Integrated Eye-Tracking Response Surface Analysis to Optimize the Design of Garden Landscapes

Abstract

Gardens not only provide people with a place for leisure and relaxation, they also contribute to improving urban ecological environments and promoting social interactions and cohesion. Additionally, from a psychological perspective, gardens play a role in alleviating stress, enhancing happiness, and improving the quality of life. Current research on gardens has primarily employed methods such as questionnaire surveys, environmental psychology analyses, and eye-tracking analyses; however, comprehensive studies on the relationships between multiple factors and levels in garden designs are lacking. Here, we propose a response surface analysis approach based on eye-tracking technology for the design and optimization of gardens. Firstly, the impacts of different garden elements on visitors’ psychology and fixation counts were analyzed using environmental psychology and eye-tracking analyses. Subsequently, the optimal range of each garden feature was determined through single-factor experiments, followed by response surface analysis to obtain the optimal value for each element. The results revealed that changes in garden elements such as the greenery ratio, number of buildings, and water saturation significantly affected visitors’ psychology. The greenery ratio had a greater impact than the number of buildings, which in turn had a greater impact than water saturation. This study is the first to analyze the relationships between multiple garden elements. A strong relationship was found between the greenery ratio and the number of buildings, as well as between the number of buildings and water saturation, while the relationship between the greenery ratio and water saturation was weaker. This approach can not only optimize garden designs but can also be widely applied in fields such as urban planning and public space transformation to enhance visitors’ comfort and satisfaction with the environment and promote sustainable urban development.

1. Introduction

Environmental psychology is a discipline that investigates the perception, cognition, and emotional responses of individuals towards environmental landscapes [1]. Its primary focus is on studying the interplay between individuals and their surroundings, as well as examining the impact of the environment on an individual’s psychology and behavior [2]. On one hand, the relationship between humans and the environment is mainly influenced by culture, which promotes an understanding of human–environment interactions (i.e., cultural environmental psychology) [3]. On the other hand, researchers have explored the impact of the environment on an individual’s psychology, specifically by examining how the environment affects environmental sensitivity [4], place satisfaction [3], and the attachment effect [5]. In the field of landscape architecture, the study of environmental psychology primarily investigates the psychological responses of individuals to various garden elements, including their attention, preferences, and feelings of comfort [6]. For instance, studies have focused on people’s comfort levels in green spaces [7,8], the extent of their preferences for architectural aesthetics [9,10], and their levels of satisfaction with water features [11]. Psychological research on these factors facilitates the optimization of garden design and the creation of exceptional aesthetic experiences for individuals while also promoting relaxation and recreational activities, with the ultimate goal of creating environments that are beneficial to people [12,13,14].

Currently, research methods in environmental psychology primarily include self-report questionnaires, verbal interview protocols such as think-aloud procedures [15,16], electroencephalography (EEG) [17], and eye-tracking analyses [18]. In recent years, eye-tracking technology has been extensively applied in various fields, including advertising, user interface design, and human–computer interactions [19,20,21]. Eye-tracking analysis is a highly accurate tool that analyzes visual attention and cognitive processes by recording eye movements. It provides precise quantitative data, capturing users’ gaze points, gaze duration, and eye movement trajectories while observing landscapes [22]. Due to its real-time and objective nature, it is an ideal tool for studying landscape design and user experiences, with significant application value in these domains [18,23]. Applying eye-tracking technology in landscape research allows for attention and interest measurements to be recorded for different landscape elements, thus reducing biases during data analysis [18]. For example, eye tracking has been used to study the influence of individual landscape features, such as the proportion, shape, or height of green spaces, on visitors’ visual preferences [24,25]. In another case, visual heatmaps were generated to show the visual completeness between two elements (buildings or trees) or multiple landscape features (buildings, rivers, or greenery) using eye tracking from a micro-to-macro perspective [26,27]. Given the unique advantages of eye-tracking technology in landscape design research, we chose to apply this technique in our study to provide an in-depth analysis of users’ visual focal points and emotional responses to different landscape elements [28]. Through the analysis of eye-tracking data, we can uncover the specific effects of varying greenery ratios, architectural layouts, and water saturation on users’ visual attention and emotional experiences. This not only contributes to optimizing landscape design and enhancing the quality of user experiences, it also provides landscape designers with scientific evidence and supporting data, facilitating the development of landscape design theory and practice.

Response surface methodology (RSM), also known as the response surface design method, utilizes a multivariate quadratic regression equation to model the functional relationship between factors and response values. In 1951, Box and Wilson first introduced the Central Composite Design (CCD) [29], which has become one of the most commonly used designs for quadratic models, alongside the Box–Behnken design [30]. By combining multiple responses, the CCD enables the study of their behaviors and has led to the development of various optimization and desirability routines [31]. Through analysis of the regression equation, the RSM seeks to identify optimal process parameters and address the challenges associated with multivariate problems [32]. The response surface analysis method enables the examination of the regression relationship between experimental indicators (dependent variables) and multiple experimental factors (independent variables), which can take the form of either a curve or a surface. Currently, response surface analyses are primarily applied in industrial structural design and parameter optimization. For instance, design optimization has been performed for turbine drill bit structures by using the response surface method to analyze the relationships between various parameters in order to enhance the drilling efficiency of the bit [33]. Response surface analysis typically involves conducting single-factor experiments and designing experiments for the response surface through which optimal ranges of multiple factors, with each containing the optimal value, can be determined and a regression equation can be derived [34]. Due to its high efficiency, the response surface methodology has witnessed rapid development in the fields of environmental protection and urban planning [35]. For instance, in relation to the components of fermentation of vegetable residues mixed with a ratio of moisture (M) to solid content (S), initial pH value, and organic loading (OL), the fertilizer yield can be determined by using a regression analysis to determine the optimal fertilizer yield formula [36]. Research on landscape design has indicated that different landscape elements in courtyards and peripheral areas have an impact on the overhead layer, which can be determined using response surface analysis (RSA) [37]. Moreover, the proportions and spatial layouts of park elements such as lawns, trees, water bodies, hard surfaces, and buildings also influence urban parks [38]. However, the application of RSA in landscape design is still in the early stages of development. Furthermore, studying gardens as holistic entities rather than focusing on specific elements or features is crucial in order to avoid biased research outcomes. A comprehensive and objective assessment of research findings considers subjective perception and the interrelationships among different garden elements, as well as the overall compositional effects on garden design. Consequently, there is a lack of research methods that incorporate multiple factors and levels. This study aimed to employ eye-tracking technology and RSA to quantify and evaluate the comprehensive impacts of different garden elements (such as the greenery ratio, architectural style, water saturation, and stone design) on people’s visual attention and emotional experiences. Ultimately, this research provides scientific evidence for optimizing landscape designs.

Here, we aimed to optimize the design of gardens by leveraging the advantages of the response surface methodology and a multifactorial analysis. Using tourists’ focal points as the dependent variable, and the vegetation, architecture, and water as the independent variables, we employed eye-tracking technology to determine the optimal values of independent variables that resulted in the optimal value of the dependent variable. This approach allowed us to achieve the goal of optimizing the garden design while significantly reducing the design costs [39]. Therefore, we employed response surface analysis based on eye-tracking data. This method allowed for analysis of the relationship between multiple factors and multiple levels using statistical modeling, resulting in the determination of the optimal values for each element. In contrast to a single-factor analysis, response surface analysis takes into consideration the relationship between multiple garden elements, making the research results more comprehensive and accurate. The integration of eye tracking and response surface analysis presents significant advantages in the field of landscape design. Firstly, it can help designers understand the extent of the impact that different elements have on people’s attention, preferences, and comfort, enabling targeted adjustments during the design process. Secondly, by considering the relationship between different elements, this method can facilitate overall coherence and harmony in landscape design, enhancing people’s experience. Additionally, eye tracking and response surface analysis can be widely applied in landscape design, urban planning, and public space transformations. Through an in-depth examination of people’s behavior and perceptions in different environments, this method can provide scientific evidence for creating more attractive and comfortable surroundings, thereby improving quality of life for urban residents.

2. Materials and Methods

2.1. Study Area

This study focused on six selected gardens that served as key representatives of the Jiangnan style (Surging Wave Pavilion, Lion Grove Garden, Humble Administrator’s Garden, Lingering Garden, Jichang Garden, and Master-of-Nets Garden; Figure 1). The criteria and rationale behind the selection of these six gardens were as follows [40,41,42]:

(1)

Representativeness: These gardens hold significant historical and cultural value in the Jiangnan region, representing the typical style and design concepts of Jiangnan gardens. Specifically, they showcase the essence of Jiangnan gardens in terms of their layout, architecture, and plant arrangements.

(2)

Geographical coverage: The chosen gardens are distributed across different cities in the Jiangnan region, including Suzhou and Wuxi, reflecting their diversity and regional characteristics.

(3)

garden elements: The presence of pavilions, towers, small bridges, flowing water, and rock formations within these gardens fully embodies the artistic features and aesthetic pursuits of the Jiangnan garden design.

Through literature review and field investigations, we examined the characteristics of these garden elements across the six chosen gardens (Table S1).

2.2. Photo Acquisition and Processing

The use of photographic images as substitutes for real landscape scenes has been shown to be reliable and has been widely adopted in landscape perception research [43,44,45,46,47]. Therefore, it is feasible to conduct experiments using photographs as stimuli. Three types of garden element images representing different landscape types—greenery, buildings, and water features—were used as stimuli. These elements were selected because they are common in Jiangnan gardens. Representative images were chosen for each type of landscape based on the similarity of the landscape structures in the images, the complexity of the garden settings, and the feasibility of modifying content, eliminating other covariates, and controlling landscape complexity by removing or adding certain landscape components. All the photographs were taken under similar weather and seasonal conditions, including sunny weather and lower traffic flow. Considerations included mounting the camera on a tripod at a fixed height and capturing photos at a resolution of 2970 × 1980 pixels. A focal length of 50 mm was maintained to ensure a consistent field of view (+31 × 21°). All images were selected from a photo library, with a focus on three key garden elements: the greenery ratio, the number of buildings, and water saturation. Collage techniques were used to manipulate the scenes and create well-integrated images by adding, removing, or synthesizing landscape elements [24,48,49,50,51]. All added landscape elements referenced actual garden photos with similar landscape structures to enhance their realism [42]. This resulted in 18 images (Figure S1), with 6 images per setting type, used to compare the differences in preference ratings and eye movement metrics across different levels of landscape complexity in each environment.

2.3. Subject

We recruited participants from diverse age groups and genders by sending requests. The criteria for eligibility included normal vision and color perception, a willingness to participate, and the absence of any assistance. The exclusion criteria included excessive eye blinking, a prolonged dwelling time, and erratic eye-tracking trajectories. The final sample consisted of 94 individuals (44 males and 50 females) between the ages of 18 and 65. Before their enrollment, the participants were provided with concise instructions about the testing procedures, while the specific research objectives remained undisclosed. Additionally, the participants were instructed to avoid using eye makeup and mascara to prevent potential interference with accurate pupil tracking by the eye-tracking glasses.

2.4. Eye Tracking and Questionnaire Data Collection

Research has shown that attractive features lead to longer gaze durations and higher fixation rates [52,53] (Table S2). Therefore, it was imperative to examine the differences in the fixation counts and dwelling times when perceiving variations in the complexity of different elements to gain a better understanding of individual differences within the general population in order to inform the future development of gardens [54,55].

Eye-tracking data were analyzed using the ETG2 Wireless Analysis Pro (18009771) software (SMI, Germany). This software utilizes infrared eye-tracking technology [56], enabling the accurate recording of both reflective signals and pupil positions. As a result, all the fixations (fixation count) and saccades (scan path) were recorded [57]. In addition, the SMI analysis software BeGaze 3.7, which facilitated the export of the eye-tracking metrics (ETMs) into well-structured Excel files, was employed.

2.5. Procedure

To analyze the visual preferences and differences in various elements in Jiangnan gardens, we conducted a specific scene-based visual and psychological cognition experiment. The participants experienced the same location continuously for five minutes at a time, and their experiences were recorded and analyzed through a questionnaire survey (Table S3). Using qualitative methods, our aim was to reveal the impact of garden elements on psychological responses and validate the relationship between garden elements and psychological reactions.

The eye-tracking landscape optimization method was used to carry out the experiments and procedures (Figure 2). The eye-tracking experiments were performed using 18 garden photographs and the participants were allowed to freely view and control the observation duration. To address potential biases introduced by a fixed order, the presentation sequence of the photographs was randomized, creating a relaxed experimental environment [50,58]. The space was carefully controlled to ensure the strict regulation of lighting and ambient noise. Prior to testing, all the participants received identical instructions (training and experimental operation guidelines), and three calibration points were utilized to ensure accurate eye-tracking measurements across the entire screen. During the experiment, the participants observed three sets of different garden elements that were recorded by an independent eye-tracking device in conjunction with a computer. After the experiment, the participants were asked to complete questionnaires (Table S4) and received corresponding rewards.

After completing the experiment, the eye-tracking data were analyzed using the BeGaze 3.7 software package (SMI). The raw data obtained from the eye tracker were transformed to derive a meaningful measure, the fixation count. Subsequently, response surface analysis was performed using the Design-Expert 8.0.5b software (Design-Expert, Minneapolis, MN, USA). The independent variables were the greenery ratio, number of buildings, and water saturation. The response surface analysis employed a Box–Behnken central composite experimental design with three factors and three levels. To validate the reliability of the experiment, another eye-tracking experiment was conducted, ultimately leading to the obtained results.

2.6. Data Analysis

Statistical analysis was conducted using SPSS to determine whether there are significant differences in preference ratings between different garden elements and visitors. A bivariate correlation analysis was performed to examine the relationships between three factors—greenery ratio, number of buildings, and water saturation—and the fixed count data. This analysis quantified the impact of these variables on different indicators and assessed the normality of data distribution. The factors included in this analysis were the” number of buildings” (rated on a six-point scale), “water saturation” (rated on a six-point scale), and “greenery ratio” (rated on a six-point scale). Origin and heatmap analyses were employed to investigate differences in visual behavior and cognitive assessment when observing the three park elements based on an assumption of normal data distribution (Figure S2). Furthermore, response surface methodology was used to evaluate the overall trend and relationships between variables in the dataset, as well as the relationships between park elements. Finally, quadratic regression and analysis of variance were applied to test the interactions between the three factors.

3. Results

3.1. Research on Environmental Psychology of Garden Elements

3.1.1. Impact of the Greenery Ratio on Visitors’ Psychology

Landscaping elements primarily comprise the greenery ratio, number of buildings, and water saturation. In order to examine the impact of these elements on visitors’ environmental psychology using field research with three-dimensional scenes (Figure 3a,b), we compared the psychological experiences of visitors under scenarios of low and high levels of vegetation. The findings revealed a significant influence of vegetation on visitors’ psychology and behavior (Table S5).

When the greenery ratio in a garden was relatively low, the absence of greenery created a monotonous and oppressive atmosphere that failed to capture the interest and attention of visitors. In contrast, scenes with a higher greenery ratio were perceived as more beautiful and pleasant. An increase in the greenery ratio added vitality to the environment, thereby enhancing visitors’ level of enjoyment. However, an excessive density of greenery often resulted in feelings of fear, similar to findings from studies on people’s fear of densely vegetated areas [59]. The greenery ratio plays a significant role in environmental psychology, and an appropriate increase in the greenery ratio is often associated with positive emotions [24,60]. This suggests that visitors’ perception, cognition, and emotional responses to their environment are directly influenced by the surrounding landscape (greenery), as observed in studies on environmental psychology.

3.1.2. Impact of the Number of Buildings on Visitors’ Psychology

By conducting field research using three-dimensional scenes (Figure 4a,b) and comparing the psychological experiences of visitors under scenarios with fewer and more buildings, it was observed that the number of buildings had a profound impact on visitors’ environmental psychology (Table S5).

Scenarios with fewer buildings, which were characterized by a low level of architectural detail and a lack of complexity in the building facades, were considered compact and devoid of variation, inducing a sense of urgency. In contrast, scenes with an excessive number of buildings led to feelings of anxiety and suppression. The presence of numerous buildings contributes to a crowded and chaotic environment, evoking a sense of oppression and unease [61,62]. Psychological studies have shown that a single architectural scene can be considered relatively dull and lonely; without the accompaniment of other buildings, such a scene lacks diversity and vitality, imparting a feeling of isolation and monotony. Inversely, scenes with a multitude of diverse buildings can be seen as oppressive and noisy, failing to capture the interest and favor of visitors [63]. In summary, the number of buildings plays a significant role in environmental psychology. A suitable increase in the number of buildings can enhance subjective comfort and sense of security [64]. Thus, it is necessary to design the interval between buildings reasonably.

3.1.3. Impact of Water Saturation on Visitors’ Psychology

In the field of color theory, saturation, also known as purity, refers to the intensity or brightness of a color. In the hue–saturation–value (HSV) color model, saturation is one of three attributes of color, alongside hue and value. In this model, saturation ranges from 0 to 100%. As saturation decreases, the color becomes duller, eventually losing its hue and becoming achromatic, with a saturation value of 0. In our analysis of numerous photographs showcasing garden water features, the highest saturation level frequency identified was 50%, which we established as the standard (Figure S3). Experimentally, we observed that when the saturation of this photograph was adjusted to 100%, the colors approached full saturation. Therefore, the definition of 50% saturation aligns with our requirements. Subsequently, based on the regular patterns observed in the single-factor experiments, we assigned saturation levels of 20%, 30%, 40%, 50%, 60%, and 70% to the water feature photographs.

Through field research using three-dimensional scenes, the impact of water saturation on the environmental psychology of visitors was compared (Figure 5a,b). It was found that changes in water saturation elicited different psychological experiences and behavioral responses (Table S5). Low water saturation (dark-colored water) evoked a sense of oppression. Black is typically perceived as a negative and heavy color, and thus, water with a dark hue may induce feelings of sadness or repression. This color may trigger negative emotional responses, such as anxiety or depression, as it may evoke associations with pollution or a dirty and disorderly environment [65]. In contrast, high water saturation (yellow-colored water) created a feeling of nausea [66]. Yellow is often associated with dirt or impurities; therefore, water with a yellow hue may cause discomfort and aversion among visitors [67]. This color can evoke associations with pollution or the presence of harmful substances and may even elicit feelings of nausea or unease. Water saturation significantly influences the environmental psychology of individuals; thus, in landscape design, it is crucial to consider the impact of water saturation on visitors’ emotions and mental states in order to create a more comfortable and pleasant environment.

Based on the analysis of the questionnaire, we validated and discussed the relationship between garden elements and psychological responses. The experimental data clearly demonstrated that varying vegetation proportions, building quantities, and water saturation levels had differential impacts on the visual attention and psychological perceptions of the participants. These findings are in line with existing theories in landscape psychology [68,69,70,71].

3.2. Eye-Tracking Analysis of Elements of Garden Landscapes

3.2.1. Eye-Tracking Analysis of Greenery Ratio

The influence of the greenery ratio on attention distribution and visual preferences was assessed using eye-tracking analysis. The eye-tracking device recorded the visitors’ eye movements, and heatmaps were generated to illustrate the influence of different greenery ratios on eye behavior. It was observed that in environments with a low greenery ratio (Figure 6a), the heatmaps exhibited scattered and sparse features, indicating that the visitors’ visual attention was not concentrated on specific areas. The visitors tended to fixate more frequently on the empty sky or buildings, as the lack of greenery failed to capture their attention. Conversely, in scenes with high greenery ratios (Figure 6b), the clear concentration of the fixation counts in areas with sparse greenery was observed, forming distinct hotspots. The visitors tended to focus on areas not covered by vegetation, as these areas presented simpler features that attracted their visual attention, accompanied by a sense of security and comfort [72,73]. However, when the visitors concentrated on highly vegetated areas, a feeling of fear emerged, driving them to seek escape.

Therefore, an appropriate greenery ratio can guide visitors’ eye tracking and enhance their overall cognition and preferences toward the landscape. Excessive or insufficient greenery may lead to visual fatigue or monotony, thereby affecting visitors’ perception and emotional experiences in response to their environment.

3.2.2. Eye-Tracking Analysis of The Number of Buildings

Using eye-tracking technology, this study investigated the influence of the number of buildings on visitors’ eye movement behavior, as well as their attention distribution and visual preferences. The results showed that the number of buildings significantly influenced participants’ eye movement patterns. On one hand, in scenes with fewer buildings (Figure 7a), the heatmaps exhibited concentrated features, indicating that the visitors’ visual attention was predominantly focused on specific areas. The visitors tended to frequently fixate on the few available buildings, but these individual buildings struggled to sustain their attention for long periods. Consequently, while the fixation counts were concentrated, they also displayed fast scanning or transient dwelling behavior, reflecting a state of uneasiness or tension. On the other hand, in scenes with an excessive number of buildings (Figure 7b), the heatmaps displayed a more concentrated eye movement trajectory around the areas where the buildings were located, forming distinct hotspots. The visitors tended to prioritize the structures and details of each individual building in these areas because they presented richer and more visually appealing features.

However, when the visitors concentrated on areas with too many buildings, they might have experienced a sense of suppression or anxiety [61,62]. Therefore, an appropriate number of buildings is crucial for diverting visitors’ attention distribution and guiding them toward a more pleasant emotional and psychological experience.

3.2.3. Eye-Tracking Analysis of Water Saturation

We conducted an eye-tracking study to examine the effects of water saturation on attention distribution and visual preferences. The results of the eye-tracking study demonstrated that water saturation significantly influenced the visitors’ eye-tracking behavior. In scenes with low water saturation (presenting dark hues) (Figure 8a), the heatmaps exhibited concentrated fixation counts on the water’s surface, forming distinct hotspots. Conversely, in scenes with high water saturation (presenting yellow hues) (Figure 8b), the heatmaps showed relatively dispersed fixation counts without clear concentration on the water’s surface. Further analysis revealed that, on dark water surfaces, the participants’ fixation counts were primarily centered on or above the water’s surface, possibly due to their discomfort or unease regarding the color of the water. Therefore, they attempted to avoid direct fixation on the water’s surface. On the yellow water surface, the visitors’ fixation counts tended to cluster more around the edges or underwater areas, potentially due to feelings of nausea or aversion towards the color of the water; thus, they avoided direct fixation on the water’s surface. Additionally, on the yellow water surface, the visitors’ fixation counts were more likely to demonstrate rapid scanning or transient dwelling behavior, as the visitors tried to quickly avert their eyes away from the uncomfortable visual stimuli as they were unwilling to remain fixated on the water’s surface for an extended period.

Collectively, the saturation level of water significantly influenced visitors’ eye movement behavior by greatly impacting the distribution of fixations. Optimal water saturation can effectively redirect visitors’ visual attention distribution, thereby guiding them toward a more enhanced emotional and psychological experience.

3.3. The Relationship between Environmental Psychology and Eye Tracking

Through the analysis of environmental psychology and eye tracking on the perception and psychology of visitors towards garden elements, it was found that the eye-tracking analysis not only encompassed the content of the environmental psychology analysis but also provided more intuitive and objective support through eye movement data. For example, in the eye movement recorded in response to two-dimensional images, we clearly observed the eye movement behavior of visitors when observing scenes with low or high degrees of vegetation. These results reflected the variations in visitors’ psychology in response to garden elements. For instance, in the study on the number of buildings, when visitors concentrated on a specific area with a single building, it induced a sense of depression or anxiety, which aligned with the results of the psychological analysis. Similarly, the psychological results of the saturation level of water were also highly consistent between the two studies.

Therefore, eye-tracking analysis can be used as a substitute for environmental psychology when conducting landscape analysis within the realms of three-dimensional analysis and psychological analysis. Not only does this allow for the study of visitors’ psychological changes in a sensory manner, but it also provides rational and intuitive eye-tracking research data for two-dimensional images. This enables a more accurate analysis and a better understanding of visitors’ perceptions and responses to garden elements, thus providing a scientific basis for the design and optimization of such elements.

3.4. Single-Factor Experiments

Using an eye-tracking device, our research examined the impact of different elements of landscaping on the psychology of visitors. This study showed that the greenery ratio, number of buildings, and water saturation significantly affected visitors’ visual fixation and psychological perceptions. However, a simple analysis using a real-world situation with an eye-tracking device cannot fully capture the influence of individual elements on visitors’ psychology. Due to the inability to accurately measure the variations in each element in the actual three-dimensional space, the research results of a single element are heavily influenced by other landscaping elements, making it difficult to precisely determine the range of each element that satisfies visitors. To better investigate the impact of individual landscaping elements on visitors’ psychology and effectively minimize the interference of other elements on visitors’ preferences, we adopted single-factor experiments to determine the optimal range of landscaping elements.

3.4.1. The Impact of the Greenery Ratio on Visitors

As shown in Figure 9a, a single-factor experiment revealed that the visual score (fixation counts) for tourists exhibited an increasing–decreasing trend with an increase in the greenery ratio (Table S6). When the greenery ratio reached 60%, the visual fixation count of visitors reached its highest value, indicating that appropriately increasing the greenery ratio to replace other elements in the garden can make the space more natural and vibrant, thus increasing visitors’ satisfaction [74,75]. This finding is consistent with the results of the eye-tracking analysis, which suggests that a certain level of greenery significantly enhances visitors’ sense of security and comfort (Figure 9b). When the greenery ratio exceeded 70%, the visual fixation counts of visitors sharply declined, in line with the results of the eye-tracking analysis. This is because a large area of green space interferes with visitors’ visual scanning and leads to feelings of fear and disgust. At the visual level, it has been observed that the hierarchy and focal points within a landscape guide the attention and gaze of visitors through the use of elements such as shape and color. A 60% ratio of greenery can potentially create the most optimal visual contrast and sense of hierarchy, thereby enhancing the attractiveness and interest of the landscape. Therefore, appropriately controlling the greenery ratio can optimize visitors’ experience in the garden, with an optimal greenery ratio of around “60%”.

3.4.2. The Impact of the Numbers of Buildings on Visitors

As depicted in Figure 10a, a single-factor experiment revealed a trend of increasing and then decreasing visual fixation counts with an increase in the number of buildings (Table S7). The optimal state of visitors’ visual fixation counts was observed when the number of buildings was three. In the case of a lower number of buildings, buildings served as decorative elements between the greenery and water saturation, providing immediate architectural information to visitors [70,76]. A moderate increase in the number of buildings enhanced the diversity of architectural styles in visitors’ perspectives, thereby attracting their visual attention and improving their garden experience (Figure 10b). Nonetheless, an excessive number of buildings may blur the boundaries between garden elements (the greenery ratio, number of buildings, and water saturation), causing a cognitive burden and triggering feelings of depression or anxiety among visitors [77,78]. Therefore, an appropriate number of buildings can provide a positive aesthetic experience, with the optimal number being around “three”.

3.4.3. The impact of water saturation on visitors

As shown in Figure 11a, the results of the single-factor experiment revealed a trend whereby the visitors’ fixation counts initially increased and then decreased as the saturation level of water increased (Table S8). The highest level of visual attention from visitors was observed when the water saturation level approached zero. Water saturation is considered a crucial factor in enhancing the aesthetic appeal of the landscape as it significantly contributes to the overall visual experience [71]. Assuming that the greenery ratio and the number of buildings remain constant, a water saturation level of zero (green water) enhanced the spatial atmosphere of the landscape. Gradually increasing the saturation level of water better integrated it into the garden environment, thereby improving visitors’ sense of comfort (Figure 11b). However, excessive water saturation (yellow water) led to a gradual decrease in visitors’ fixation counts, potentially due to the intense visual stimulation caused by the overly bright yellow water surface, which discourages a prolonged gaze. This phenomenon demonstrates the mechanism of a color–emotion response. Color is a significant component in landscape design, as different colors elicit distinct emotional reactions. For instance, green evokes feelings of tranquility and relaxation, while red can elicit excitement and arousal. Consequently, the color of water can also influence an individual’s psychological and physiological state, leading to stress reduction or enhancing happiness [79]. Therefore, an appropriate water color provides a positive visual experience, with the optimal saturation level at approximately “60”.

3.4.4. Validation of the Results of Single-Factor Experiments

We re-evaluated novel approaches for collecting and analyzing eye-tracking data to explore the changes in participants’ attention in comparison to real observations. This was achieved by optimizing real-life scenarios and varying the placement of elements in multiple scenes while conducting multiple experiments and reproducing the scenes. For instance, we investigated the influence of pavilion positioning on eye-tracking heatmaps of visitors and found that the pavilion’s placement slightly impacted the distribution of participants’ visual attention, although the overall effect was minor (Table S9). Hence, the distribution of participants’ attention in the eye-tracking heatmaps was convincing compared to actual circumstances.

Additionally, we conducted a questionnaire study to further validate the reliability of the findings. This research revealed that the results of the single-factor experiment aligned with the trends observed in the questionnaire survey. Therefore, the conclusions drawn from the single-factor experiment regarding the garden elements can serve as experimental parameters for subsequent response surface experiments (Table S10).

3.4.5. Mechanisms and Principles

The potential health benefits of exposure to novel natural landscapes may be better understood via the lens of the supportive environment theory (SET). Landscapes that are simple to comprehend and maintain are referred to as supportive environments. According to the SET, people require these kinds of surroundings—garden components—in order to preserve both physical and mental health [80].

Like all natural landscapes, natural landscapes devoid of plants, rocks (buildings), and water have various limitations and potential hazards despite to their many potential health advantages [81]. Thus, in order to effectively plan and create natural landscapes, we need to investigate the relationships between garden elements in more detail.

3.5. Response Surface Analysis of Garden Elements

3.5.1. Response Surface Optimization Simulation

After establishing the optimal range for each element of a garden, further investigation was conducted using response surface analysis to explore the interrelationships among these elements to determine the optimal value for each garden element. The specific research process is as follows: The experimental factors and levels are shown in

Table 1. Factors and levels of response surface experience.

Number	Greenery Ratio (%)	B Number of Buildings	C Water Saturation (%)
−1	50	2	40
0	60	3	50
1	70	4	60

, while

Table 2. Design and results of response surface experiments.

Number	A Greenery Ratio (%)	B Number of Buildings	C Water Saturation (%)	Combination (Fixation Count)
1	−1 (50%)	−1 (2)	0 (50%)	14
2	1 (70%)	−1 (2)	0 (50%)	13.3
3	−1 (50%)	1 (4)	0 (50%)	13.5
4	1 (70%)	1 (4)	0 (50%)	12.8
5	−1 (50%)	0 (3)	−1 (40%)	15.1
6	1 (70%)	0 (3)	−1 (40%)	14.5
7	−1 (50%)	0 (3)	1 (60%)	14.8
8	1 (70%)	0 (3)	1 (60%)	14.3
9	0 (60%)	−1 (2)	−1 (40%)	13.2
10	0 (60%)	1 (4)	−1 (40%)	12.8
11	0 (60%)	−1 (2)	1 (60%)	13
12	0 (60%)	1 (4)	1 (60%)	12.6
13	0 (60%)	0 (3)	0 (50%)	17.9
14	0 (60%)	0 (3)	0 (50%)	17.8
15	0 (60%)	0 (3)	0 (50%)	17.7
16	0 (60%)	0 (3)	0 (50%)	17.9
17	0 (60%)	0 (3)	0 (50%)	17.9

, Tables S11 and S12 contain the design plans and analysis of the results.

Multiple quadratic regression and analysis of variance were applied to the data presented in Table 2. These analyses yielded a quadratic regression equation (Equation (1)):

Y = 11.66 − 2.22A + 1.06B − 0.84C − 0.73AB − 0.27AC − 0.05BC + 1.79A² − 2.52B² + 3.37C²

(1)

To analyze the effectiveness of the quadratic regression equation, further variance analysis was conducted on the regression model.

Table 3. Analysis of variance of the response surface model.

Sources of Variance	Sum of Squares	Degrees of Freedom	Mean Square	F-Value	p-Value
Model	69.71	9	7.75	1095.30	<0.0001 **
A	0.78	1	0.78	110.48	<0.0001 **
B	0.41	1	0.41	57.27	<0.0001 **
C	0.10	1	0.10	14.32	0.0069
AB	0.000	1	0.000	0.000	1.0000
AC	2.500 × 10−3	1	2.500 × 10−3	0.35	0.5708
BC	0.000	1	0.000	0.000	1.0000
A2	7.48	1	7.48	1057.22	<0.0001 **
B2	40.66	1	40.66	5749.78	<0.0001 **
C2	14.14	1	14.14	1999.48	<0.0001 **
Residuals	0.049	7	7.071 × 10−3
Lack-of-fit terms	0.017	3	5.833 × 10−3	0.73	0.5860
Pure error	0.032	4	8.000 × 10−3
Total variation	69.76	16

The ’**’ signifies highly significant differences (p < 0.01).

shows that the garden model was highly significant (p < 0.01). The lack-of-fit terms (AB, AC, and BC) for all three models were not significant (p > 0.05), indicating that the influence of non-experimental factors on the results was minimal and the models were reasonable and acceptable. The coefficient of determination (R²) for the particle size distribution was 0.9984, denoting a strong linear relationship between the factors and visual attractiveness to visitors. The adjusted coefficient of determination (R² = 0.9953) indicated that the model explained 99.53% of the variation in the response variable. The experimental coefficient of variation (CV) was 0.56% (<10%), and the signal-to-noise ratio was 81.827 (>4). In conclusion, this model can be used for predicting and simulating visitors’ psychological preferences toward garden elements. Furthermore, it was found that variables A, B, and C had extremely significant effects on Y (p < 0.01). The interaction terms AB, AC, and BC did not significantly impact Y (p > 0.05), while the quadratic terms A², B², and C² had highly significant effects on Y (p < 0.01). Further analysis based on the F-value revealed that the order of influence in the garden was A > B > C, suggesting that vegetation coverage had the greatest impact, followed by the number of buildings and then water saturation.

3.5.2. Analysis of the Relationships between Garden Elements

Moreover, to enhance the design of garden elements, a further analysis was conducted to investigate the synergistic effects between different elements in the courtyard.

• The relationship between the greenery ratio and the number of buildings

The relationship between the greenery ratio and the number of buildings exhibited a steep trend in the response surface model, indicating significant impacts of the greenery ratio and the number of buildings on visitors’ psychological preferences. Additionally, there existed a certain relationship between them (Figure 12a,b).

Research has revealed a coupled relationship between the greenery ratio and the number of buildings, where a higher greenery ratio generally implies greater vegetation coverage, consequently imposing potential constraints on the quantity and layout of buildings. In environments with a high greenery ratio, designers may be inclined to select a smaller number of buildings with distinctive features to avoid excessive interference with the visual effect of vegetation and maintain a natural sense of the green environment. Simultaneously, the density and layout of buildings can influence people’s perception of the greenery ratio. Excessive numbers of buildings may reduce the visibility and perception of green spaces, making the greenery ratio appear lower. Conversely, a moderate layout of buildings can highlight green spaces and enhance the perception of the greenery ratio. Additionally, a building’s design and height also affect the perception of the surrounding greenery. For instance, tall buildings may obstruct sunlight and hinder plant growth, thereby impacting the actual effect of the greenery ratio [82].

2.

The relationship between the greenery ratio and water saturation.

The relationship between the greenery ratio and water saturation exhibited an elliptical trend in the response surface model plot, indicating significant influences of both factors on visitors’ psychological preferences. However, the relationship was relatively weak (Figure 13a,b).

By further analyzing the correlation between the greenery ratio and water saturation, it was found that the impacts of the greenery ratio and water saturation on environmental landscapes are relatively independent. In the absence of alterations to water and vegetation colors, this implies that increasing or decreasing the greenery ratio does not significantly affect the water saturation level and vice versa. Therefore, in this study, which focused on investigating plant color without any alterations, the greenery ratio and water saturation in the environment can be adjusted independently without considering their relationship [83].

3.

Relationship between the number of buildings and water saturation

The relationship between the number of buildings and water saturation exhibited a steep trend in the contour plot, indicating a significant impact of both factors on visitors’ psychological preferences. Moreover, there existed a clear mutual influence between the number of buildings and water saturation (Figure 14a,b).

Analysis of the correlation between the greenery ratio and the number of buildings showed that an increase or decrease in the number of buildings may change the overall sense of balance of the environmental landscape. For example, too many buildings may suppress the visual effect of water saturation and make it appear less prominent, while a moderate number of buildings may better emphasize water saturation and enhance the visual appeal of the landscape.

Additionally, changes in the number of buildings may influence a visitor’s perception of space. A lower number of buildings can create a more open and expansive spatial experience, with water saturation being more prominent within it. Inversely, a higher number of buildings can lead to a crowded and congested perception of space, diminishing the significance of water saturation. Hence, during the planning and design process, holistic consideration should be given to the layout, height, and form of buildings, as well as the positioning, size, and characteristics of water saturations, to achieve an optimal combination of the number of buildings and water saturation [79].

3.6. Optimizing the Design of Garden Elements

Thus, by analyzing the regression model for gardens, we obtained the following optimal conditions for garden elements: a greenery ratio of “58.82%”, number of buildings of “3”, and water saturation level of “47.2%”. Under these conditions, the predicted visual attractiveness of the buildings was “17.86”. We conducted three independent parallel experiments to validate the effectiveness and reliability of the model (Table S13). The experimental results indicated that the garden’s visual attractiveness to tourists was rated as “18”. The deviation of the experimental results from the predicted value was only “0.22%”, suggesting that the model demonstrated a strong predictive capacity to capture the psychological preferences of garden tourists.

4. Discussion

4.1. The Integration of Landscape Psychology and Environmental Psychology

The field of environmental psychology involves extensively examining individuals’ perception, cognition, and emotional responses to environmental landscapes, with a primary focus on the interplay between humans and their surrounding environment [84,85]. In the realm of horticulture, landscape psychology investigates individuals’ psychological reactions, such as their attention, preferences, and comfort, towards various garden elements [86,87,88]. Adopting this perspective allows for a deeper understanding of the relationship between individuals and landscape features and enhances our comprehension of the dynamic relationship between humans and the environment. Our research findings support this theory, demonstrating the significant influence of garden elements such as the vegetation ratio, architectural layout, and water color on individuals’ psychological and emotional responses.

4.2. The Impact of Different Garden Elements on Visitors’ Psychology

The greenery ratio, number of buildings, and water saturation in garden landscapes have multifaceted effects on visitors’ psychological experiences. Firstly, a high greenery ratio generally creates a more pleasant and vibrant environment, but excessive greenery may also induce feelings of oppression and unease. Therefore, striking a balance between visual pleasure and spatial comfort is necessary when designing green areas [24,60]. Secondly, the number of buildings also significantly influences visitors’ psychology. Fewer buildings may result in a monotonous environment, while an excess of buildings can lead to visual confusion and a sense of oppressiveness. Consequently, architectural designs need to consider the diversity and openness of spaces to avoid a singular or overcrowded layout [63,64]. Lastly, water saturation has a noticeable impact on emotions. Low saturation of dark-colored water often triggers feelings of depression, whereas high saturation of yellow water may induce aversion and discomfort. This highlights the importance of selecting appropriate colors and transparency levels when designing water features [65]. Collectively, landscape design needs to take these factors into account in order to create an environment that is both aesthetically pleasing and meets the psychological needs of visitors.

4.3. The Impact of Different Garden Elements on Visitors’ Visual Preferences

The greenery ratio, number of buildings, and water saturation in landscapes have complex and far-reaching effects on visitors’ visual preferences. First of all, a greenery ratio of about 60% can most effectively attract visitors’ attention and increase their visual interest and pleasure; however, too high a greenery ratio can lead to visual fatigue and a sense of oppression, indicating the need to balance the relationship between visual comfort and natural elements in design [72,73]. Furthermore, an appropriate number of buildings not only adds visual focus and diversity to the landscape but also enhances the overall aesthetic experience of visitors; however, too many buildings may cause visual confusion and depression, suggesting that the number of buildings needs to be carefully controlled during design to avoid visual overload [61,62]. Finally, moderate control of water saturation can effectively enhance visual experience and environmental aesthetics, while excessive saturation may cause discomfort and aversion due to strong visual stimulation. The balance and coordination between these elements are crucial, indicating that landscape design needs to comprehensively consider the visual and psychological needs of visitors in order to create both aesthetic and comfortable garden environments.

4.4. The Advantages of Integrated Eye-Tracking Response Surface Analysis

The integrated eye-tracking response surface analysis demonstrates significant advantages in landscape design. Primarily, this approach investigates the complex relationships between different topographic garden elements and visitor emotional preferences through multi-factor and multi-level analysis, effectively guiding topography design and planning. Moreover, by evaluating visitor preferences for garden elements and considering their coupling effects, optimal ranges for multiple elements are established, enhancing the comprehensiveness and scientific basis of the design. In addition, the combination of comprehensive analysis methods, eye-tracking technology, and psychological theories enables the analysis of visitor behaviors and preferences using big data, providing a scientific basis for landscape design that aligns with visitors’ visual preferences and psychological needs. Importantly, this approach accurately reveals the interaction and influence mechanisms between different elements. Compared to traditional single-factor qualitative analysis, it enhances the accuracy of design decisions, reduces design costs, and facilitates more effective and economic landscape design optimization.

4.5. Directions for Future Research

Although our study preliminarily demonstrated the potential application of eye tracking and response surface analysis in landscape design, there are still some limitations that require further investigation. For example, the collection and analysis of eye-tracking data may have been influenced by external factors such as changes in lighting conditions and fluctuations in pedestrian flow. Future research could explore the impacts of these factors and consider other variables that may affect eye-tracking data. In addition, we plan to expand the scope of our study to encompass a wider range of garden elements and environmental conditions, allowing for further exploration of applications in landscape ecology, urban planning, biodiversity conservation, and health promotion.

5. Conclusions

Herein we propose an eye-tracking-based response surface analysis method for the design and optimization of gardens with the aim of enhancing visitors’ comfort and satisfaction. Firstly, we performed a study of different garden elements using landscape psychology and identified that variations in certain factors, such as the greenery ratio, number of buildings, and water saturation, have an impact on visitor psychology. We also found that eye-tracking analysis was able to accurately capture visitors’ psychological activities and provide intuitive eye-tracking data. Subsequently, through single-factor experiments and heatmap analysis, we established the optimal ranges for each garden element. The optimal values for the greenery ratio, number of buildings, and water saturation were around “60%” (greenery ratio), “3” (number of buildings), and “50%” (water saturation), respectively. Finally, we conducted the first-ever response surface analysis to investigate the relationships between different garden elements. The results revealed a significant relationship between the greenery ratio and the number of buildings, as well as between the number of buildings and water saturation, while the relationship between the greenery ratio and water saturation was relatively low. Among these elements, the greenery ratio was the primary factor influencing visitors’ garden experiences, followed by the number of buildings and the level of water saturation, which aligned with actual visitor experiences. The optimized values of greenery ratio, number of buildings, and water saturation were determined, providing guidance for the optimization of landscape design.

The eye-tracking response surface analysis optimization method integrates visual perception analysis, multi-factor multi-level design, and nonlinear data processing, offering broad prospects for diverse landscape designs. This method provides a novel approach to multi-factor, multi-level landscape design that incorporates visual, sensory, and psychological analyses. It is not only applicable to garden design optimization but also has potential applications in landscape design, urban planning, and public space renovation. While this study demonstrates the significant potential of combining eye tracking and response surface analysis in garden design, there are still some limitations. For instance, this study selected only greenery ratio, number of buildings, and water saturation as the main variables and did not consider other potential influencing factors. Future research will further explore the impact of these factors on eye-tracking data and expand the scope of the study to include more landscape elements and environmental conditions.