Rokuon-ji Temple Garden (commonly known as Kinkaku-ji or Golden Pavilion) was selected as the primary research site due to its exemplary status as a classical Japanese landscape garden with documented healing properties (Brown and Corry, 2011). Established in the late 14th century as a retirement villa for Shogun Ashikaga Yoshimitsu, this UNESCO World Heritage site exemplifies the sophisticated spatial principles and aesthetic ideals of traditional Japanese garden design. The garden features several key characteristics that make it particularly suitable for color analysis research: a circumambulatory layout that provides multiple viewing perspectives, a large reflective pond (Kyoko-chi) that occupies approximately 60% of the garden area, and exceptionally diverse vegetation that creates a rich chromatic palette varying with seasonal changes (Takakura, 2008).

The garden’s design incorporates all major elements of traditional Japanese landscape architecture: the shinden-zukuri style architecture of the Golden Pavilion itself, carefully positioned stone arrangements (iwakura), meticulously pruned vegetation representing both native and symbolic plant species, and the central water feature that serves as both a reflective surface and spatial organizing element (Yoshikawa, 1990). These features collectively create a complex visual environment with distinctive color characteristics worthy of detailed investigation. Additionally, the garden’s well-documented history and relatively stable maintenance practices provide a consistent research environment, mitigating variables that might otherwise complicate color analysis (Hayakawa, 2009).

Field data collection was conducted during late September 2023, specifically selected for optimal research conditions: (1) maximum color contrast between red autumn maple foliage and evergreen vegetation, (2) stable lighting conditions with minimal atmospheric interference, (3) reduced visitor density facilitating standardized photography, and (4) alignment with traditional Japanese garden peak viewing periods when chromatic compositions achieve intended aesthetic effects (Miyamoto, 2005). All photography was performed during three consecutive days (September 25–27, 2023) under clear sky conditions between 10:00 AM and 2:00 PM to maintain consistent natural lighting. The standardized equipment setup included a Canon 5D Mark III camera fitted with an EF-35 mm f/1.4 L USM lens, mounted on a carbon fiber tripod positioned at 1.6 meters height (approximating average human eye level).

Photographic documentation followed the official visitor pathway that circumnavigates the central pond, with systematic image capture at 5-meter intervals and additional photographs taken at designated viewing points (officially marked within the garden). This methodology ensured comprehensive coverage of all significant viewsheds while adhering to the intended visitor experience designed by the garden’s creators (Suzuki, 2012). At each photography position, three exposures were captured: standard exposure according to camera metering, plus one overexposed and one underexposed image (+/−1.0 EV) to ensure accurate color data across varying light conditions. All images were captured in RAW format (CR2) to preserve maximum color information for subsequent analysis.

The initial data collection yielded 150 high-resolution landscape photographs documenting the garden’s complete visitor experience. These images were processed using Adobe Lightroom with minimal adjustments (white balance correction only) to maintain color fidelity while standardizing viewing conditions for the evaluation phase.

Power analysis using G*Power 3.1.9.7 indicated a minimum sample size of 20 participants for detecting medium effect sizes (Cohen’s d = 0.5) with 80% power at α = 0.05. To identify landscape views with the strongest healing potential, a preliminary evaluation study was conducted with 23 participants recruited from the Department of Regional Environmental Sciences at Tokyo University of Agriculture. All participants completed a demographic questionnaire and color preference pre-survey to assess potential cultural biases. The sample comprised 14 female and 9 male students (age range: 20–24 years, mean age: 22.1 ± 1.3), with 87% having Japanese cultural background and 78% reporting previous experience with traditional Japanese gardens. This participant group was selected for their intermediate knowledge of landscape architecture (90% had completed at least one course in Japanese garden design), providing informed assessments while avoiding potential biases of professional designers (Nakamura, 2011). The group consisted of 14 female and 9 male students between 20 and 24 years of age, all with normal color vision as verified by Ishihara color test screening.

Evaluation sessions were conducted in a controlled environment laboratory with calibrated displays (EIZO ColorEdge CG319X, ΔE < 1.5) under standardized lighting conditions (5,000 K, CRI > 95). Each participant independently evaluated all 150 photographs on a 7-point Likert scale measuring perceived “healing quality” (where 1 = not at all healing, 7 = extremely healing). To mitigate order effects, images were presented in randomized sequences for each participant. Additionally, participants were asked to identify specific elements within each image that contributed to their rating to provide qualitative context for the quantitative evaluations (Watanabe, 2013).

From this evaluation process, 42 photographs were selected as the final research sample based on rigorous selection criteria: mean rating score ≥5.0, median rating score ≥5.0, and standard deviation <1.2 (indicating strong consensus among evaluators). The complete evaluation results for all 42 selected photographs are presented in Table 2, which shows the landscape types, primary elements, dominant colors, and shooting conditions for each high-rated sample.

The selected photographs represent diverse spatial experiences within the garden, capturing various combinations of architectural elements, vegetation, stone arrangements, and water features (Kato, 2010). This diversity ensures comprehensive analysis of color characteristics across the full range of healing landscape typologies present within Rokuon-ji Garden. Figure 3 demonstrates the variety of sample landscape photographs from Rokuon-ji Garden, showing various healing environments and their distinct color compositions.

The final photograph selection encompasses multiple viewing distances (from intimate detail views to broad panoramas), various spatial arrangements (enclosed, semi-enclosed, and open spaces), and different relationships between built and natural elements. This range of visual experiences provides a robust dataset for analyzing how color compositions contribute to healing effects across diverse landscape conditions within a single coherent garden system.

The quantitative analysis of color characteristics in Rokuon-ji Garden landscapes required development of a systematic procedure for extracting standardized color data from the selected photographic samples. Building upon methodologies established by Zhang et al. (2019), we implemented a multi-stage digital image processing protocol to isolate and quantify specific color components. The initial processing utilized Adobe Photoshop CC 2023 for color calibration against X-Rite ColorChecker Passport reference standards photographed under identical lighting conditions to ensure accurate color reproduction across all samples.

Color extraction proceeded via a binary segmentation process applied to each sample photograph, isolating six predominant color categories identified through preliminary spectral analysis and traditional Japanese landscape color theory (Sasaki, 2008). These categories—red, yellow, brown, gray, white, and green—encompass the essential chromatic elements of traditional Japanese gardens and are particularly prominent in Rokuon-ji Garden’s composition. The specific RGB value ranges for each color category and their corresponding landscape elements are detailed in Table 3.

The binary segmentation process utilized a customized MATLAB script that converted each RGB image into six separate binary masks, one for each color category (MATLAB, 2023). Each pixel in the original image was classified based on its RGB values falling within the specified ranges for each color category, with binary thresholding applied to create discrete color masks. This method enabled precise quantification of color distribution patterns, relative proportions, and spatial relationships across the landscape composition. The color extraction process is demonstrated in Figures 4, 5 which shows the transformation from original landscape photography to isolated color components.

Three primary metrics were calculated to characterize the color properties of each landscape sample: fractal dimension, diversity index, and concentration index. These metrics were selected for their demonstrated relevance to human perceptual and psychological responses to landscape composition (Nasar, 1998).

The fractal dimension (D) quantifies spatial complexity and self-similarity across scales, properties that have been associated with restorative landscape effects in numerous studies (Mandelbrot, 1982). We calculated the fractal dimension for each color component using the box-counting method, which determines how the number of boxes containing the feature of interest changes with box size. The mathematical representation is given by:

Where N ( ε ) represents the number of boxes of side length ε required to cover the color pattern. In practical implementation, the slope of the log–log plot of N ( ε ) versus ( 1 / ε ) was computed through linear regression analysis to determine D. Higher D values (approaching 2) indicate greater complexity and have been associated with enhanced cognitive engagement and restorative effects (Taylor et al., 2011).

The diversity index (H′) quantifies the evenness of color distribution using a modified Shannon–Wiener formula:

Where p i represents the proportion of image area occupied by color category i, and n represents the total number of color categories (six in this study). Higher H′ values indicate more balanced distribution of colors across the landscape view, a characteristic associated with visual harmony in traditional Japanese garden design (Shannon, 1948).

The concentration index (C) measures the degree to which specific colors are spatially clustered rather than dispersed throughout the image. This was calculated using Moran’s I spatial autocorrelation statistic:

Where N is the number of spatial units (pixels in this case), w ij is the spatial weight between locations i and j, x i and x j are color presence values (binary) at locations i and j, x _ is the mean color presence, and W is the sum of all spatial weights. Values approaching +1 indicate strong clustering of colors, while values approaching −1 indicate dispersed patterns (Moran, 1950). Intermediate values have been associated with balanced visual compositions that support focused attention without inducing visual fatigue (Kaplan, 1995).

The box counting method for fractal dimension calculation was implemented through a multi-scale grid overlay procedure. Each binary color mask was analyzed at 12 different grid scales, with box sizes ranging from 2 pixels to 2048 pixels in geometric progression (Falconer, 2003). For each grid scale, the number of boxes containing at least one pixel of the target color was counted.

The practical implementation involved several key steps: (1) converting the color-specific binary masks to boundary representations, (2) overlaying grids of progressively finer resolution, (3) counting occupied boxes at each resolution, and (4) calculating fractal dimension through linear regression analysis of the log-transformed data pairs. This approach follows established protocols in landscape structure analysis while providing specific adaptations for color pattern analysis (McGarigal and Marks, 1995).

The resulting fractal dimension values characterize the spatial complexity of each color component across multiple scales, providing quantitative measures that correlate with subjective perceptions of visual richness and engagement. These metrics enable comparative analysis of how different color distributions within the garden landscape contribute to overall healing effects through their structural complexity patterns (Purcell et al., 2001).

To assess the psychological impact of color characteristics in Rokuon-ji Garden landscapes, we employed the Semantic Differential (SD) method, which has been extensively validated in environmental psychology research for quantifying subjective perceptions of spatial environments (Osgood et al., 1957). Following the methodological framework established by Osgood et al. (1957) and refined by Kuller and Mikellides (2013), a comprehensive SD questionnaire was developed incorporating 24 bipolar adjective pairs specifically selected to evaluate healing and restorative qualities of landscape environments. The semantic pairs were systematically chosen based on previous environmental perception studies (Nasar, 1998; Stamps, 2004) and validated through pilot testing with 15 participants to ensure cultural appropriateness and semantic clarity. These semantic pairs were carefully chosen based on previous environmental perception studies and refined through pilot testing to ensure cultural appropriateness for evaluating traditional Japanese garden aesthetics.

The questionnaire utilized a five-point rating scale for each adjective pair, allowing participants to indicate their perceptual response along a continuum between opposing semantic concepts. Table 4 presents the complete set of semantic differential pairs employed in this study, organized according to their hypothesized relationship to different dimensions of environmental perception.

The semantic pairs were randomized and counterbalanced in presentation order to minimize potential biases, with approximately half of the positive terms appearing on the left side of the scale and half on the right side (Stamps, 2004). Additionally, the questionnaire included demographic items and questions regarding participants’ familiarity with Japanese gardens to control for potential influence of prior knowledge and experience.

The psychological evaluation study involved 58 participants (32 female, 26 male, age range 22–45 years, mean age 28.7) with backgrounds in landscape architecture, psychology, or Japanese cultural studies. The evaluation sessions were conducted in a controlled laboratory environment at Tokyo University, with participants seated at individual workstations equipped with calibrated 27-inch displays (Dell UltraSharp UP2720Q, calibrated to Adobe RGB color space) (Dell Inc., 2023).

Each participant evaluated all 42 selected landscape photographs from Rokuon-ji Garden, presented in individually randomized order to control for sequence effects. A maximum viewing time of 2 min per image was imposed to maintain consistent cognitive processing conditions while allowing adequate time for thoughtful evaluation. The randomization sequence was generated independently for each participant using R software (version 4.3.0). For each image, participants completed the entire 24-item SD questionnaire, with brief rest periods introduced after every 10 images to minimize fatigue effects. The average completion time for the entire evaluation session was 75 min, including instruction and rest periods.

The SD evaluation data underwent a series of statistical analyses to identify underlying perceptual dimensions and their relationship to color characteristics. First, exploratory factor analysis was performed using the principal factor method with varimax rotation to maximize the variance of squared loadings for each factor (Tabachnick and Fidell, 2019). This approach helps identify independent perceptual dimensions with minimal overlap.

Factor extraction criteria included eigenvalues greater than 1.0 (Kaiser criterion), scree plot inspection for natural breaks in eigenvalue distribution, and minimum cumulative variance explanation of 70% (Kaiser, 1960). The resulting factor structure was evaluated for interpretability based on semantic coherence of high-loading adjective pairs within each factor. Items with factor loadings less than 0.40 or with cross-loadings (loading differences less than 0.20 between factors) were excluded from factor interpretation to ensure dimensional clarity.

Hierarchical cluster analysis was subsequently performed using Ward’s minimum variance method with squared Euclidean distance as the dissimilarity measure (Ward, 1963). This approach identified groupings of landscape scenes based on similarity of their psychological evaluation profiles across the extracted perceptual dimensions. The optimal number of clusters was determined through examination of the dendrogram structure, agglomeration schedule coefficients, and silhouette width indices measuring within-cluster cohesion and between-cluster separation.

To establish relationships between psychological responses and quantitative color characteristics, correlation analyses were conducted between factor scores derived from the SD evaluations and the color metrics described in section 2.2 (Cohen, 1988). Pearson’s correlation coefficients were calculated to quantify relationships between each perceptual dimension (factor) and the corresponding color metrics (fractal dimension, diversity index, and concentration index) for each primary color category.

Multiple regression analyses were then performed using perceptual factors as dependent variables and color metrics as predictors to identify the most influential color characteristics for each psychological dimension (Field, 2018). This approach enabled quantification of how specific color attributes (e.g., complexity of green elements, diversity of red components, concentration of yellow features) contribute to different aspects of healing perception in the garden landscape.

Additionally, canonical correlation analysis was employed to examine multivariate relationships between the set of psychological factors and the set of color metrics (Thompson, 2000). This method identifies linear combinations of variables in each set that maximize correlation between the sets, revealing more complex patterns of association between color characteristics and psychological responses than would be apparent through bivariate analyses alone.

These analytical approaches collectively provide a comprehensive framework for understanding how specific quantifiable color characteristics in Rokuon-ji Garden relate to subjective perceptions of healing qualities, thereby establishing an empirical foundation for translating traditional Japanese garden design principles into contemporary therapeutic landscape applications (Bell et al., 2008).

The Semantic Differential method questionnaire data underwent principal factor analysis with varimax rotation to identify underlying perceptual dimensions in participants’ evaluations of Rokuon-ji Garden landscapes. Statistical analysis identified six primary factors with eigenvalues exceeding 1.0, collectively explaining 64.313% of the total variance in psychological responses. These six factors represent distinct perceptual dimensions that characterize visitors’ experiences of the garden’s healing qualities (Table 5).

Factor 1, labeled “Openness,” explained the largest portion of variance (13.153%) and was strongly associated with semantic pairs such as “opened-enclosed” (0.849), “restorative-draining” (0.735), “relaxing-tense” (0.723), “meditative-distracting” (0.684), and “calming-stimulating” (0.657). This factor represents the psychological dimension of spatial expansiveness and its associated restorative qualities, suggesting that perceived openness plays a crucial role in the healing experience of Rokuon-ji Garden.

Factor 2, “Decorativeness,” accounted for 11.831% of variance and was characterized by high loadings on “gorgeous-plain” (−0.730), “warm-cold” (0.723), “vibrant-dull” (0.687), and “bright-dark” (0.642). This dimension captures the ornamental and sensory richness aspects of garden perception, particularly relating to color intensity and visual warmth.

Factor 3, “Clarity,” explained 11.485% of variance and was primarily defined by “neat-squalid” (0.763) and “orderly-messy” (0.728). This factor represents the perceptual dimension of visual organization and coherence, reflecting the garden’s meticulous design principles and maintenance.

Factor 4, “Naturalness,” contributed 10.576% to total variance with high loadings on “traditional-contemporary” (0.818), “natural-artificial” (0.745), “spiritual-secular” (0.679), and “intimate-distant” (0.583). This dimension captures the authentic, traditional, and spiritual qualities of the landscape experience.

Factor 5, “Unity,” accounting for 9.040% of variance, was characterized by high loadings on “balanced-unbalanced” (0.713), “coherent-fragmented” (0.675), and “harmonious-discordant” (0.642). This factor represents the perceptual integration and compositional harmony of landscape elements.

Factor 6, “Complexity,” explained 8.230% of variance and showed high loadings on “complex-simple” (0.834), “dynamic-static” (0.767), and “mysterious-obvious” (0.613). This dimension captures the intricacy, visual richness, and cognitive engagement aspects of the garden experience.

Hierarchical cluster analysis using Ward’s method with Euclidean distance measure was applied to the factor scores of the 42 landscape samples, resulting in the identification of eight distinct landscape types within Rokuon-ji Garden. As shown in Figure 6, the dendrogram visualization reveals the hierarchical relationship and relative similarity between these landscape clusters.

Type I landscapes (n = 6) were characterized by exceptionally high scores on the Clarity factor (mean z-score: 1.32) and moderately high scores on Unity (0.78), with average scores on other dimensions. These scenes typically featured well-defined structural elements with clear visual organization, such as carefully arranged stone paths and precisely trimmed vegetation, exemplified by sample photographs No. 27 and No. 135.

Type II landscapes (n = 5) demonstrated high Complexity scores (1.45) but low Openness (−0.92). These scenes featured intricate arrangements of multiple landscape elements within visually enclosed spaces, creating intimate environments rich in detail but limited in spatial extent, as seen in samples No. 42 and No. 93.

Type III landscapes (n = 7) scored highly on both Naturalness (1.24) and Unity (0.86) factors. These views captured harmonious compositions of predominantly natural elements with minimal evidence of human intervention, yet exhibiting clear organizational principles consistent with traditional Japanese aesthetics, represented by samples No. 102 and No. 118.

Type IV landscapes (n = 4) were distinguished by exceptionally high Decorativeness scores (1.67) combined with above-average Openness (0.73). These scenes typically featured the golden pavilion reflected in the pond with vibrant seasonal foliage, creating visually striking compositions with rich color contrasts, as exemplified by samples No. 13 and No. 71.

Type V landscapes (n = 5) scored highly on Openness (1.38) and Clarity (0.92) but below average on Complexity (−0.68). These expansive views featured clean, uncluttered visual compositions with extended sight lines and minimal visual obstruction, as seen in samples No. 56 and No. 149.

Type VI landscapes (n = 4) demonstrated balanced scores across all factors, with no extreme values (all mean z-scores between −0.4 and 0.4). These scenes represented transitional spaces that incorporated multiple garden elements in balanced proportions, creating neutral but cohesive visual experiences.

Type VII landscapes (n = 6) scored highly on Complexity (0.89) and Decorativeness (0.76) but below average on Clarity (−0.73). These visually rich scenes featured multiple overlapping elements with intricate details and vibrant color compositions but lacked clear organizational structure, as exemplified by samples No. 126 and No. 142.

Type VIII landscapes (n = 5) were characterized by high Naturalness (1.15) and low Decorativeness (−1.21). These scenes emphasized subdued, monochromatic natural elements with minimal ornamental features, creating contemplative environments that evoked traditional Zen aesthetic principles.

The distribution of landscape types throughout the garden reveals a deliberate spatial sequence that alternates between contrasting perceptual experiences. This pattern suggests an intentional design strategy to create a rhythm of varying psychological states as visitors move through the garden, potentially enhancing the overall therapeutic effect through controlled contrast and complement of experiential qualities.

The identification of these eight distinct landscape types provides a framework for understanding how different spatial arrangements and visual compositions within a single garden can evoke varied psychological responses. The subsequent analysis will examine how specific color characteristics correlate with these experiential dimensions to elucidate the relationship between color properties and healing effects.

Quantitative analysis of color characteristics across the eight identified landscape types revealed distinctive patterns in the distribution, complexity, and spatial arrangement of color elements within Rokuon-ji Garden. Each landscape type exhibited a unique color signature as measured by fractal dimension (D), diversity index (H′), and concentration index (C) for the six primary colors identified in the garden composition. Table 6 presents the comprehensive quantitative metrics for each landscape type.

Type I landscapes exhibited notably high fractal dimension values for brown elements (D = 1.849), indicating complex spatial distributions of wooden architectural features, exposed soil areas, and tree trunks. This complexity was counterbalanced by relatively simple, dispersed white elements (D = 1.628, C = 0.314), creating a visual rhythm through contrast between complex and simple components. The moderate concentration of green elements (C = 0.476) with high complexity (D = 1.726) suggests carefully arranged vegetation with intricate patterns that maintain visual coherence.

Type II landscapes were distinguished by exceptionally high diversity values for brown (H′ = 0.381) and green (H′ = 0.415) elements, indicating even distribution of these colors throughout the scenes. Gray elements displayed high complexity (D = 1.650), reflecting the intricate textures of stone pathways and decorative rocks. The combination of high diversity with moderate concentration for green elements (H′ = 0.415, C = 0.296) created a balanced distribution of vegetation that avoided both monotony and visual dominance.

Type III landscapes featured the highest complexity values for green elements (D = 1.977) among all landscape types, combined with relatively low concentration (C = 0.315), creating visually rich vegetation patterns distributed throughout the scenes. These landscapes also exhibited moderately complex red elements with high concentration (D = 1.331, C = 0.628), typically manifested as focal points of autumn foliage or ornamental features against the green background.

Type IV landscapes were characterized by high concentration values across multiple color categories, particularly brown (C = 0.624), white (C = 0.517), and green (C = 0.523). This clustering of color elements created strong focal points and visual anchors within the scenes. The high complexity of white elements (D = 1.775) combined with their concentrated distribution reflected the pavilion’s gold leaf surfaces and their reflections in the water, creating visually dominant features with intricate detail.

Type V landscapes displayed exceptionally high complexity values for brown elements (D = 1.832) with relatively low concentration (C = 0.387), creating intricate patterns of wooden and soil features distributed throughout the scenes. The moderate complexity and concentration values for other colors suggested balanced compositions without strong chromatic dominance, supporting the high Openness and Clarity scores identified in the psychological evaluation.

Type VI landscapes featured a balanced combination of highly complex green elements (D = 1.867) and concentrated brown features (C = 0.562). This arrangement created a dynamic visual relationship between background vegetation and focal architectural elements, resulting in balanced scores across all psychological factors as noted in the clustering analysis.

Type VII landscapes combined low complexity gray elements (D = 1.478) with highly complex, concentrated white features (D = 1.781, C = 0.583) and concentrated green areas (C = 0.596). This distinctive pattern created strong visual contrast between simple stone elements and intricate white surfaces, with concentrated vegetation providing compositional structure.

Type VIII landscapes were defined by highly concentrated white elements (C = 0.775) with moderate complexity (D = 1.653), typically representing focused areas of sand, sky reflections, or ceremonial spaces. The relatively low complexity values for yellow (D = 1.317) and red (D = 1.276) elements reflected the subdued, monochromatic quality of these scenes that scored high on Naturalness but low on Decorativeness.

The relationship between quantified color characteristics and psychological factors revealed several significant patterns. Landscapes scoring high on the Openness factor (Types I, IV, and V) consistently displayed moderate to high complexity values for brown elements (D > 1.78) combined with distributed rather than concentrated patterns. The Decorativeness factor correlated strongly with yellow complexity and concentration, particularly evident in Type IV landscapes where gold leaf elements created visually striking focal points.

The Clarity factor showed strong association with white element distribution patterns, with high-clarity landscapes (Types I and V) featuring moderate complexity (D ≈ 1.65) and low concentration (C < 0.37) of white elements. Naturalness correlations were most evident in green element characteristics, with high-Naturalness landscapes (Types III and VIII) displaying either exceptionally high green complexity (Type III) or distinctive white concentration patterns (Type VIII).

The Unity factor showed notable correlation with balanced complexity values across multiple color categories, particularly in Type VI landscapes where no single color dominated in terms of complexity or concentration. The Complexity psychological factor correlated most strongly with the mathematical complexity (fractal dimension) of green and brown elements, particularly in landscape Types II and III.

These findings demonstrate that specific quantifiable color characteristics correlate with distinct psychological responses, suggesting that the therapeutic effects of traditional Japanese gardens may be partially explained through the deliberate orchestration of color complexity, diversity, and concentration patterns. The spatial distribution of landscape types across different psychological dimensions is visualized in Figure 7 (Factors 1 and 2) and Figure 8 (Factors 3 and 4).

To establish quantitative relationships between psychological perception and specific color attributes in Rokuon-ji Garden landscapes, Pearson correlation analysis was conducted between the six psychological factors identified through factor analysis and the 18 color metrics (three quantitative indicators for each of the six color categories). This analysis revealed statistically significant associations between specific color characteristics and psychological dimensions, providing empirical evidence for the relationship between quantifiable color properties and healing effects (Table 7).

The Openness factor demonstrated the strongest correlations with brown element characteristics, showing significant positive associations with both brown fractal dimension (r = 0.455, p < 0.01) and brown diversity index (r = 0.549, p < 0.01). Additionally, moderate positive correlations were observed with gray fractal dimension (r = 0.332, p < 0.05) and gray diversity index (r = 0.383, p < 0.05). Significant negative correlations were observed between Openness and both white concentration index (r = −0.369, p < 0.05) and green concentration index (r = −0.385, p < 0.05). These findings suggest that perceptions of openness in Rokuon-ji Garden are positively associated with complex, evenly distributed brown and gray elements (typically representing architectural features and pathways), while concentrated patches of white and green elements show negative associations with perceived openness. These correlational relationships require experimental validation to establish potential causal mechanisms.

The Decorativeness factor showed only one significant correlation: a negative association with brown concentration index (r = −0.312, p < 0.05). This suggests that scenes perceived as more decorative feature distributed rather than clustered brown elements, potentially indicating that decorative quality in Japanese gardens relies more on the arrangement of wooden architectural elements than on their concentration in focal points.

The Clarity factor displayed a complex pattern of correlations, with significant negative associations with brown fractal dimension (r = −0.363, p < 0.05), gray fractal dimension (r = −0.380, p < 0.05), red diversity index (r = −0.532, p < 0.05), and gray diversity index (r = −0.398, p < 0.05). Conversely, Clarity showed positive correlations with white fractal dimension (r = 0.425, p < 0.05), white concentration index (r = 0.410, p < 0.05), and green concentration index (r = 0.341, p < 0.05). These relationships suggest that perceptions of clarity in garden landscapes are enhanced by simpler (less complex) brown and gray elements, concentrated rather than dispersed green elements, and complex, concentrated white elements.

The Naturalness factor demonstrated significant positive correlation with green fractal dimension (r = 0.402, p < 0.05) and significant negative correlation with green concentration index (r = −0.449, p < 0.01). This pattern suggests that landscapes perceived as more natural are associated with complex green elements (typically vegetation) distributed throughout the scene rather than concentrated in specific areas. The correlational nature of these findings indicates associations rather than causal relationships between color characteristics and psychological responses. The absence of significant correlations with other color categories suggests that naturalness perception in Japanese gardens is primarily influenced by the complexity and distribution patterns of green elements.

The Unity factor showed only one significant correlation: a positive association with white concentration index (r = 0.350, p < 0.05). This suggests that concentrated white elements, such as sand patterns, ceremonial spaces, or focused reflections, contribute to perceptions of compositional harmony and integration in garden landscapes.

The Complexity factor demonstrated significant negative correlation with white fractal dimension (r = −0.343, p < 0.05), indicating that landscapes perceived as more complex paradoxically feature simpler (less fractally complex) white elements. This seemingly counterintuitive relationship suggests that perceived complexity in Japanese garden composition may arise from contrast between complex and simple elements rather than uniform complexity across all color components.

These correlation patterns reveal several important principles underlying the psychological effects of color characteristics in traditional Japanese garden design. First, the perception of openness—a key factor in restorative environments—is strongly influenced by the complexity and distribution of brown and gray elements, suggesting that the arrangement of pathways, wooden structures, and stone elements plays a crucial role in creating mentally expansive spaces despite physical constraints.

Second, the clarity of garden composition appears dependent on the interplay between structural simplicity (in brown and gray elements) and focal complexity (in white elements), reflecting the Japanese aesthetic principle of “controlled contrast” that creates visual order through deliberate juxtaposition of simple and complex forms.

Third, naturalness perception is primarily driven by the characteristics of green elements, with complexity enhancing naturalness while concentration diminishing it. This finding aligns with biophilic design principles that emphasize the restorative benefits of distributed, complex natural patterns.

Fourth, the single significant correlation for the Unity factor suggests that concentrated white elements serve as compositional anchors that enhance perceptual integration of diverse landscape elements. This supports traditional Japanese design concepts that utilize focused negative space (ma) to create coherence within visual complexity.

Finally, the inverse relationship between perceived complexity and white element fractal dimension indicates that psychological complexity arises from relational rather than absolute complexity—a principle central to Japanese garden philosophy where simplicity and complexity are viewed as complementary rather than opposing qualities.

These empirical findings provide quantitative support for traditional Japanese garden design principles while establishing specific relationships between measurable color characteristics and psychological healing effects. Figure 9 further illustrates the distribution patterns across Unity and Complexity factors, completing the visualization of all six psychological dimensions identified in this study. Such relationships offer both theoretical insight into environmental perception mechanisms and practical guidance for designing contemporary healing landscapes based on traditional wisdom.

The quantitative analysis of Rokuon-ji Garden landscapes has revealed specific color characteristic combinations that contribute systematically to healing environmental qualities. The significant relationships between color metrics and psychological factors suggest that traditional Japanese garden designers intuitively developed sophisticated color orchestration principles that enhance restorative experiences. These relationships operate through multiple parallel mechanisms mediated by cognitive and perceptual processes.

The interaction between high-complexity brown elements (typically wooden architectural features with D > 1.8) and dispersed, lower-complexity white elements (such as sand patterns or reflective surfaces with C < 0.35) consistently correlated with elevated Openness and Clarity psychological factors. This combination creates a perceptual dynamic where complex structural elements provide cognitive engagement while dispersed white elements create visual “resting points” that prevent cognitive overload. This balanced visual stimulation facilitates what attention restoration theory describes as “soft fascination”—environmental engagement that supports cognitive recovery without demanding directed attention resources.

Non-uniformly distributed green elements (typically vegetation with high fractal dimension D > 1.9 and moderate concentration C ≈ 0.4) surrounding concentrated brown and gray structural features creates a distinctive spatial composition observed particularly in landscape Types III and VI. This arrangement generates simultaneous perceptions of Complexity and containment, establishing what environmental psychology terms “prospect-refuge balance”—the co-presence of expansive views (prospect) and sheltered spaces (refuge) that evolutionarily signals environmental safety conducive to restoration (Table 8).

The three quantitative dimensions of color characteristics—complexity (fractal dimension), uniformity (diversity index), and concentration—influenced psychological perceptions through distinct but complementary mechanisms. Fractal dimension demonstrated the strongest correlations with psychological factors, particularly Openness and Naturalness, suggesting that the mathematical complexity of color patterns plays a fundamental role in environmental restoration processes. This aligns with theories of fractal aesthetics that identify mid-range fractal dimensions (D ≈ 1.7) as optimal for human visual preference and processing efficiency.

The diversity index, measuring uniformity of color distribution, showed significant correlations with Openness and Clarity factors. High diversity values for brown elements (H′ > 0.35) consistently enhanced perceived Openness, while low diversity values for green elements enhanced Unity. This pattern suggests that evenly distributed structural elements create spatial comprehensibility while concentrated vegetation elements establish visual coherence—a principle consistent with traditional Japanese garden philosophy emphasizing the balance between structure (shin) and texture (so).

The concentration index, quantifying spatial clustering of color elements, demonstrated inverse relationships with psychological factors compared to diversity measures. High concentration values for white elements (C > 0.5) enhanced Unity and Clarity, while low concentration values for green elements enhanced Naturalness. This pattern reveals the importance of strategic color clustering in creating perceptual anchors that organize visual experience while maintaining the natural quality of the overall composition.

The relationship between natural elements (predominantly green) and artificial elements (brown and gray) revealed a crucial balance mechanism underlying the healing effects of Rokuon-ji Garden. Landscapes scoring highest on combined Naturalness and Unity factors (particularly Type III) exhibited specific ratios of green complexity to brown/gray complexity, suggesting an optimal proportional relationship rather than maximization of either quality independently.

High complexity green elements (D > 1.9) combined with moderate complexity brown elements (D ≈ 1.7) created perceptual dynamics that simultaneously supported natural immersion and spatial coherence. This balance echoes the fundamental Japanese aesthetic principle of shizen-kan (natural appearance achieved through human intervention) that distinguishes Japanese gardens from both wilderness landscapes and formally geometric gardens. The correlation data demonstrated that this balance—rather than maximization of natural elements—optimizes healing effects.

The distribution patterns of colors further revealed mechanisms of natural-artificial integration. When green elements displayed moderate diversity (H′ ≈ 0.3) and low concentration (C < 0.4), while brown elements showed high diversity (H′ > 0.35) and moderate concentration (C ≈ 0.45), landscapes consistently scored high on both Naturalness and Unity factors. This pattern suggests that healing environments benefit from natural elements distributed throughout the visual field while architectural elements provide strategic focal points that organize spatial experience.

The empirical relationships between color characteristics and psychological factors identified in this study suggest several fundamental principles of color application in therapeutic landscape design. First, the complexity contrast principle—juxtaposing high-complexity elements (typically brown and green) with moderate or low-complexity elements (typically white and gray)—creates perceptual dynamics that enhance both cognitive engagement and restoration simultaneously.

Second, the concentrated distribution principle—strategic clustering of accent colors (typically red and yellow) while dispersing background colors (typically green)—establishes visual hierarchy that guides attention without creating cognitive strain. This principle manifests particularly in landscape Types IV and VII, which scored highest on combined Decorativeness and Openness factors.

Third, the natural-structural balance principle—maintaining specific ratios of natural color complexity to structural color complexity—optimizes the simultaneous perception of Naturalness and Unity that characterizes highly restorative landscapes. This principle manifests most clearly in landscape Type III, which achieved the highest combined scores on these factors.

These principles demonstrate that the therapeutic effects of Rokuon-ji Garden derive not from individual color characteristics but from sophisticated relationships between colors, specifically their complexity, distribution, and proportional balance. These relationships established through centuries of intuitive design practice now receive empirical validation through quantitative analysis, offering evidence-based guidance for contemporary therapeutic landscape design.

The quantitative relationships between color characteristics and psychological factors identified in Rokuon-ji Garden suggest potential design principles for contemporary therapeutic landscapes that require further empirical validation in clinical and therapeutic contexts. These preliminary findings should be considered as hypothetical design guidelines pending verification through controlled intervention studies with physiological and clinical outcome measures. These empirical findings can be systematically translated into practical design strategies that target specific healing objectives across various therapeutic contexts.

For gardens prioritizing Clarity—a psychological dimension particularly valuable in healthcare environments where cognitive restoration is essential—the research findings suggest implementing combinations of high-complexity brown elements (D > 1.8) with low-complexity, dispersed white elements (D < 1.65, C < 0.35). This combination creates visual environments that are simultaneously engaging and cognitively accessible, reducing mental fatigue while maintaining sufficient environmental legibility. Practical applications include using intricately textured wooden structures as organizational elements while incorporating dispersed light-colored gravel or sand features as visual counterpoints in rehabilitation garden settings.

In therapeutic landscapes seeking to balance Complexity and containment—qualities particularly beneficial in mental health treatment contexts—the research supports centrally positioned, concentrated brown and gray elements (C > 0.5) surrounded by non-uniformly distributed green elements with variable complexity gradients (D ranging from 1.7–1.9). This arrangement creates spatial compositions that provide both cognitive engagement and psychological security, supporting what environmental psychology terms “fascination with boundaries”—a perceptual state conducive to stress reduction and emotional regulation. Implementation strategies include creating central stone or wooden features with peripheral plantings that vary in density and complexity based on distance from the center.

For therapeutic environments aiming to simultaneously achieve Openness, Clarity, and Naturalness—an optimal combination for general wellness spaces—the data supports implementing high-complexity, dispersed green elements (D > 1.9, H′ > 0.35) with strategic red accents (C > 0.6), while maintaining low-complexity uniform distributions of gray, white, and brown elements. This arrangement creates perceptually expansive environments that remain visually coherent and naturalistically authentic, supporting multiple dimensions of psychological restoration simultaneously. Practical applications include creating open lawn areas framed by complex vegetation with seasonal color accents, supported by simple architectural elements that provide spatial definition without visual dominance (Table 9).

The application of Rokuon-ji Garden’s color principles to contemporary therapeutic landscapes must account for significant contextual factors that influence color perception and effectiveness. Seasonal variation—a defining characteristic of traditional Japanese gardens—substantially alters color compositions throughout the year, with corresponding shifts in psychological impact. Research findings suggest that gardens designed for year-round therapeutic function should incorporate Type III and Type VI landscape characteristics, which demonstrated the most stable psychological effects across simulated seasonal variations during sensitivity analysis testing.

Lighting conditions significantly modify perceived color characteristics, particularly affecting fractal dimensions of reflective surfaces such as water features. The strong correlations between white element complexity and psychological factors suggest that therapeutic garden designs should incorporate specific lighting strategies that enhance white element definition under various natural light conditions. Practical implementations include strategic positioning of reflective surfaces relative to primary viewing positions and consideration of diurnal lighting patterns when determining the orientation of key color elements.

Viewing angle and distance—variables systematically documented in the Rokuon-ji Garden analysis—significantly influence perceived color characteristics and their psychological effects. The research data indicates optimal viewing distances for different landscape types, with Type I landscapes demonstrating strongest therapeutic effects at middle distances (15–25 meters), while Type IV landscapes were most effective at closer viewing positions (5–15 meters). These findings suggest that contemporary therapeutic garden designs should incorporate carefully planned circulation systems that guide visitors through predetermined viewing sequences optimized for specific psychological objectives.

Different therapeutic objectives require specific adaptations of Rokuon-ji Garden’s color principles to meet particular functional requirements. For meditation-focused gardens, research findings support emphasizing the color characteristics found in Type I and Type VIII landscapes, with their balanced complexity patterns and strategic use of white concentration. Implementation strategies include creating compositions with moderate complexity green elements (D = 1.7–1.8) surrounding focused areas of highly concentrated, low-complexity white elements (D < 1.5, C > 0.7) that serve as meditation focal points.

For rehabilitation gardens supporting recovery from physical or cognitive impairment, the characteristics of Type II and Type V landscapes offer evidence-based design templates. These landscape types, with their combination of high brown complexity and strategically distributed green elements, create environments that simultaneously support physical navigation and cognitive engagement. Practical applications include designing rehabilitation pathways that progress through color complexity gradients, beginning with simple, clearly defined color compositions and advancing to more complex arrangements as treatment progresses.

Social interaction gardens benefit from adapting the color characteristics of Type IV and Type VII landscapes, which demonstrated highest correlations with combined Decorativeness and Openness factors. These landscape types, featuring complex color contrasts and strategic concentration of accent colors, create environments that stimulate conversation while maintaining psychological comfort. Implementation strategies include creating social gathering spaces defined by concentrated yellow or red elements with moderate complexity (D = 1.6–1.7, C > 0.5) that serve as conversation catalysts while maintaining sufficient background green complexity to ensure perceptual comfort.

The empirical relationships between specific color characteristics and psychological factors identified in Rokuon-ji Garden provide a scientific foundation for therapeutic landscape design that transcends cultural and historical contexts. By systematically adapting these evidence-based principles to contemporary needs, designers can create healing environments that combine the time-tested wisdom of traditional Japanese garden design with the specific requirements of modern therapeutic applications. This research-to-application approach represents a significant advancement in evidence-based design for therapeutic landscapes, offering quantifiable metrics and specific implementation strategies for enhancing healing environments through color characteristics.