How Environment Features Affect Children’s Emotions in Natural Playgrounds: A Context-Specific Case Study in China

Abstract

Natural playgrounds have garnered growing attention as supportive environments for children’s mental health. This study develops an analytical framework grounded in affordance theory and incorporates the Pleasure–Arousal–Dominance (PAD) model to examine the relationships between physical environmental features—and their combinations—in natural playgrounds and children’s emotional perceptions. Using the Yunhu Natural Playground in Fuzhou, China, as a case study, we selected seven typical behavior setting units. Environmental features were assessed through UAV imagery and on-site observations, while PAD-based visual questionnaires were employed to collect emotional responses from 159 children. By applying correlation analysis, random forest, and regression tree models, this study identified key environmental predictors of children’s emotional responses and revealed heterogeneous mechanisms across the three emotional dimensions. The results indicated that seasonal flowering/fruiting plants, accessible lawns, and structured play facilities were critical in supporting children’s pleasure, arousal, and dominance. Specifically, pleasure was primarily associated with sensory enjoyment and contextual aesthetics, arousal favored open grassy areas, and dominance was linked to environments with clear structure and manipulability. Based on these findings, this study proposes a spatial configuration strategy characterized by “nature as foundation, play encouraged, and structure clarified” to promote the positive development of children’s multidimensional emotional experiences. This research contributes empirical evidence on the role of physical environmental features in supporting children’s play behaviors and expands the theoretical understanding of the “emotional effects” of green spaces. While the findings are exploratory and context-specific, they emphasize the critical role of the sensory–behavioral–emotional chain in shaping children’s well-being and provide theoretical and practical guidance for the design of emotionally supportive, child-friendly, natural play environments in schools, parks, and residential areas.

1. Introduction

1.1. Reconnecting Children with Nature for Mental Health and Well-Being

According to the United Nations, 66% of the global population is projected to live in urban areas by 2050 [1]. High-density urban environments are often characterized by spatial compression and highly artificial built settings, which have significantly reduced opportunities for children to experience nature. Furthermore, modern urban lifestyles have intensified children’s physical disconnection from the natural world. The widespread use of digital devices, parental concerns over safety, and the scarcity of outdoor spaces have led children to spend increasing amounts of time indoors, gradually losing everyday contact with nature. Richard Louv coined the term “nature-deficit disorder” to describe this phenomenon, which may contribute to attention deficits, emotional dysregulation, anxiety, and social difficulties, particularly in childhood [2]. One of the most pronounced and developmentally salient outcomes is children’s emotional disturbance, often manifested as anxiety, aggression, and behavioral problems such as tantrums, bullying, and rule-breaking.

In this urban context, blue and green spaces serve not only as critical mediators of children’s contact with nature but also as important settings for emotional regulation. A growing body of research has demonstrated that active engagement with nature can promote children’s physical activity, social interaction, and emotional adjustment, thereby contributing to their mental well-being [3,4]. However, despite the well-established long-term benefits of blue and green spaces, their provision remains structurally inadequate and often underutilized in practice. Therefore, optimizing the design of urban green and blue spaces to rebuild children’s connections with nature is a pressing issue for both child-friendly and healthy city agendas.

1.2. Natural Playgrounds as a Strategy to Restore Children’s Nature Engagement in Urban Contexts

Natural playgrounds, as child-specific spaces within urban green and blue infrastructures, integrate natural elements, spatial openness, and unstructured layouts, and are considered an effective strategy for restoring children’s engagement with nature in cities. Compared with conventional “KFC” (kit–fence–carpet) playgrounds, natural playgrounds emphasize free exploration and material plasticity, which are more likely to stimulate children’s autonomy, creativity, and physical participation [5]. The design concept draws on the nature-based educational philosophies of Friedrich Fröbel and John Dewey, who emphasized that children learn through sensory interaction with nature and grow through social experience [6,7].

While a growing body of practice has accumulated around natural playgrounds, most empirical studies to date have focused on their contributions to children’s play behaviors or long-term mental health outcomes [8,9,10]. Limited attention has been paid to how natural playgrounds influence children’s emotional perception. This research gap needs to be addressed to fully understand how nature-based playground environments contribute to the promotion of children’s positive emotions. In particular, in high-density urban areas, natural design may offer valuable emotional support alternatives for children with limited access to everyday nature.

1.3. Applying Affordances and Behavior Settings to Interpret Children’s Emotional Perception in Natural Playgrounds

Understanding how children benefit psychologically from their interactions with the environment has become a central concern in environmental–behavior research. Recent studies suggest that the mental health benefits of environmental exposure depend not only on the presence of natural elements but also on how individuals interact with their surroundings [11]. Therefore, in natural playgrounds, which are inherently dynamic and unstructured, children may tend to actively identify and utilize environmental features based on their interests, abilities, and contexts—leading to embodied, personalized behavioral experiences that foster emotional meaning and perception.

In this process, affordance theory and behavior setting theory jointly provide a theoretical foundation for understanding meaningful engagement as a key mechanism linking environment, behavior, and emotion. Affordance theory emphasizes the functional opportunities arising from dynamic person–environment interactions [12], while behavior setting theory focuses on the spatiotemporal organization of behavior under social norms [13]. These frameworks are complementary in natural playground settings; affordances highlight how the environment supports possible actions, whereas behavior settings delineate discrete ecological units that connect spatial features with patterned child activities.

To strengthen the empirical applicability of these theories, researchers have developed typologies of affordances based on physical functions, such as flat surfaces, slopes, sheltered spaces, fixed structures, loose objects, malleable materials, and water elements. These components are often associated with prototypical forms of children’s activities and provide a more child-relevant vocabulary for describing environmental features [14,15,16,17]. In parallel, the concept of behavior setting units enables the integration of physical space, activity patterns, and social meaning into bounded ecological units. This analytical approach facilitates the systematic identification of potential linkages between environmental features, child behaviors, and emotional perceptions in natural playgrounds [18].

1.4. Aims of This Study

Grounded in the theoretical perspectives discussed above, this study adopted an affordance-based analytical framework to examine the relationship between environmental features and children’s emotional perceptions in a representative natural playground in China. Two research questions guided this investigation:

(1) What are the relationships and their strengths between specific environmental features and children’s emotional perceptions in natural playgrounds?

(2) Which combinations of environmental features are more likely to foster more positive emotional responses in children?

Theoretically, this study integrated emotional perception with affordance theory to explore how environmental support for children’s activity behaviors may constitute a fundamental component of their emotional experiences and well-being. This integration may introduce a complementary lens to understand how environmental affordances may relate to children’s psychological experiences and long-term well-being. From a practical standpoint, the findings may provide valuable insights for designing and planning urban green and blue spaces that are supportive of children’s mental well-being.

2. Methodology

2.1. Study Area

In recent years, an increasing number of newly developed playgrounds have adopted the defining features of natural playgrounds, most of which are concentrated in major first-tier cities such as Beijing, Shanghai, Shenzhen, Guangzhou, and Chengdu. These playgrounds generally follow several shared design principles, as follows:

(1) Design philosophy—emphasizing the core concept of “returning to nature, encouraging free exploration, and integrating play with learning,” with a focus on naturalized spatial layouts and material selection to facilitate children’s sensory experiences, physical development, and social interaction through self-directed exploration;

(2) Facility types—integrating natural environmental elements with built play structures to create a synergistic relationship between natural features and functional installations;

(3) Activity orientation—promoting unstructured play by providing open and varied spaces that meet the diverse activity needs of children across different age groups.

This study was conducted in the Yunhu Natural Playground, located in Fuzhou, Fujian Province, China. The site embodies the typical characteristics of natural playgrounds while also presenting three distinctive features compared to similar sites elsewhere in China:

(1) High environmental heterogeneity—within a limited area, the playground incorporates multiple types of affordances, including flat surfaces, slopes, sheltered areas, fixed structures, loose objects, moldable materials, and water features, providing a diverse range of play opportunities for children;

(2) Unique locational context—situated on the urban fringe, most of the participating children visited the playground for the first time, thereby reducing potential bias from prior experiences and enabling a more accurate capture of immediate emotional responses;

(3) Enclosed management model—as a closed and controlled play environment, external disturbances were minimized, allowing this study to focus more directly on the relationships between environmental features, children’s behaviors, and emotional perceptions.

Based on the site’s design logic and preliminary behavioral observations, the research team identified seven main behavior setting units (BSUs): DNA Loop, Youlan Pool, Fern Mist, Infinity Dome, Shell Hill, Calla Lily Slide, and Dengyun Overlook (Figure 1). The boundaries of each BSU were defined by natural or artificial elements, such as vegetation screens or pathway junctions, providing spatially explicit units for subsequent analyses of environmental features, behavioral patterns, and emotional perceptions.

2.2. Data Collection

2.2.1. Participant Recruitment

Data were collected through on-site field surveys combined with random questionnaire distribution. To ensure the rigor of the questionnaire design, a small-scale pilot test was conducted prior to formal implementation to assess its reliability and validity. The formal survey was carried out over two days in spring under clear weather conditions without rain or strong wind, including one weekend day and one weekday. Each survey day comprised three time periods (08:00–10:00, 12:00–14:00, and 16:00–18:00), covering different times of day and various areas of the playground. Four researchers participated in questionnaire distribution simultaneously and provided assistance to respondents who had difficulty completing the survey, ensuring the comprehensiveness and accuracy of data collection.

All procedures involving child participants were conducted after obtaining informed consent from parents or guardians and voluntary assent from the children. All data were anonymized and used solely for research purposes.

2.2.2. Assessment of Children’s Emotional Perception

To assess children’s emotional responses across different BSUs, a perception-based questionnaire was developed following the PAD model proposed by Mehrabian and Russell [19]. While the PAD model originates from environmental psychology and is traditionally used for affective evaluation, its integration here is theoretically justified by framing the three dimensions—pleasure, arousal, and dominance—as affective correlates of actualized affordances. In natural playgrounds, pleasure can be interpreted as the intrinsic enjoyment derived from sensory-rich and aesthetically engaging affordances; arousal reflects the behavioral activation triggered by affordances supporting movement and exploration; and dominance indicates the perceived agency afforded by controllable or structured elements. These emotional states are not only immediate outcomes but also influence children’s motivation, learning, and future interactions with the environment.

The model characterizes emotional states along three dimensions: (1) Pleasure–Displeasure, representing positive versus negative affective states; (2) Arousal–No arousal, indicating the level of physiological activation and alertness; and (3) Dominance–Submissiveness, reflecting the perceived degree of control over the environment. To adapt the PAD model for younger children, visual illustrations were used instead of verbal descriptions. Each dimension was rated on a 9-point scale using a cartoon-based semantic differential format representing low, medium, and high emotional states. Opposing adjectives were used to define the poles of each dimension. Low pleasure represented sadness, fear, or anger, while high pleasure indicated happiness and contentment; low arousal denoted calmness or fatigue, and high arousal reflected excitement or tension; low dominance reflected helplessness, while high dominance represented a sense of environmental control.

To facilitate memory recall of specific spatial experiences, a combination of the se-mantic differential method and photo-assisted recall technique was adopted [20,21]. On a sunny spring day, the research team captured 3–4 images per BSU between 8:00 a.m. and 5:00 p.m., forming an image set. All photos were taken from a child’s perspective (1–1.3 m above ground) and showcased typical play features, natural elements, and child activities in the space. During the survey, trained interviewers guided children to recall their actual experiences based on the image sets and asked them to select the face that best represented their emotional state for each of the three PAD dimensions. Consistency in understanding and scoring was ensured throughout the process (Figure 2).

2.2.3. Assessment of Environmental Features in the Playground

To systematically identify the physical characteristics of the natural playground, we applied an affordance-based classification framework to select relevant environmental feature indicators. As noted earlier, this assessment is grounded in the premise that children’s interactions with the physical environment are mediated by the functional opportunities it affords. The resulting multidimensional indicator system provided sufficient spatial and functional resolution to support subsequent analyses of behavior–environment relationships.

The assessment followed an adapted version of the classical affordance typology, comprising eight types of environmental features. Adjustments were made based on field conditions. Specifically, (1) flat and sloped surfaces were merged as “open surfaces,” represented by accessible green areas or paved ground; (2) shelter was defined based on the presence of usable shade or built canopies; (3) attached objects included immobile landscape elements such as rocks, sculptures, or tree trunks; (4) non-rigid, attached objects referred to flexible or swingable elements such as hanging branches or soft frames; (5) graspable/detached objects encompassed loose paving materials or naturally occurring play items such as flowers or fruits; (6) moldable materials referred primarily to sand or soil; and (7) water features included ponds or shallow water bodies with accessible edges.

These indicators were then coded and evaluated for each BSU using drone-captured orthophotos, digital base maps, and on-site field observations (

Table 1. Indicators of natural playground features based on affordance typology.

Affordance-Based Typology		Environmental Feature Indicators
Flat and sloped surfaces	Artificial features	X1—proportion of hard paving (%)
	Natural features	X2—proportion of vegetation coverage (%)
		X3—proportion of accessible lawn (%)
Shelter	Artificial features	X4—types of artificial shelters
		X5—count of artificial shelters
		X6—proportion of usable shaded area (%)
	Natural features	X7—canopy closure (%)
Attached objects	Artificial features	X8—types of fixed play equipment
		X9—count of fixed play equipment
Non-rigid, attached objects	Artificial features	X10—types of movable play equipment
		X11—count of movable play equipment
Graspable/detached objects	Artificial features	X12—types of loose ground
		X13—proportion of loose ground (%)
	Natural features	X14—types of flowering/fruiting plants
		X15—types of usable seasonal flowering/fruiting plant
Moldable material	Natural features	X16—types of malleable paving
		X17—proportion of malleable paving (%)
Water	Natural features	X18—types of waterscape
		X19—proportion of waterscape (%)

). Each affordance type was quantified using three dimensions: (1) area-based indicators, such as the percentage of paved surface, green cover, accessible lawn, and total usable shade; (2) structural attributes, including types of associated facilities or vegetation; and (3) count-based indicators, such as the number of fixed/movable structures or the variety of usable plant species. This multidimensional indicator system ensured adequate spatial and functional resolution for subsequent behavior–environment analyses.

2.3. Data Analysis

2.3.1. Reliability and Validity Testing

To assess the internal consistency of the PAD-based questionnaire, Cronbach’s alpha was calculated across 21 emotion-rating items. An alpha coefficient above 0.8 was considered good, and an alpha coefficient above 0.7 was considered acceptable for the child sample in this study.

Construct validity was tested using the Kaiser–Meyer–Olkin (KMO) measure and Bartlett’s test of sphericity. The KMO value of 0.755 indicated satisfactory sampling adequacy (>0.70), and the Bartlett test result (p < 0.001) confirmed the appropriateness of the data for exploratory factor analysis.

2.3.2. Correlation Between Environmental Features and Emotional Perception

To identify environmental factors potentially influencing children’s emotional perception, Pearson correlation coefficients were calculated between 19 environmental feature variables (X1–X19) and scores on the three emotional dimensions (pleasure, arousal, and dominance). Correlation strengths were interpreted as follows: 0.8–1.0 (very strong), 0.6–0.8 (strong), 0.4–0.6 (moderate), 0.2–0.4 (weak), and <0.2 (very weak or none). To enhance the robustness of subsequent modeling, variables with high collinearity were screened and excluded to reduce multicollinearity risks.

2.3.3. Variable Importance and Key Predictor Identification

After identifying significantly correlated variables and controlling for collinearity, random forest models were constructed for each emotional dimension (pleasure, arousal, and dominance). The models ranked variables based on their predictive importance. The percentage of variance explained (%Var explained) was used to assess model performance, while the increase in mean squared error (%IncMSE) served as the key indicator of variable importance. %Var explained reflects the proportion of variance in the target variable accounted for by the model. %IncMSE indicates how much predictive error increases when a variable is permuted, making it a widely used measure of feature importance in ensemble learning.

2.3.4. Pathway Identification and Emotional Threshold Analysis

To explore how combinations of environmental features affect children’s emotional perceptions, CART (Classification and Regression Tree) models were developed for each emotional dimension. These regression trees identified key variable combinations and dominant interaction pathways that produced the highest or lowest emotional scores. The models were constructed using the rpart algorithm, with continuous variables automatically split and the complexity parameter set to 0.0001 to allow for more detailed branching. Final trees provided clear visual outputs of terminal node emotion scores, supporting the interpretation of environmental conditions associated with high or low emotional responses.

3. Results

3.1. Reliability and Validity Analysis

This study collected affective ratings from 159 children across seven sets of images, covering the dimensions of pleasure, arousal, and dominance. The reliability analysis indicated a high internal consistency of the scale, with a Cronbach’s alpha of 0.895. Test–retest reliability also showed strong temporal consistency, with Pearson correlation coefficients for all three affective dimensions exceeding 0.7. Validity analysis demonstrated acceptable construct validity, with a KMO value of 0.755. Bartlett’s test of sphericity was significant (p < 0.001), indicating sufficient inter-item correlations for exploratory factor analysis.

3.2. Descriptive Statistics

Among the respondents, boys accounted for 56.2% and girls for 43.8%, with boys generally displaying higher levels of physical activity. Guardians assisted younger children (aged 0–6) in completing the questionnaire, while those aged 7–12 completed it independently. In terms of age distribution, children aged 4–6 comprised the largest proportion of participants.

Overall, children reported high levels of affective perception across different spatial settings, with notable variation between behavior setting units (Figure 3). Pleasure, arousal, and dominance scores tended to co-vary at high levels. However, correlations among the three dimensions were not consistently significant. When pleasure scores were low, arousal and dominance often diverged in opposite directions. High levels of dominance were not always accompanied by high pleasure or arousal.

3.3. Correlations Between Environmental Features and Emotional Perception

All environmental features were significantly correlated with at least one affective dimension (Figure 4).

Among them, the availability of seasonal flowering or fruit-bearing plants (X15) showed strong and positive correlations with all three dimensions (r = 0.385, p < 0.001 for pleasure; r = 0.323, p < 0.001 for arousal; r = 0.379, p < 0.001 for dominance). In contrast, greening coverage (X2) was strongly and negatively associated with pleasure (r = −0.408, p < 0.001), and canopy closure (X7) was negatively related to dominance (r = −0.312, p < 0.001).

A correlation matrix revealed high multicollinearity among some variables (Figure 5). To ensure model stability, variables at risk of multicollinearity were excluded, and only representative predictors—X2, X3, X6, X7, X9, X10, X12, X13, X15, X16, and X18—were retained for subsequent modeling.

3.4. Importance Ranking of Environment Features

The random forest models demonstrated good explanatory power across all three dimensions (Figure 6), with % variance explained values of 37.83% for pleasure, 27.16% for arousal, and 27.36% for dominance.

Plant-related features were the strongest predictors of children’s positive affect. Seasonal flowering/fruiting plant availability (X15) ranked first across all models. Greening coverage (X2) and canopy closure (X7) also ranked among the top five variables. Comparing the top five predictors across affective dimensions revealed differentiated mechanisms.

Pleasure was most strongly influenced by sensory and atmospheric features, with a notable role for water features (X18), suggesting pathways related to tranquility, imagination, or restorative experiences.

Arousal was more dependent on spatial openness, particularly accessible grass areas (X3), which support dynamic activities like running or exploration.

Dominance was linked to structured and controllable elements, such as fixed play equipment (X9), underscoring the importance of organized environments for perceived control—an aspect less relevant for pleasure or arousal.

Conversely, analysis of the least important predictors indicated that loose surface materials (X12), although minimally related to pleasure or arousal (incMSE = 6.2% and 4.9%, respectively), played a larger role in dominance (incMSE = 8.1%). On the other hand, water features (X18) contributed more substantially to pleasure and arousal (both incMSE = 15.9%) but had minimal influence on dominance (incMSE = 8.2%).

3.5. Environmental Feature Combinations and Affective Perception

The random forest results suggested that all environmental features contributed at least 5% to affective outcomes, indicating the value of a “multi-dimensional design” approach in natural playgrounds. Such spaces may benefit from combinations of rich seasonal vegetation, high greening coverage, moderate canopy density, open grass areas, and well-structured play equipment. These insights justified retaining all variables from the random forest model for decision tree analysis. A comparison of key variables across both models revealed consistency in feature importance.

In the pleasure model (Figure 7), predicted values ranged from 4.78 to 8.50. The lowest pleasure was associated with few flowering/fruiting plants (X15 < 2.5), low loose surface coverage (X13 < 0.091), and few mobile play features (X10 < 0.5). The highest pleasure corresponded to rich vegetation (X15 ≥ 2.5), low canopy density (X7 < 0.4185), and high greening coverage (X2 ≥ 0.6945). These results suggest that natural diversity combined with visual openness promotes pleasurable experiences.

In the arousal model (Figure 8), predictions ranged from 3.76 to 7.73, with the widest range (3.97) among the three dimensions, indicating strong discrimination. The lowest arousal occurred when flowering/fruiting plants were limited (X15 < 3.5) and greening coverage was extremely high (X2 ≥ 0.9145), suggesting visual monotony or constrained activity space. The highest arousal was observed under conditions of rich seasonal vegetation (X15 ≥ 3.5), low loose surface coverage (X13 < 0.053), and limited grass area (X3 < 0.156), indicating the supportive role of diverse but balanced natural elements.

In the dominance model (Figure 9), predicted values ranged from 5.39 to 8.42. The lowest scores occurred with limited vegetation (X15 < 3.5) and minimal grass area (X3 < 0.016), reflecting restricted opportunities for action. Conversely, dominance was highest in areas with sufficient vegetation (X15 ≥ 3.5), low loose surface coverage (X13 < 0.053), and limited mobile play equipment (X10 < 4), suggesting that clarity and structure facilitate feelings of control and autonomy.

The decision trees revealed dimension-specific interaction pathways. Overall, all three dimensions were shaped by nature-dominant configurations, modulated by secondary spatial elements. The dominance model exhibited the most complex tree, with nine terminal nodes, implying greater environmental sensitivity. The pleasure model showed a more balanced, simplified structure (six terminal nodes).

Seasonal vegetation (X15) and loose surface coverage (X13) were cross-dimensionally influential. X15 consistently appeared as the first-level node in all trees, indicating a dominant predictive role. X13 also showed stable, multi-dimensional importance across models.

Notably, X13 demonstrated distinct moderating mechanisms. In the pleasure model, X13 helped enhance pleasure under conditions of limited vegetation. In the arousal and dominance models, X13 acted as an amplifier in combination with grass area (X3) or mobile features (X10), supporting different emotional pathways. Thus, X13 served as a sensory enhancer in the pleasure and arousal models and as a manipulable resource in the dominance model.

4. Discussion

4.1. Overview

This study, grounded in the affordance perspective, systematically examined the influence of physical environmental features in natural playgrounds on children’s emotional perceptions. By integrating correlation analysis, random forest, and decision tree models, we identified how distinct environmental characteristics—individually and in combination—contribute to children’s positive emotional experiences. Plant-related features provided foundational support for multidimensional emotional responses. Beyond this, each emotional dimension was linked to different spatial cognition and behavior-support mechanisms. Pleasure was primarily influenced by sensory and experiential elements; arousal favored open grassy areas conducive to movement and exploration; and dominance was associated with structured and manipulable features. High-scoring paths across all emotional dimensions revealed a pattern of “nature-dominant with secondary modulation,” where pleasure was enhanced by highly natural yet visually open settings, arousal benefited from rich natural features with moderately bounded spaces, and dominance thrived in spatial combinations that were structured, resource-abundant, and minimally distracting. These findings suggest that a natural playground designed to simultaneously support pleasure, arousal, and dominance should incorporate a composite spatial configuration that balances sensory richness, spatial clarity, and behavioral support.

4.2. Interpretation of Main Findings

4.2.1. Natural Elements as Triggers of Positive Emotional Perceptions

Natural elements that afford children’s play behaviors—such as playable vegetation, accessible lawns, and loose ground surfaces—emerged as consistent triggers of pleasure, arousal, and dominance. Notably, playable vegetation exhibited the most significant positive effects across all emotional dimensions, underscoring its priority in design considerations. In contrast, overly “green” environments lacking playability and experiential engagement may lead to boredom and fail to elicit positive emotional responses. Prior studies have similarly indicated that reductions in depressive symptoms are associated not only with higher average greenness and native vegetation but also with a greater number of distinct natural features in the neighborhood [22].

This suggests that children’s perception of nature extends beyond visual exposure and is closely tied to behavioral affordances. Other natural features, such as the presence of insects and small animals, may also foster exploration and bodily engagement, reinforcing the link between sensory experience, behavior, and emotional involvement.

4.2.2. Differentiated Spatial Mechanisms Across Emotional Dimensions

As previously discussed, pleasure tends to rely on sensory enjoyment and atmospheric cues. This aligns with the affective primacy theory, which posits that visual elements in natural settings can swiftly trigger pleasant emotions [23]. Thus, the design of natural play spaces should emphasize the elicitation of “first emotional impressions.”

In contrast, arousal appears to be driven by spatial openness that supports physical movement and exploration. Our findings showed that high-arousal pathways involved moderate enclosure and open grassy areas. These findings resonate with Veitch et al. [24], who emphasized that spaces with good visibility, smooth lawns, and minimal obstructions promote high-intensity physical activity in children.

Dominance, meanwhile, was more responsive to spatial features characterized by structural clarity and manipulability. This finding corresponds with Bandura’s theory of self-efficacy [25], which highlights that environments perceived as controllable and predictable support children’s development of autonomy and agency.

In sum, the spatial mechanisms underlying pleasure, arousal, and dominance diverge. While pleasure is largely triggered by sensory cues and situational immersion, arousal and dominance highlight the dynamic interplay among perception, behavior, and emotion, suggesting the need to reframe emotional generation through the lens of affordances.

4.2.3. Multidimensional Spatial Strategies for Emotionally Diverse Playgrounds

Our findings indicate that emotional mechanisms rarely operate in isolation; rather, they interact to produce multifaceted emotional experiences. Overall, the spatial mechanisms underlying pleasure, arousal, and dominance were distinct. Pleasure was primarily triggered by sensory cues and immersive contexts, whereas arousal and dominance reflected the dynamic interplay among perception, behavior, and emotion. This pattern suggests the need to reconceptualize emotional generation from an affordance-based perspective. Such an interpretation aligns with Lester and Russell’s [26] proposition that children’s emotions are often activated through “compound pathways,” in which sensory stimuli, action opportunities, and environmental controllability converge to shape complex emotional responses.

Children’s emotional responses were not determined by single spatial features but were activated through specific combinations and interactions. The results support a composite design strategy that integrates naturalness, play encouragement, and structural clarity. Natural attributes consistently played a central role across all emotional dimensions—not only serving as sources of sensory pleasure but also forming the basis for behavioral engagement and a sense of control. Loose surfaces, as well as the structure and functionality of play equipment, exhibited differentiated moderating effects across emotional dimensions, underscoring the importance of nuanced spatial configurations.

4.3. Contributions and Implications

This study contributes to the growing discourse on the emotional effects of urban green spaces by offering a refined understanding of how physical environmental features in natural playgrounds influence children’s emotional experiences. Unlike prior research focusing predominantly on green exposure or greenness indices, our study emphasizes micro-scale, perceptible, accessible, and manipulable environmental features. The findings underscore the critical role of natural elements that afford children’s activities in fostering psychological well-being and confirm the significance of the perception–behavior–emotion chain. Moreover, they demonstrate the heterogeneity of emotional pathways across pleasure, arousal, and dominance dimensions.

In practical terms, the results provide actionable guidance for designing child-friendly green and blue spaces (Figure 10). First, to enhance pleasure, designers should incorporate a high diversity of plant species—including flowering and seasonal plants—arranged to provide rich visual and tactile stimuli, complemented by moderately open grassy areas that support free movement. Second, to stimulate arousal in a positive manner, moderately challenging play structures and varied terrain (e.g., gentle slopes, small mounds) can be integrated to encourage exploration and active engagement, while maintaining safety and accessibility. Third, to foster dominance—a sense of control and competence—spaces should include adaptable and manipulable facilities, such as movable play materials or modular seating, that allow children to shape their environment and activities.

Collectively, these principles can be synthesized into a spatial configuration strategy characterized by “nature as foundation, play encouraged, and structure clarified”. This approach can be adapted to diverse everyday settings, including school campuses, public parks, and residential green spaces. By aligning specific environmental features with distinct emotional outcomes, the strategy offers empirically grounded and context-sensitive guidance for planners and designers aiming to support the multi-dimensional emotional needs of children.

4.4. Limitations and Future Opportunities

Despite the methodological and theoretical contributions of this study, several limitations warrant consideration.

First, the data were derived from a single natural playground in one city. While the Yunhu Natural Playground offers a high diversity of affordance types and a controlled environment suitable for focused observation, it represents only one example within the broader spectrum of natural playgrounds in China. Its specific geographical, cultural, and climatic conditions may influence children’s emotional responses, and the findings should be interpreted within these contextual boundaries. Although the sampling strategy ensured representativeness within the site, the results may not be fully generalizable. They should be viewed as exploratory and context-dependent. Future research should consider cross-city or multi-regional comparative studies to position individual playgrounds within a broader national context and to test the universality and contextual adaptability of the identified spatial mechanisms.

Second, this study adopted a broad age range (0–12 years) without further stratification into narrower developmental stages. While this approach maximized sample size for statistical robustness, it may mask age-related differences in sensory perception and emotional cognition. Future research should recruit larger, balanced samples to enable age-stratified analyses (e.g., 3–5, 6–8, and 9–12 years), which could yield more precise insights into developmental variations in environmental–emotional responses.

Third, while this study focused on the influence of physical environmental features, it did not account for social environmental variables, such as peer interactions or adult behavior. These non-physical factors may also significantly shape children’s emotional experiences, suggesting the need for a more layered and cross-scale analytical framework in future research.

Fourth, although the combination of random forest and decision tree models facilitated the identification of important predictors and interaction effects, these models do not explicitly test the causal pathways between environmental features, behavioral mediators, and emotional outcomes. Future research could integrate structural equation modeling (SEM) or mediation analysis to verify hypothesized mechanisms.

Fifth, while multicollinearity was addressed in variable selection, this study did not further examine potential interdependencies or redundancies among environmental variables, such as the overlap between green coverage and the presence of flowering plants. Variable clustering techniques or interaction-term modeling could be adopted in future studies to refine predictor relationships and enhance explanatory capacity.

5. Conclusions

This study, informed by the theory of affordances, systematically examined how physical environmental features in natural playgrounds affect children’s emotional perceptions across pleasure, arousal, and dominance. Our findings highlight the central role of natural elements in eliciting positive emotions and reveal that each emotional dimension responds to environmental features through distinct mechanisms; pleasure is driven by sensory stimulation and situational aesthetics, whereas arousal and dominance reflect the importance of movement opportunities and environmental control.

Based on these insights, we propose a composite spatial configuration strategy—grounded in nature, supportive of play, and structurally clear—to guide the design of emotionally supportive, child-friendly play environments. Future studies should expand on this framework by incorporating multi-regional samples and additional influencing factors to develop more context-sensitive and robust explanatory models.