below 5 kΩ to ensure high-quality EEG signals. During recording, FCz served as the online reference electrode, with bilateral mastoid signals (M1, M2) recorded simultaneously for subsequent offline re-referencing. Stimulus markers identifying the four experimental conditions (high culture–high naturalness, high culture–low naturalness, low culture–high naturalness, low culture–low naturalness) were automatically recorded via a trigger interface between E-Prime and Neuroscan upon landscape image presentation. Two trained researchers monitored data acquisition quality throughout the experiment to ensure stable electrode contact and signal integrity.

Data preprocessing was conducted using the EEGLAB toolbox [82] within the MATLAB R2022a environment. The preprocessing pipeline comprised the following steps: importing raw continuous EEG data into EEGLAB; re-referencing to the average of the bilateral mastoids (M1, M2) to enhance the signal-to-noise ratio; downsampling data from 1000 Hz to 500 Hz to reduce computational load while preserving essential information; applying a third-order Butterworth band-pass filter (0.1–40 Hz) to remove high-frequency noise and low-frequency drifts, supplemented by a 48–52 Hz notch filter to eliminate line noise interference; epoching continuous data relative to event markers, extracting segments from 200 ms pre-stimulus to 1000 ms post-stimulus onset; performing baseline correction using the average amplitude from the −200 ms to 0 ms pre-stimulus interval; applying Independent Component Analysis (ICA) for artifact separation, utilizing the ICLabel plugin [83] for automated identification and removal of non-neural components (e.g., ocular, muscular artifacts). All preprocessing steps and parameter selections adhered to established best practices in the field, ensuring data quality and analytical reliability. The preprocessed data were used for subsequent statistical analyses examining the effects of landscape types on neural responses. Based on prior research, this study focused on two ERP components associated with landscape perception and evaluation: P2 and the LPP. For P2 analysis (reflecting early attentional processing), data from three central–parietal electrodes (CZ, C1, C2) were extracted to investigate the effects of cultural landscape type and naturalness. For LPP analysis (reflecting sustained affective processing and preference formation), data from nine electrodes across central, parietal, and occipital regions (C1, Cz, C2, CP1, CPz, CP2, P1, Pz, P2) were analyzed [15].

5.5. ERPs Results

EEG waveform plots and topographical maps were generated using data from the central electrode Cz and parietal electrode Pz within the −200 to 1000 ms time window (see Figure 4). Based on the waveform plots, the P2 component time window was defined as 190–220 ms. The average amplitude of the P2 component was calculated across three electrodes (CZ, C1, C2), followed by a 2 × 2 repeated-measures analysis of variance (ANOVA) on the mean P2 amplitude.

The 2 × 2 ANOVA on P2 mean amplitude revealed a significant main effect of cultural landscape type (F(1, 26) = 11.197, p = 0.003, ηp² = 0.301), while the main effect of landscape naturalness was not significant (F(1, 26) = 0.509, p = 0.482, ηp² = 0.019). The interaction effect between cultural landscape type and landscape naturalness was significant (F(1, 26) = 7.697, p = 0.010, ηp² = 0.228). Post hoc comparisons indicated that high-culture landscapes elicited significantly lower P2 mean amplitudes (M = 0.423, SE = 0.938) than low-culture landscapes (M = 1.740, SE = 0.874, p = 0.003). Simple effect analysis showed that for high-culture landscapes, landscape naturalness had no significant effect on P2 mean amplitude (p = 0.230). For low-culture landscapes, low naturalness elicited marginally higher P2 mean amplitudes (M = 2.314, SE = 0.938) than high naturalness (M = 1.165, SE = 0.898, p = 0.053). Additionally, under low naturalness conditions, low-culture landscapes elicited significantly higher P2 mean amplitudes (M = 2.314, SE = 0.938) than high-culture landscapes (M = 0.143, SE = 0.937, p < 0.001).

Based on the waveform plots, the LPP component time window was defined as 400–700 ms. The average amplitude of the LPP component was calculated across nine electrodes (PZ, CPZ, CZ, P1, P2, C1, C2, CP1, CP2), followed by a 2 × 2 repeated-measures ANOVA on the mean LPP amplitude. The main effect of cultural landscape type was not significant (F(1, 26) = 1.518, p = 0.229, ηp² = 0.055), while the main effect of landscape naturalness was significant (F(1, 26) = 5.603, p = 0.026, ηp² = 0.177). The interaction effect between cultural landscape type and landscape naturalness was significant (F(1, 26) = 10.826, p = 0.003, ηp² = 0.294). Post hoc comparisons revealed that low-naturalness landscapes elicited significantly higher LPP mean amplitudes (M = 4.788, SE = 0.848) than high-naturalness landscapes (M = 3.611, SE = 0.855, p = 0.026). Simple effect analysis indicated that for high-culture landscapes, landscape naturalness had no significant effect on the LPP mean amplitude (p = 0.394). For low-culture landscapes, low naturalness elicited significantly higher LPP mean amplitudes (M = 5.983, SE = 1.139) than high naturalness (M = 3.086, SE = 0.962, p < 0.001). Furthermore, under low naturalness conditions, low-culture landscapes elicited significantly higher LPP mean amplitudes (M = 5.983, SE = 1.139) than high-culture landscapes (M = 3.593, SE = 0.757, p = 0.016).

6. Study 3

6.1. Experimental Objectives

This study focuses on utilizing machine learning methods to classify individuals’ cognitive responses when viewing high-naturalness versus low-naturalness landscapes based on P2 and LPP features in EEG signals. EEG-based machine learning has been widely applied in diverse fields, such as predicting consumer purchase decisions in online shopping environments [84], hazardous driving detection [85], driving fatigue monitoring [86], trust assessment [87], safety perception [88], and food umami evaluation [89]. This study systematically evaluates the predictive performance of three machine learning models, namely Decision Tree (DT), Support Vector Machine (SVM), and Gradient Boosting Decision Tree (GBDT), in this binary classification task. The core objectives are as follows: (1) to leverage machine learning methods for classifying viewing experiences of high-naturalness versus low-naturalness landscapes based on EEG P2 and LPP features; (2) to validate the differential patterns of P2 and LPP observed in Study 2 between high and low-naturalness landscapes from a data-driven perspective; and (3) to systematically assess the predictive capability and applicability of three representative machine learning models (DT, SVM, GBDT) in EEG signal classification, providing a basis for selecting appropriate analytical tools in subsequent research. The experimental procedure comprises two main phases, hyperparameter tuning and performance evaluation, ensuring methodological rigor and result reproducibility.

6.2. Data Preprocessing and Partitioning

The dataset consists of two classes: high-naturalness landscapes and low-naturalness landscapes. Features were extracted from EEG signals within the P2 (190–220 ms) and LPP (400–700 ms) time windows. A total of 18 EEG features were derived from nine electrodes: PZ, CPZ, CZ, P1, P2, C1, C2, CP1, and CP2. The dataset was partitioned into training (80%) and testing (20%) sets at an 8:2 ratio, with a fixed random seed (random_state = 42) to ensure reproducibility.

6.3. Hyperparameter Tuning and Performance Evaluation

Decision Tree (DT): Key hyperparameters were optimized via GridSearchCV with 10-fold cross-validation. Tuning occurred in two rounds: Round 1 optimized max_depth (range: 1–10) and min_samples_split (range: 1–10); Round 2 optimized min_samples_leaf (range: 2–30) based on Round 1’s optimal parameters. The final optimal parameters are max_depth = 6, min_samples_split = 8, and min_samples_leaf = 17.

Support Vector Machine (SVM): Grid search optimized kernel type (‘linear’/’rbf’), regularization parameter C (range: 1–10), and gamma (range: 1e–5 to 1). The final optimal parameters are C = 1, kernel = ‘rbf’, and gamma = 1e–5.

Gradient Boosting Decision Tree (GBDT): Hyperparameter tuning occurred in three rounds: Round 1 optimized learning_rate (range: 1e–5 to 1) and n_estimators (range: 2–100), yielding learning_rate = 0.4445 and n_estimators = 16; Round 2 optimized max_depth (range: 1–15), yielding max_depth = 1; Round 3 optimized min_samples_split (range: 2–20) and min_samples_leaf (range: 2–20), yielding min_samples_split = 4 and min_samples_leaf = 3.

After obtaining optimal parameters for each model, they were retrained on the full training set to maximize data utilization. To comprehensively evaluate model stability and generalization capability, the dataset partitioning (8:2 ratio) and model training processes were repeated 1000 times. In each iteration, models were trained using optimal parameters on the training set, and classification accuracy was computed on the independent test set. Test accuracy from each iteration was recorded, and the mean, standard deviation, and maximum accuracy were calculated to quantify model performance.

6.4. Results

The Decision Tree model achieved the highest mean classification accuracy of 0.6529 (standard deviation = 0.1310), significantly higher than the chance level (50%). Its maximum test accuracy reached 1.0, indicating perfect discrimination between high-naturalness and low-naturalness landscapes under certain data partitions. The SVM model exhibited the lowest mean classification accuracy at only 0.3876 (standard deviation = 0.0779), falling below chance level, with a maximum test accuracy of merely 0.4545, demonstrating insufficient classification capability for the current task. The GBDT model attained a mean classification accuracy of 0.5596 (standard deviation = 0.1322), performing intermediately between the Decision Tree and SVM. Its maximum test accuracy also reached 1.0, indicating high discriminative capability under specific data partitions. The performance ranking of the three models was Decision Tree (0.6529) > GBDT (0.5596) > SVM (0.3876). These results demonstrate that the Decision Tree model achieved the highest mean classification accuracy, followed by the GBDT model, while the SVM model performed worst, with its mean accuracy even below the chance level (0.5 for binary classification tasks). Both the DT and GBDT model achieved maximum accuracy of 1.0 under certain data partitions, but their relatively large standard deviations indicate sensitivity to data partitioning. Although the SVM model had the smallest standard deviation, this reflects its consistently poor performance across different data partitions. The ROC curves and confusion matrices for the three models are shown in Figure 5.

7. Discussion

This study aims to systematically investigate how different cultural landscape types influence tourists’ recommendation intention through two pathways, place attachment and perceived restorativeness, based on Place Attachment Theory, ART, and the APE model, while examining the moderating role of landscape naturalness. Study 1 employed a scenario-based experiment, designing tourism scenarios combining different cultural landscape types and landscape naturalness levels to collect self-reported data from tourists. This validated the impact pathways of cultural landscape type on place attachment and perceived restorativeness, their effects on recommendation intention, and the moderating effect of landscape naturalness. Study 2 utilized EEG technology, focusing on tourists’ neurophysiological responses to different landscape combinations, analyzing differences in early attentional allocation (P2 component) and affective arousal (LPP component). Adopting a 2 (cultural landscape type: high culture, low culture) × 2 (landscape naturalness: high naturalness, low naturalness) within-subjects experimental design, Study 2 recorded and analyzed tourists’ EEG activity while viewing different landscape images. Key analyses focused on the P2 component associated with early attentional processing [14] and the LPP component linked to subsequent processing depth and emotional arousal [90]. Repeated-measures ANOVA examined the main effects of cultural landscape type and landscape naturalness and their interaction effect on these ERP components. Study 3 systematically evaluated the predictive capability and applicability of three representative machine learning models (DT, SVM, and GBDT) in the classification task of EEG signals induced by landscape naturalness.

The main findings of this study provide robust support for the core mediating role of place attachment in the process through which cultural landscapes influence recommendation intention. Specifically, the results confirm H1: high-culture landscapes significantly enhance tourists’ place attachment levels. This finding aligns with Place Attachment Theory, which posits that the meaning and emotional bonds formed through individual interactions with specific environments are key to place attachment [36]. Cultural landscapes, as symbolic spaces embodying history, art, and lifestyle, are more likely to evoke emotional resonance and identification among tourists, thereby fostering stronger place attachment [43]. Furthermore, place attachment had a significant positive effect on recommendation intention, consistent with prior research showing that place attachment effectively predicts word-of-mouth communication [40] and tourist recommendation intention [41]. This result indicates that place attachment plays a pivotal role in how cultural landscapes influence tourist behavioral intentions. Additionally, high-naturalness landscapes also positively influence place attachment and recommendation intention, further validating the importance of natural environments in fostering emotional bonds [50]. Integrating these points, H1 is confirmed: place attachment plays a key mediating role in the impact of high-culture landscapes on recommendation intention, highlighting the importance of cultivating tourists’ emotional connections to enhance destination appeal. An unexpected finding from the mediation path analysis revealed a significant negative direct effect of cultural landscape type on recommendation intention after controlling for the indirect effect of place attachment, despite a positive total effect driven by strong place attachment. This counterintuitive result suggests that high-culture landscapes, when failing to evoke deep emotional bonds, may lead to hesitation in direct recommendations due to perceived complexity or aloofness.

Secondly, regarding the hypothesis on the mediating role of perceived restorativeness and its moderation by landscape naturalness, the results reveal a more complex dynamic relationship. The significant main effect of landscape naturalness on perceived restorativeness aligns with Attention Restoration Theory [57] and the view that natural environments promote psychological restoration [58]. The interaction effect between cultural landscape type and landscape naturalness was highly significant. Further simple effect analysis and moderated mediation analysis using PROCESS Model 7 confirmed that cultural landscape type had a significant positive effect on perceived restorativeness, as did landscape naturalness. Crucially, their interaction had a significant negative effect on perceived restorativeness. This means that the H2 presupposition, “under high naturalness conditions, the positive influence of cultural landscape type on perceived restorativeness would be stronger, leading to the highest level of perceived restorativeness in high culture-high naturalness landscapes”, is not fully supported. This indicates that in environments with lower naturalness (e.g., cultural sites within urban built-up areas), enhancing cultural attributes can significantly boost tourists’ perceived restorativeness, thereby promoting recommendations. However, in environments that are already highly natural (e.g., natural scenic areas with cultural accents), the marginal contribution of adding cultural elements to perceived restorativeness may diminish, or could even interfere with, the restorative experience provided by the pure natural environment by introducing excessive artificial elements [58]. In such cases, solely emphasizing cultural attributes might not be the optimal strategy for enhancing perceived restorativeness. The mediating role of perceived restorativeness between cultural landscape type and recommendation intention is significantly moderated by landscape naturalness, exhibiting a complex pattern. Under low naturalness conditions, high-culture landscapes exert a significant positive indirect effect on recommendation intention via perceived restorativeness; conversely, under high naturalness conditions, this indirect effect was negative and non-significant. This unexpected finding suggests that high naturalness may attenuate the unique contribution of cultural landscapes to perceived restorativeness, possibly because high-naturalness environments inherently provide sufficient restorative experiences [54].

Analysis of the P2 component revealed that low-culture landscapes, particularly under low naturalness conditions, captured individuals’ early attentional resources more readily. This may reflect rapid detection and vigilance toward potentially negative or undesirable environmental features. The P2 analysis indicated that high-culture landscapes elicited significantly lower mean P2 amplitudes than low-culture landscapes. According to attentional resource allocation theory, increased P2 amplitude typically reflects automatic attentional capture by negative or potentially threatening stimuli [14]. The higher P2 amplitudes evoked by low-culture landscapes suggest that this category may activate stronger negative attentional biases, requiring more cognitive resources for processing. This finding aligns with Huang and Luo’s (2006) research demonstrating that negative stimuli induce larger P2 amplitudes, reflecting the brain’s prioritized processing mechanism for potential threat information [71]. Specifically, under low culture–low naturalness conditions, the P2 amplitude was significantly higher than that under other conditions, indicating that these landscapes likely contain more discomforting elements or features demanding additional cognitive effort. This supports evidence that visually discordant or unattractive environmental elements automatically attract attentional resources, manifesting as enhanced P2 amplitudes [81]. In contrast, high-culture landscapes, regardless of naturalness level, evoked lower P2 amplitudes, suggesting that cultural elements may reduce negative attentional processing by providing familiarity and meaning.

The LPP is typically associated with affective processing and subjective evaluation of stimuli, with more emotionally evocative stimuli eliciting stronger LPP responses [90]. Analysis of the LPP component demonstrated that low-naturalness landscapes, especially when combined with low-culture landscapes, triggered more intense sustained cognitive processing directed toward negative emotional experiences. The results show significantly higher LPP amplitudes for low-naturalness landscapes compared to high-naturalness landscapes. The LPP is widely recognized as relating to emotional processing, particularly with negative emotional experiences [90]. Higher LPP amplitudes generally indicate stronger negative emotional reactions or emotion regulation demands. Thus, the elevated LPP amplitudes evoked by low-naturalness landscapes suggest that these environments likely induced more intense negative emotional experiences. A significant interaction was found between cultural landscape type and naturalness. Under low culture–low naturalness conditions, LPP amplitude peaked, indicating that this landscape combination likely provoked the strongest negative emotional responses. This negative valence may stem from low-naturalness landscapes lacking restorative qualities, resonating with Attention Restoration Theory [57], which posits that the absence of natural environments may weaken tourists’ emotional bonds and thereby affect overall destination evaluations. Conversely, high-culture elements appeared to mitigate negative emotional responses evoked by low-naturalness landscapes, evidenced by significantly lower LPP amplitudes under high culture–low naturalness conditions compared to low culture–low naturalness. This finding supports Brown and Raymond’s (2007) perspective that landscape values can enhance visitor experiences by providing meaning and place attachment [91].

The combined P2 and LPP results delineate the dynamic process of tourists’ information processing for different landscape combinations. When encountering “low culture–low naturalness” landscapes, tourists exhibited not only stronger early attentional capture but also sustained heightened neural activity extending into later deep processing and affective evaluation stages. Low culture–low naturalness landscapes demanded greater attentional resources for processing and simultaneously elicited stronger negative emotional responses. This neurophysiological evidence supports Stress Reduction Theory [92], indicating that environments lacking cultural significance and natural elements increase cognitive load and induce negative emotions. Culturally and artistically relevant landscape components with high aesthetic value serve as crucial restorative attributes that effectively improve visitors’ physiological responses and emotional states [93]. The low culture–low naturalness combination may exacerbate tourists’ negative emotional reactions, suggesting that landscape planning should avoid dual deficiencies in cultural and natural elements to reduce emotional aversion. Notably, our study found that neural responses evoked by high-culture landscapes remained relatively stable across different naturalness levels, indicating that rich cultural elements may act as environmental “stabilizers”. This aligns with Li et al.’s (2023) finding that culturally significant landscapes enhance visitor attractiveness by providing cognitive coherence and place identity [46]. This discovery offers neurophysiological evidence for understanding how cultural landscapes enhance place attachment by reducing cognitive conflict and negative emotional responses. In summary, the EEG findings not only confirm the joint influence of cultural and natural attributes on visitor experiences, emphasizing the importance of avoiding “double-negative” (low culture and low naturalness) landscape combinations, but also provide neuroscientific insights for optimizing landscape design to enhance the positivity of tourists’ immediate experiences in destination planning and management.

Study 3 compared the performance of three machine learning models (DT, SVM, and GBDT) in classifying high versus low-naturalness landscapes using P2 and LPP EEG features. The Decision Tree model demonstrated the best average classification performance (mean accuracy 0.6529), significantly exceeding the chance level. This suggests that P2 and LPP features could be utilized by the DT model to distinguish the two landscape types to some extent, potentially reflecting nonlinear relationships between these features and the landscape naturalness perception that the model captured. However, its high standard deviation (0.1310) and maximum accuracy reaching 1.0 under certain conditions indicate strong sensitivity to data partitioning. This may relate to dataset size, feature distribution heterogeneity, or inherent individual differences in EEG signals. As an ensemble learning method, the GBDT model achieved intermediate mean accuracy (0.5596), also above the chance level. Similar to the DT, the GBDT’s performance showed considerable fluctuation (standard deviation 0.1322) and likewise achieved perfect classification under specific data partitions. Theoretically, the GBDT can improve performance through iterative optimization, but its slightly inferior performance compared to the DT in this study may relate to limitations in parameter tuning, feature set signal-to-noise ratio, or model complexity relative to the current data scale. While SVMs are effective in high-dimensional EEG data spaces, with different kernels exploring linear and nonlinear decision boundaries [94], the SVM model performed least optimally in this task. Its mean accuracy (0.3876) fell far below expectations and even below the chance level. Despite the SVM’s strong performance in many classification problems, it failed to effectively utilize P2 and LPP features for classification in this specific task, even after parameter optimization (using RBF kernel). The low accuracy and relatively small standard deviation (0.0779) indicate that the SVM model stably underperformed. This may suggest poor linear separability of the P2 and LPP features, or inadequate separability in the high-dimensional space mapped by the RBF kernel. It could also indicate the insufficient informativeness or unsuitability of the current feature set for the SVM model. The ability of the DT and GBDT models to achieve 1.0 accuracy under certain data partitions indirectly validates the conclusion from “Study 2” that P2 and LPP features potentially exhibit differences in distinguishing the two landscape types, confirming that these features indeed carry neural activity information related to landscape naturalness. However, the performance fluctuations across all models (particularly the high standard deviations of the DT and GBDT) and the overall inefficiency of the SVM also reveal limitations in the current approach.

7.1. Theoretical Implication

The theoretical contribution of this study lies in successfully integrating Place Attachment Theory, Attention Restoration Theory, and the Associative–Propositional Evaluation Model, providing a more comprehensive and multidimensional theoretical framework for understanding how cultural and natural factors jointly shape tourist experiences and subsequent behavioral intentions. Specifically, the research not only reveals that characteristics of cultural landscapes evoke emotional bonds but also demonstrates that their integrated cultural and natural elements may collectively promote attentional restoration. Meanwhile, rapid cognitive and affective responses elicited by landscape images serve as key associative-based evaluative evidence, ultimately influencing behavioral decisions such as recommendation intention through individuals’ propositional evaluation processes. This study pioneers the introduction of the APE model [16] into tourism research, addressing limitations in traditional tourism experience studies that focused exclusively on “propositional attitudes” by capturing tourists’ “associative attitudes” toward landscapes through EEG technology. This dual-path approach measuring both explicit and implicit attitudes enhances the scientific rigor and objectivity of tourism experience and attitude research, responding to theoretical demands for understanding the multidimensional nature of attitude structure. It represents the first distinction and integration of “propositional attitudes” and “associative attitudes” in tourism experience research, revealing the dual structure of tourists’ explicit evaluations (via self-reports like questionnaires) and implicit reactions (captured via EEG) toward destinations. This perspective addresses previous over-reliance on self-reports and neglect of subconscious affective associations, providing a new theoretical framework for understanding the complexity of authentic tourist experiences. By combining measurements of explicit and implicit attitudes, this study advances scientific understanding of the multidimensional structure of tourism experiences.

This study extends Place Attachment Theory [36] from traditional resident–place bonding to World Heritage tourism contexts, systematically examining the mechanism through which cultural landscape type influences tourists’ behavioral decisions via place attachment. It reveals the indirect effects on recommendation intention through a dual-path mediation mechanism (place attachment and perceived restorativeness). This study also indirectly suggests that cultural attributes themselves can serve as significant “place characteristics” that evoke place attachment, providing concrete exemplification for the constituent elements of “place” within Place Attachment Theory. This extension not only deepens the explanatory power of Place Attachment Theory regarding tourist psychological mechanisms but also establishes a foundation for interdisciplinary research. By confirming the critical mediating role of place attachment in how cultural landscapes influence recommendation intention, this study further enriches theoretical applications in cross-cultural tourism contexts.

This study makes significant extensions to Attention Restoration Theory [57]. While traditional ART primarily emphasizes the restorative efficacy of natural environments, this study innovatively explores the interaction between “naturalness” and “cultural attributes” in high-culture landscapes and their combined effects on individuals’ affective responses and perceived restorativeness quality. Notably, the finding that high naturalness may attenuate the contribution of cultural landscapes to perceived restorativeness suggests a complex dynamic relationship between natural and cultural elements in restorative experiences. The discovery that high-culture landscapes enhance perceived restorativeness quality indicates that not only pristine natural environments but also landscapes integrating profound cultural heritage with certain naturalness can provide significant attentional restoration. This expands ART’s application boundaries, suggesting that cultural elements may function as or synergize with natural elements to promote psychological restoration. Furthermore, through neurophysiological measurements such as EEG, this study objectively reveals neural mechanisms through which cultural landscapes engage neural pathways to elicit restorative experiences, providing new empirical evidence dimensions for ART.

7.2. Managerial Implication

The managerial contribution of this study lies in providing scientific guidance for tourism destination managers and marketers. Through design and experiential guidance that promotes tourists’ personalized emotional engagement and meaning-making, cultural value can be transformed into recommendation motivation, thereby enhancing destination appeal. Future scenic area managers should design tourism products and souvenirs that align with tourists’ cultural needs and natural experience preferences to strengthen place attachment and perceived restorativeness, ultimately boosting destination recommendation intention and overall competitiveness. However, the very success of cultural tourism, which brings significant economic and social benefits, can lead to the critical issue of “overtourism”, especially in historic cities with fragile infrastructure and limited carrying capacity. Overtourism is a phenomenon where the influx of tourists negatively impacts the quality of life for residents and diminishes the quality of the experience for visitors, threatening the very cultural and natural heritage that attracts them. Therefore, managerial strategies must extend beyond promotion to include proactive urban and spatial planning to ensure long-term sustainability. This study highlights the importance of avoiding dual deficiencies in cultural and natural elements during landscape planning, as low-culture–low-naturalness landscape combinations may exacerbate tourists’ negative emotional reactions and reduce emotional aversion. The findings provide empirical support for destination managers to facilitate synergistic development between cultural/natural heritage conservation and sustainable tourism. They advocate transforming the tourism industry from subjective evaluations toward more scientific, objective, and multidimensional comprehensive assessment methods, such as promoting scenario experiments and neuroscientific approaches in tourism research and visitor feedback evaluation, to enhance operational decision-making quality. This study also reveals that high-culture landscapes induce relatively stable neural responses across different naturalness levels, indicating that rich cultural elements may serve as “stabilizing agents” for environmental experiences. It is recommended to avoid dual deficiencies in cultural and natural elements during landscape planning to reduce tourist emotional aversion. Additionally, the machine learning model for landscape naturalness recognition developed in this study offers references for future brain–computer interface algorithm development in tourism.

7.3. Research Limitations and Future Directions

Although this study provides a reference for cultural landscape tourism, it also has some limitations. The research mainly focuses on specific World Heritage sites or specific tourist groups, and its conclusions may have limitations when extended to other types of cultural landscapes, natural environments, or tourists with different cultural backgrounds and demographic characteristics. Although neuroscientific methods such as electroencephalography (EEG) provide objective data, the environment of “scenario experiments” may differ from tourists’ experiences in real and complex tourism scenarios, which may affect the ecological validity of the research results. There are significant limitations in using static images as the experimental stimulus materials in this study. Static images cannot fully capture the dynamic features and immersive experiences of cultural landscapes, which is quite different from the multi-sensory experiences of tourists in real tourism scenarios. The complexity and presentation of the images may distract the participants’ attention and introduce potential confounding variables. For example, the differences in visual elements may lead to an uneven distribution of participants’ attention to certain stimuli. Although we attempted to control the basic features of the images in the research design, there is still room for improvement in image selection and standardization. There are some unresolved limitations in this study when adopting a multi-method design and interpreting the results. For example, although ERP measurements in a laboratory environment can provide neurophysiological data with high temporal resolution, their ecological validity is relatively limited. The real cultural landscape experience involves complex factors such as multi-sensory stimulation, social interaction, and environmental atmosphere, all of which cannot be fully reproduced in a controlled experimental environment. Importantly, this study’s reliance on visual stimuli (photographs) represents a limitation, as it fails to capture the multi-sensory nature of actual tourism experiences. Beyond visual perception, soundscapes play a crucial role in shaping human perception and behavior in tourism contexts. With the emergence of virtual reality and similar technologies, realistic simulations increasingly incorporate not only visual but also auditory and other sensory experiences, suggesting that future tourism research must adopt a more holistic sensory approach. Furthermore, in this study, dividing the landscape into “high/low culture” and “high/low naturalness” may simplify the complexity and diversity of cultural and natural attributes. Tourists’ perception of culture and nature is highly individualized and contextualized. Although the research explored the mediating roles of place attachment and perceived restorativeness as well as the moderating role of naturalness, there might still be other important psychological mechanisms or situational factors that affect tourists’ experiences and behavioral intentions and that were not included in the model for investigation. The research mainly adopted cross-sectional data, making it difficult to capture the changes in tourists’ attitudes and emotional connections over time or through multiple visits.

Future research should supplement samples from different cultures, regions, and ethnic groups to expand the application scope of the research. Future research could attempt to combine multimodal data such as eye tracking, facial expression analysis, and physiological indicators (such as heart rate variability) to capture tourists’ immediate reactions and deep experiences in a more three-dimensional way. Future research should pay more attention to the standardization of visual stimuli to ensure the consistency of images in terms of resolution, composition, and content complexity, so as to reduce the interference of irrelevant variables. Future research should further explore how to collect neurophysiological data in a more natural environment and combine field studies to verify the transferability of laboratory results. Crucially, future studies should strive to incorporate a wider range of sensory inputs. Given that soundscapes significantly impact perception and behavior, research could explore the influence of auditory elements, perhaps by integrating authentic soundscape recordings with visual stimuli or by utilizing virtual reality (VR) and similar immersive technologies that increasingly simulate multimodal real-world experiences, including both visual and auditory cues. We encourage the application of other neuroscience methods in a broader field of tourism research and actual operation assessment, with the aim of promoting the transformation of the tourism industry from subjective evaluation to a more scientific, objective, and multidimensional comprehensive assessment approach. Future studies should also extend the research framework and conclusions to different types of tourist destinations (such as intangible cultural heritage sites, urban parks, etc.), test the universality of the theoretical model, and enrich the application of Place Attachment Theory in cross-cultural tourism contexts. It is possible to track the dynamic change process of tourists’ place attachment, perceived restorativeness, and behavioral intention after multiple visits or long-term experiences. The long-term impact of the interaction between culture and natural elements on the visitor experience should be explored, and a longitudinal research design to track the formation and evolution process of place attachment should be established. Future research could also incorporate more potential individual difference variables (such as tourists’ cultural involvement, previous experiences, environmental values, etc.) and situational factors to more comprehensively reveal the complexity of the interaction mechanism between tourists and destinations. In addition to EEG and questionnaires, future research could attempt to integrate multimodal data such as eye movement tracking, facial expression analysis, and physiological indicators (such as heart rate variability) to capture tourists’ immediate reactions and deep experiences in a more three-dimensional way.

8. Conclusions

This study systematically explores how cultural landscape types influence tourists’ recommendation intentions through place attachment and perceived restorativeness, while revealing the moderating role of landscape naturalness. Integrating Place Attachment Theory, Attention Restoration Theory, and the APE model, this study employs a multi-method design (scenario experiments, ERP, and machine learning) to offer new insights into cultural heritage tourism. The results show that high-culture landscapes significantly enhance recommendation intentions via place attachment, with stronger effects of perceived restorativeness under low naturalness conditions. Theoretically, this study extends the application of Place Attachment Theory in tourism contexts and uncovers the dual structure of tourist attitudes through neurophysiological data. Practically, it provides strategies for heritage site managers to optimize landscape design and balance cultural and natural elements. Limited by specific heritage sites and static image use, future research should incorporate multi-sensory stimuli and cross-cultural samples to enhance generalizability and ecological validity.