Gender Differences in Visual Perception of Park Landscapes Based on Eye-Tracking Technology: A Case Study of Beihai Park in Beijing

Abstract

Previous landscape design mostly relies on general standards, failing to fully consider gender differences in landscape visual perception, with relevant research still needing further exploration. This study takes Beijing’s Beihai Park as the research object, using five types of on-site-collected photos (water landscape, plant landscape, architectural landscape, path landscape, and square landscape) as stimuli. Twenty males and twenty females participated in an eye-tracking experiment and a questionnaire survey to analyze gender differences in the visual perception of these five landscapes. The results show the following: (1) females show a “core–radiation” pattern, focusing on mid-short vision and environmental details; males focus on distant views and functional areas. (2) Females have slightly higher APD and fixation counts, with stronger cognitive/emotional fluctuations; males have longer total fixation time and more sustained attention. (3) Males prefer architectural/square landscapes, emphasizing functionality; females favor water/plant landscapes, prioritizing emotional connection with nature. (4) The total fixation time significantly impacts subjective evaluations; the average fixation duration is gender-neutral but uniquely affects evaluations of certain landscape types. This study has guiding significance for enhancing park landscapes’ inclusiveness and attractiveness, promoting different genders’ participation and satisfaction, and boosting space vitality and utilization efficiency.

1. Introduction

As an important part of urban green space, parks assume multiple functions in cities, such as improving the ecological environment, promoting economic development, social progress, and cultural prosperity, enhancing public health, elevating urban landscapes and images, and providing disaster prevention sites. They are an indispensable component of cities. In recent years, research on urban park landscapes has made remarkable progress, showing a multi-dimensional and in-depth development trend. The content covers multiple aspects, including ecological adaptive design [1,2], sustainable development [3,4,5], human-centered design [6,7], cultural integration [8,9,10], health orientation [11,12], climate adaptability [13,14], and big data analysis [15].

Landscape perception refers to the comprehensive perception and interaction with landscapes through human sensory systems such as vision, smell, hearing, taste, and touch [16], reflecting the complete process from information reception to cognitive formation. Visual perception accounts for 80% of the human multisensory system [17], and it plays a leading role in connecting humans with the external environment. Landscape visual perception has always been an important part of urban park landscape research and has played a positive role in promoting the practice of park landscape design and renovation.

Research indicates that there are significant gender differences in the visual perception of park landscapes, with complex influencing mechanisms supported by multi-disciplinary theories in biology, psychology, sociology, and environmental behavioral science. Broadly, these differences stem from both social and cognitive roots: Social Constructivism Theory highlights that gender-specific behavioral norms, aesthetic preferences, and environmental perception patterns are shaped through socialization—for instance, women’s reinforced identity as “caregivers” may heighten their focus on landscape “safety” and “social convenience” (e.g., seat layout, path lighting), while men often prioritize “functionality” and “exploratory potential” (e.g., adaptability of open activity spaces) [18,19]. Complementing this, theories of information processing differences note that men tend to use “systematic thinking”, emphasizing spatial structure and functional logic in landscapes, whereas women lean toward “empathic thinking”, with greater sensitivity to details, colors, and emotional atmospheres—observable in how men gaze more at building outlines and road networks, while women focus on plant colors and street facilities [18,20].

More directly tied to landscape perception mechanisms, the “Prospect–Refuge Theory” (rooted in evolutionary psychology) confirms gendered survival instinct patterns. Women prefer enclosed, plant-sheltered seating and resting spaces, showing a strong refuge tendency, and they often choose seats with plants at their backs and open paths in front (balancing refuge and prospect needs). In contrast, men prioritize visual control of open spaces (e.g., mountain-top viewing platforms, central squares), reflecting the prospect tendency of “explorers” and “protectors”, and they exhibit higher tolerance for “exposure levels” of seats [18,21]. Eye-tracking experiments further validate this; when viewing forest landscapes, women’s gaze lingers more on shaded areas under tree crowns (refuge elements), while men focus more on prospect areas such as forest clearings [18].

Attention Bias Theory further elaborates on gendered focus on specific landscape elements as follows: women visually identify children’s activity areas and rest facilities more quickly, while men pay greater attention to sports fields and landmarks [22].

Additionally, the Care Theory underscores that gendered care responsibilities shape perception—for example, in Carmona, Spain, women attach greater importance to all urban green space features than men, linked to their heavier family care duties and heightened sensitivity to public space safety [23].

Additionally, studies have proven that women exhibit lower visual field sensitivity than men, with an average deviation difference of 0.28 dB [24]. In terms of gender differences in visual persistence, men maintain visual sensations longer than women in Ganzfeld experiments and afterimage experiments [25]. Ruochen Ma’s [18] study on landscape image data of Mizumoto Park in Tokyo shows that men and women have different eye movement patterns, and the impact of element proportions on women’s eye movement behavior is stronger than on men. Men spend more time gazing at trees, while their gazing time on shrubs, flowers, and artificial elements is significantly shorter than that of women. Wang et al. [26] and Wang et al. [27] found that women tend to evaluate landscapes as a whole—preferring low, more colorful trees and hierarchical features—while men prefer taller trees, open spaces, and recreational facilities. Shi et al. [28] showed that women exhibit more and shorter gazes compared to men’s fewer and longer gazes. Women generally feel less safe than men in urban parks [29,30] and tend to be active during the daytime, while men’s frequency of park use is more significantly affected by objective distance. These findings highlight the importance of considering gender differences in the design and construction of urban parks to create more inclusive public spaces that meet the needs of all users [29,30]. In conclusion, gender differences exist in landscape visual perception, but their specific manifestations and influencing mechanisms still need in-depth study.

In terms of research methodologies, previous studies on landscape visual perception predominantly employed subjective evaluation systems, including questionnaires and interviews [27,31], landscape aesthetic assessment [32], psychological prediction with behavioral analysis [33,34], image-content and landscape imagery analysis [15,35], and semantic differential analysis [36]. With the advancement of computer information technology, cutting-edge techniques such as virtual reality (VR), 3S technology, visual perception simulation, landscape visualization [37], and eye-tracking technology [38] have been increasingly integrated into the field. These methodological advancements have propelled landscape visual perception research from qualitative to quantitative paradigms, significantly enhancing research accuracy and efficiency. Among these innovative methods, eye-tracking technology plays a pivotal role in unraveling human visual behavior and cognitive patterns. By measuring participants’ eye movements in experimental setups, it enables the intuitive quantification of visual attention disparities across genders [39,40] and illuminates the cognitive processing mechanisms [41]. The application of eye-tracking has not only enhanced the comprehensiveness of landscape cognition and evaluation but also elevated research precision, demonstrating distinct advantages in exploring gender differences in park landscape visual perception.

Based on the above reasons, this study uses eye-tracking technology to take Beijing’s Beihai Park as the research object and explore the gender differences in human visual perception of park landscapes. The objectives are as follows:

(1)

Analyzing the differences in the distribution of visual hotspots for different landscape types between males and females, and exploring the focusing characteristics and preference differences of visual attention between the two genders.

(2)

Through eye-tracking experiments, obtaining four eye-tracking indicators for males and females, including average pupil diameter (APD), number of fixations (NF), total fixation duration (TFD), and average fixation duration (AFD), and analyzing the differences in visual landscape perception represented by these indicators.

(3)

Through questionnaires, obtaining ratings from males and females on 13 factors, such as landscape spatial sensation, continuity, and spatial coordination, and analyzing the preference degrees of males and females for different landscape types.

(4)

Verifying the correlation between the four eye-tracking indicators representing physiological preferences and the landscape perception questionnaire representing psychological preferences, and comprehensively analyzing the gender differences in park landscape perception between males and females.

This study can provide more accurate and scientific data support for the optimized design and quality improvement of Beihai Park and similar landscapes based on gender differences. It can also provide an effective planning and design basis and guidance for improving the inclusiveness, fairness, attractiveness, and user experience of parks.

2. Materials and Methods

2.1. Study Area

After extensive field research, this study finally selected Beihai Park in Beijing as the research object. Beihai Park was first constructed in the sixth year of the Dading reign of the Jin Dynasty (1166 AD). After continuous construction and renovation during the Jin, Yuan, Ming, and Qing dynasties, it has formed a unique garden art style. It is the oldest imperial garden in the world and the most well-preserved imperial garden with the longest history in China [42].

Beihai Park is located in the center of Beijing. To the north, it adjoins Shichahai Lake; to the south, it is connected to Zhonghai Lake and Nanhai Lake; to the east, it is adjacent to Jingshan Park and the Forbidden City; to the west, it is near a branch of the Beijing Library. It is the core part of the Three Seas (Beihai, Zhonghai, and Nanhai) in Xiyuan. The park covers an area of 68.2 hectares, of which 38.9 hectares are water areas and 29.3 hectares are land areas. Beihai Park is centered around Qionghua Island, which is surrounded by water on all sides, and the White Pagoda serves as a landmark building. It is a representative of the traditional Chinese garden layout of “one pool and three mountains” (Figure 1). From September to October 2024, the authors carried out multiple field investigations in Beihai Park and collected a large number of landscape photos.

2.2. Stimuli

The eye-movement experiment used park landscape photos collected on-site as static stimulus materials. Based on the landscape composition elements and their interrelationships, the landscapes in Beihai Park are divided into five categories in this study: water landscape, plant landscape, architectural landscape, path landscape, and square landscape. After classifying the collected landscape photos, four photos were selected from each type, with a total of 20 photos (Figure 2). These photos were uniformly made into slide videos in a standard 16:9 aspect ratio. To eliminate the possible primacy effect during the experiment, two photos were randomly selected from the remaining photos to be used as warm-up pictures.

2.3. Selection of Participants

Previous investigations have established that professional backgrounds—including backgrounds as designers, ecologists, and the general public—significantly influence the consistency of landscape preferences [43]. Notable disparities in landscape preferences have been documented between design students and non-design students. For instance, design students exhibit significantly stronger preferences for “high-naturalness” landscapes (e.g., wild meadows) than non-professionals, with a greater focus on landscape dynamics (e.g., seasonal variations) [16].

Thus, this study recruited 40 undergraduate and graduate students majoring in architecture, urban planning, and landscape architecture (20 males and 20 females) for an eye-tracking experiment, aiming to effectively explore gender differences in landscape visual preferences among design professionals. Additionally, to ensure clear capture of visual signals, participant screening adhered to the Minimum Reporting Guidelines for Eye-Tracking Research (2023 Edition) [44] and the visual acuity requirements of Tobii Pro Spectrum equipment; participants were required to have uncorrected visual acuity of ≥0.7, excluding individuals with myopia or those wearing corrective lenses.

2.4. Experimental Instruments

The experiment utilized the Tobii Pro Glasses 3 wearable eye-tracker with an eye movement sampling frequency of 50 Hz. The compatible recording software was Glasses 3, and the display device featured a 23.8-inch monitor with a screen resolution of 1920 × 1080 pixels.

2.5. Experimental Procedure

(1)

The participants were briefed on the experimental objectives, procedures, and safety guidelines. They were directed to sit approximately 1 m away from the display, and they were assisted in securing the eye-tracker’s head-mounted module.

(2)

The head-mounted module was properly connected to the recording module. The participant’s name was entered, and the calibration procedure for the Tobii Pro eye-tracker was initiated, ensuring that the participant’s sitting posture and head position were accurately logged. During calibration, the participant was instructed to fix their gaze on the center of the calibration card. The device compensated for individual variations by optimizing the 3D eye model—even if participants differed in sitting height, each calibration session established a new mapping relationship based on their real-time head position [45].

(3)

Two warm-up images were presented to mitigate the primacy effect.

(4)

The experiment formally began. The first set of landscape photos was displayed, and then participants were instructed to complete the questionnaire. This sequence was repeated until all five sets of images had been presented and all questionnaires had been filled out. Each photo was shown for 8 s, and the entire experiment lasted approximately 10 min (Figure 3).

2.6. Selection of the Eye-Tracking Index

After screening, 2 samples with <70% valid eye-tracking data and 6 outliers were removed, retaining 34 valid datasets (17 male, 17 female). According to the research needs, this study selected four eye-tracking indicators—average pupil diameter, number of fixations, average fixation duration, and total fixation duration (

Table 1. Definitions and implications of eye-tracking metrics.

Metrics	Abbr.	Unit	Definition	Implication
Average pupil diameter	APD	mm	Mean dimensions of pupil dilation or constriction during the observation period [46]	Indicating cognitive processing demand and emotional fluctuations. This metric demonstrates high sensitivity, where a 0.1 mm variation in pupil diameter can signify alterations in cognitive states.
Number of fixation	NF		Fixation number within a defined area [47]	This index reflects participants’ attentional number towards specific regions or objects. Higher fixation counts indicate greater attentional engagement, which may suggest that the target area/object has higher cognitive relevance or information density during cognitive processing.
Total fixation duration	TFD	s	Fixation duration at specific coordinates [47]	This metric indicates participants’ attentional engagement and temporal allocation toward specific areas. Prolonged fixation durations generally demonstrate enhanced attractiveness or informational salience within the targeted zones.
Average fixation duration	AFD	s	Calculated by dividing the total fixation duration within a defined area by the fixation count, which yields the mean dwell time per fixation point [48]	This parameter reflects the attentional allocation and cognitive load of participants under specific task/stimulus conditions [49].

)—to analyze the participants’ visual perception characteristics of five types of park landscapes.

2.7. Subjective Evaluation System

This study employed a five-point Likert scale to evaluate and assign scores to 13 subjective landscape evaluation factors (

Table 2. Questionnaire.

Serial Number	Evaluation Factor	Assign Scores (Likert Five-Point Scale)
		−2	−1	0	1	2
1	Spatial openness	Extremely Closed	Relatively Closed	Neutral	Relatively Open	Extremely Open
2	Spatial Continuity	Extremely Discontinuous	Relatively Discontinuous	Neutral	Relatively Continuous	Extremely Continuous
3	Spatial Coherence	Extremely Incoherent	Relatively Incoherent	Neutral	Relatively Coherent	Extremely Coherent
4	Aesthetic Feeling	Extremely Unbeautiful	Relatively Unbeautiful	Neutral	Relatively Beautiful	Extremely Beautiful
5	Spatial distinctiveness	Extremely lacking in distinctiveness	Fairly lacking in distinctiveness	Neutral	Relatively distinctive	Extremely distinctive
6	Historical and Cultural Heritage	Completely Unperceivable	Difficult to Perceive	Neutral	Obviously Perceivable	Extremely Profound
7	Color Coherence	Extremely Incoherent	Relatively Incoherent	Neutral	Relatively Coherent	Extremely Coherent
8	Color Purity	Extremely Cluttered	Relatively Cluttered	Neutral	Relatively Pure	Extremely Pure
9	Color Richness	Extremely Monotonous	Relatively Monotonous	Neutral	Relatively Rich	Extremely Rich
10	Curiosity	Extremely Ordinary	Relatively Ordinary	Neutral	Relatively Novel	Extremely Novel
11	Pleasure	Extremely Unpleasant	Relatively Unpleasant	Neutral	Relatively Pleasant	Extremely Pleasant
12	Comfort	Extremely Uncomfortable	Relatively Uncomfortable	Neutral	Relatively Comfortable	Extremely Comfortable
13	Attractiveness	Completely Unattractive	Relatively Unattractive	Neutral	Relatively Attractive	Extremely Attractive

). Participants were instructed to complete the questionnaire immediately after finishing the eye-tracking experiment for each set of landscape images.

2.8. Data Processing and Analysis

All data processing for this study was conducted on the SPSS 27.0 software platform. First, the descriptive statistical analysis method was applied to calculate the skewness and kurtosis of the eye-movement indicators. The results indicated that the data were non-normally distributed. Second, the Levene statistic in one-way analysis of variance was utilized for testing, and it was found that the eye-movement data did not exhibit homogeneity of variance. Considering that the eye-movement data were neither normally distributed nor had homogeneous variance, the non-parametric Mann–Whitney U test was employed to systematically compare the significant differences in eye-tracking indicators between participants of different genders (α = 0.05), and the direction of the differences was explained by means of mean comparison. Subsequently, the Spearman correlation analysis method was used to analyze the correlation between eye-tracking experiment data and subjective perception evaluations. Finally, regression models were used to quantify the interaction effects of gender variables on subjective landscape ratings and eye-tracking indicators.

3. Results

3.1. Gender Differences in Visual Hotspots

(1)

Overall visual hotspots

In the ErgoLAB (Version 3.17.20) analysis software, a visual analysis was conducted on the viewpoint positions and durations of all participants across 20 photos, and an overlay map of the visual heatmaps for all participants was generated (Figure 4). In the heatmap, red represents the areas with the most concentrated browsing and gaze fixation, while yellow and green represent the areas with less eye gaze. The visual hotspot maps show that for the water landscape, the visual hotspots are mainly concentrated in the core area of the water and the waterfront building interface. The transition area between the center of the water area and the shoreline buildings shows high-density visual aggregation, reflecting the continuous visual attention of observers to the integrity of the water shape and the waterfront space interface. With respect to the plant landscape, visual hotspots display a distinct characteristic of selective focusing. Homogeneous patches of plants are more likely to prompt visual fixation and protracted dwelling. Moreover, plant units characterized by a pronounced morphological distinctiveness or a strong color contrast manifest a highly significant visual priority effect. Regarding the architectural landscape, the visual hotspots on building facades are concentrated in entrance regions and decorative elements such as window grilles and eaves. The fixation and dwell times associated with building detail elements are markedly higher than those of the overall facade. In the path landscape, the visual hotspots present a linear distribution pattern. The rate at which landscape elements at the terminus of the path capture visual fixation is substantially higher than that of common elements along the path. For the square landscape, decorative elements such as sculptures and water features in the core area constitute high-intensity fixation hotspots. The frequency of fixation on these elements is significantly greater than that on the paving area.

Furthermore, among all landscape types, the number of gazes at the square landscape was the most (2607 times), and the number of gazes at the water landscape was the least (2479 times). Regarding the total fixation duration, the water landscape had the longest duration (721.583 s), whereas the architectural landscape had the shortest duration (671.683 s), as depicted in Figure 5.

(2)

Gender differences in visual hotspots

Throughout the experimental procedure, remarkable disparities emerged in the visual hotspots of male and female participants while they were engaged in the viewing of five distinct categories of landscape photographs, as illustrated in Figure 6.

In the visual hotspot map of the water-body landscape (Figure 6a), the spatial distribution of male fixation hotspots is relatively dispersed. These hotspots expand in a discontinuous, leap-like pattern, stretching from the water–land interface towards the distant view. In contrast, the female visual hotspots cover a substantially larger area. They continuously span from the water–land intersection to the middle-and near-range views, presenting a more prominent focusing characteristic with more well-defined boundaries. With respect to the objects corresponding to the visual hotspots, both male and female participants demonstrate visual inclinations towards buildings and the water–land intersection. Additionally, female participants display notably strong visual preferences for building reflections, the water itself, and aquatic plants. This finding implies that females possess higher visual sensitivity and a more potent ability to capture micro-features.

Based on the plant landscape hotspot map presented in Figure 6b, it is evident that the overall spatial distribution extent of male visual hotspots is more restricted than that of females. Only in the first photo of tree–shrub combinations within this group of landscape photos does the scope of male visual hotspots surpass that of females. This particular observation strongly implies that males tend to have a relatively elevated visual preference for composite plant-based landscapes. In marked contrast to the male pattern, female visual hotspots display an expansion from the core plant landscape to the adjacent surrounding structures. When considering the landscape elements associated with these visual hotspots, males manifest a relatively greater visual preference for the plant landscape in isolation. Females, on the other hand, demonstrate a more pronounced visual preference not only for the plant landscape but also for the entire complex of surrounding landscape elements.

Upon examination of the architectural landscape hotspot map presented in Figure 6c, it becomes apparent that the distribution range, dimensions, and morphology of male and female visual hotspots exhibit remarkable similarities. Moreover, the directions of their expansion also display a high degree of resemblance. The landscape elements corresponding to these visual hotspots primarily encompass architectural details and decorative components, including eaves, plaques, windows, and red lanterns. With respect to the White Pagoda, which serves as a prominent landmark within Beihai Park, both male and female participants demonstrate a relatively pronounced visual preference for its overall architectural form.

According to the path landscape hotspot map depicted in Figure 6d, within the linear path landscape scenario, the visual hotspots of both male and female participants converge at the visual focal point located at the terminus of the path. Notably, the spatial extent of male visual hotspots is markedly larger compared to that of females. In the cases of curved-path, forest-trail, and building-corridor landscapes, the visual hotspots of both male and female participants are situated at middle-and close-range positions. Significantly, the spatial extent of female visual hotspots is substantially larger than that of males. Regarding the landscape elements associated with male visual hotspots, they encompass the path surface, trees at the path’s end, decorative lanterns suspended in the sky, the demarcation of the forest trail, buildings in the distance within the building corridor, and far-off columns. As for the landscape elements corresponding to female visual hotspots, they comprise the path surface, the edge and paving stones of the forest trail, buildings in the distance within the building corridor, and adjacent columns.

Through analysis of the square landscape hotspot map presented in Figure 6e, it can be seen that the distribution pattern of visual hotspots for males and females is roughly the same. Across all four photographs within this particular group, the visual scope of females is marginally larger than that of males. When considering the landscape elements corresponding to these visual hotspots, a high degree of similarity is exhibited between males and females. These elements encompass building entrances, incense burners positioned in front of doors, tree trunks, building bases, courtyard pavements, and billboards affixed to the wall.

3.2. Differences in Eye-Tracking Indices

As depicted in the analysis of Figure 7, the eye-tracking indices conform to a non-normal distribution and lack homogeneity of variance. The comprehensive findings of the analysis regarding the eye-tracking indices are presented in

Table 3. Gender differences in all landscape photos.

Eye-Tracking Indices	Z	Sig.
APD (mm)	−0.396	0.692
NF	−2.015	0.044
TFD (s)	−2.807	0.005
AFD (s)	−1.464	0.143

. When considering all landscape types, the average pupil diameter and the average fixation duration do not display significant differences. The asymptotic significance of both the number of fixations and the total fixation duration are less than or equal to 0.05, thereby signifying significant differences. This implies that male and female participants demonstrate comparable cognitive loads and patterns of attention allocation for diverse landscapes in Beihai Park. Nevertheless, disparities exist in the frequency and intensity of their attention directed towards these landscapes.

Figure 8 presents a comparative analysis of gender differences in four eye-tracking metrics: average pupil diameter (APD), number of fixations (NF), total fixation duration (TFD), and average fixation duration (AFD). Specifically, female participants exhibited an APD of 3.78 mm, surpassing the male mean of 3.75 mm. In terms of NF, females recorded 379.53 fixations, outnumbering males’ 374.59 fixations. Conversely, males demonstrated a significantly longer total fixation duration of 100.44 s compared to females’ 82.33 s. For the average fixation duration, females showed a marginal advantage (0.29 s) over males (0.27 s). These quantitative results vividly illustrate sexually dimorphic patterns in oculomotor indices associated with visual information processing.

Table 4. Eye-tracking indices of male and female participants across the five landscape types.

Landscape Type	Eye-Tracking Indices	Mean (Male)	Mean (Female)	Sig.
Water landscape	APD (mm)	3.526	3.553	0.718
	NF	66.529	79.294	0.042 *
	TFD (s)	18.33	24.117	0.007 **
	AFD (s)	0.273	0.316	0.117
Plant landscape	APD (mm)	3.925	3.94	0.931
	NF	71.118	79.588	0.221
	TFD (s)	18.361	23.008	0.017 *
	AFD (s)	0.278	0.314	0.293
Architectural landscape	APD (mm)	3.704	3.701	0.986
	NF	83.353	67.941	0.005 **
	TFD (s)	22.73	16.781	0.008 **
	AFD (s)	0.275	0.243	0.06
Path landscape	APD (mm)	3.747	3.726	0.85
	NF	72.824	80.118	0.162
	TFD (s)	18.493	22.786	0.014 *
	AFD (s)	0.257	0.293	0.134
Square landscape	APD (mm)	3.863	3.981	0.744
	NF	80.765	72.588	0.459
	TFD (s)	22.53	18.644	0.031 *
	AFD (s)	0.289	0.267	0.163

** and * represent significance levels of 1% and 5%, respectively.

displays the means and significance levels of eye-tracking indices for male and female participants across the five landscape types. Statistical analysis shows that the difference in average pupil diameter between men and women when viewing the five landscape types is not significant, which indicates that there is no substantial difference in emotional fluctuations or cognitive load between male and female participants when perceiving different landscapes. Minor changes in pupil diameter may not be sufficient to reflect significant gender differences in emotional responses or attention concentration.

Table 4 further reveals that for the number of fixations (NF), females exhibited significantly higher numbers of fixations than males for the water landscape (Z = −2.035, p = 0.042), while females also demonstrated a greater NF for the architectural landscape (Z = −2.792, p = 0.005). This observation suggests that males directed more concentrated attention toward architectural scenes, whereas females allocated more fixations to water environments, implying that males may perceive architectural landscapes as more informationally salient or cognitively prioritized, while females attribute greater informational value to water features. Conversely, statistical analyses of plants, path, and square landscapes showed no significant gender differences in the NF, indicating that males and females displayed comparable frequencies of attentional engagement with these landscape types.

Regarding the total fixation duration (TFD), males exhibited longer total fixation durations than females in architectural (Z = −2.635, p = 0.008) and square landscapes (Z = −2.153, p = 0.031), whereas females demonstrated a greater TFD in the water (Z = −2.704, p = 0.007), plant (Z = −2.394, p = 0.017), and path landscapes (Z = −2.463, p = 0.014). Statistical analyses confirmed the significance of all these differences (p < 0.05), indicating that males allocated more attentional time to architectural and square scenes while observing diverse landscapes, whereas females directed greater focus toward natural-element landscapes (water, plants) and path environments.

The difference in average fixation duration (AFD) between men and women across different landscape types is extremely small, and statistical analysis shows that the difference is not significant, indicating that there is no substantial difference in attention allocation or cognitive load between men and women when observing landscapes.

3.3. Gender Differences in Questionnaire Scores

The questionnaire scores of each participant when viewing the five types of landscapes are shown in Figure 9. The systematic collation of questionnaire data reveals gender-based disparities in subjective visual preferences for various landscape types (

Table 5. Gender differences in the general score.

Gender	Sample Size	Mean	SD
Male	17	6.29	38.85
Female	17	−13.88	41.48

). In terms of the overall ratings of Beihai Park’s comprehensive landscape, males exhibited a higher mean score than females, indicating that males generally demonstrated stronger preference intensity for the landscape of Beijing’s Beihai Park. Moreover, the standard deviation of male scores was smaller than that of female scores—a data feature that reflects that inter-individual variations in landscape preference were more pronounced among females, with a significantly higher degree of dispersion compared to the male group.

Based on data collation and statistical analyses, an examination of gender score disparities across five landscape types (

Table 6. Gender differences in scores of diverse landscape types.

Landscape Type	Gender	Sample Size	Mean	SD	Z	Sig.
Water landscape score	Male	17	0.06	15.25	−0.64	0.52
	Female	17	2.82	12.23
Plant landscape score	Male	17	−0.35	14.93	−0.17	0.86
	Female	17	−0.29	13.05
Architectural landscape score	Male	17	3.94	17.58	−2.14	0.03 *
	Female	17	−9.24	13.29
Path landscape score	Male	17	−1.76	14.45	−0.12	0.90
	Female	17	−1.53	16.62
Square landscape score	Male	17	4.41	14.47	−1.74	0.08
	Female	17	−5.65	17.43

* represent significance levels of 5%.

) reveals that women had higher average scores than men in visual preference ratings for water, plant, path, and square landscapes, with no significant differences. This indicates that in Beijing Beihai Park, women have a slightly stronger overall preference for these landscape types; whereas in the ratings of architectural landscape, men’s scores were significantly higher than women’s (Z = −2.14, p = 0.03). Notably, males showed larger standard deviations than females in ratings of water, plant, and architectural landscapes, suggesting greater inter-individual variations among male participants in their perception of these three landscape types. Conversely, females displayed more pronounced individual differences in path and square landscape scores.

3.4. Correlation Analysis Between Eye-Tracking Indices and Questionnaire Scores

Using SPSS 27.0 statistical software, Spearman’s correlation analyses were conducted on questionnaire ratings and experimental eye-tracking data to calculate correlations between overall eye-tracking metrics and total questionnaire scores for all landscape images, as well as correlation coefficients between the eye-tracking indices of the five landscape types and their corresponding questionnaire scores (Figure 10). The analyses revealed significant correlations between the total questionnaire scores and both total fixation duration (TFD) and average fixation duration (AFD). When analyzing the five landscape categories separately, the questionnaire scores for each category exhibited significant correlations with the TFD. Given the previously documented significant differences in TFDs across the five landscape types, this correlational analysis validates the consistency between subjective landscape preferences and objective eye-tracking data, demonstrating that the TFD exerts a significant impact on landscape preference. Notably, subjective visual evaluations of water, architectural, and path landscapes also showed significant correlations with the AFD, suggesting that while the AFD does not vary significantly by gender, it is associated with the questionnaire scores. This implies that gender is not a primary determinant of the AFD, despite its correlation with subjective preference ratings.

Further stepwise regression analysis was conducted to construct an evaluation model for objective eye movement data.

First, the regression model for the main effects of eye movement indicators used the following expression:

QS = 237.167 − 2.405·APD − 0.944·NF + 4.360·TFD − 205.687·AFD

In the formula, QS represents the questionnaire score. The constant term 237.167 denotes the baseline predicted value of QS when APD, NF, TFD, and AFD are all 0 (significance p = 0.063, marginally significant). Among the effects of independent variables, APD (coefficient −2.405, p = 0.361 > 0.05, VIF = 1.066, no collinearity) has no significant effect on QS; for each one-unit increase in NF (coefficient −0.944, p = 0.011 < 0.05, VIF = 17.071, moderate collinearity), QS decreases by an average of 0.944; TFD (coefficient 4.360, p = 0.003 < 0.01, VIF = 34.889, high collinearity, standardized Beta = 2.606) has the strongest positive impact on QS, with QS increasing by an average of 4.360 for each one-unit increase; AFD (coefficient −205.687, p = 0.020 < 0.05, VIF = 24.706, high collinearity) has the largest negative impact on QS, with QS decreasing by an average of 205.687 for each one-unit increase.

Second, the regression model for the main effect of gender was constructed with the following linear regression model:

QS = 26.471 − 20.176·G

In the formula, G represents gender (G = 1 for male, G = 2 for female). The intercept term (26.471) represents the predicted value of the dependent variable QS when the independent variable G takes a value of 0; the regression coefficient of G is −20.176, reflecting the marginal effect of G on QS, i.e., for each one-unit increase in G, the predicted value of QS decreases by an average of 20.176 units, which is statistically significant (t = −2.164, p = 0.033 < 0.05).

Third, the regression model for the interaction effects between eye movement data and gender was constructed as follows:

QS = 12.272 + 0.656GAPD − 0.270GNF + 1.423GTFD − 47.247GAFD

In the formula, GAPD, GNF, GTFD, and GAFD are interaction effect variables between gender and the four eye movement indicators. Each one-unit increase in GNF (p = 0.012) leads to an average decrease of 0.270 in QS, and each one-unit increase in GTFD (p = 0.002) leads to an average increase of 1.423 in QS; the effects of both on QS are significant at the 95% confidence level (standardized coefficients show that GTFD has the strongest positive driving effect, followed by GNF with a negative effect). The effects of GAPD (p = 0.684) and GAFD (p = 0.063, close to the critical value) do not pass the statistical significance test.

4. Discussion

4.1. Visual Landscape Preferences in Traditional Chinese Parks

At the theoretical basis level, the explanatory power of the “Prospect–Refuge Theory” is validated. The literature points out that humans prefer “scenes where the environment is observable and there are sheltered spaces” [50]. In this study, women’s focus on plant-shaded resting spaces and details of water features (reflecting a “refuge tendency”) and men’s preference for open squares and distant views of buildings (reflecting a “prospect tendency”) are precisely the concrete manifestation of this theory in the gender dimension. This gender-differentiated survival instinct pattern further supports the view in the literature that “perception of natural environments is related to evolutionary adaptation” [51], i.e., the differences in visual strategies formed by different genders during long-term evolution may be deeply related to the response mechanisms of environmental mathematical structures such as fractal characteristics.

The common laws of landscape preference are partially confirmed in this study, but gender differences make them more complex. Existing studies have shown that the public generally prefers semi-open park environments [52,53], and in natural landscapes, structure-rich types such as “lakes” are strongly correlated with high aesthetic perception, while monotonous areas such as “grasslands” receive low ratings [50]. In this study, women’s high attention to water features and plant landscapes and men’s preference for buildings and squares are consistent with this law—women tend to focus on the rich details of natural elements, while men pay attention to the functional structure of artificial landscapes. However, the difference is that the “universal preference” in existing studies is filtered by gender in this study; women’s need for emotional connection with natural elements strengthens their preference for detail-rich landscapes, while men’s pursuit of functionality makes them more concerned about the structural clarity of artificial landscapes, which mutually verifies the conclusion in the literature that “historical buildings and characteristic projects are related to high aesthetic perception” [50].

Regarding the perception of landscape value, the literature emphasizes that the public pays the most attention to social values (such as cultural communication and urban image) [54]. Although this study did not directly measure the value dimension, gender differences indirectly echo this; men’s high ratings of architectural landscapes (questionnaire results) may be related to their attention to the cultural symbols carried by historical buildings and the function of urban landmarks, while women’s preference for water features and plants is more related to the emotional experience of the environment. These two, respectively, point to the cultural value and social–emotional value of landscapes, forming an implicit echo with the value ranking in the literature.

Studies on traditional Chinese gardens have shown that “water features, buildings, and ornaments enhance visual aesthetics” [55,56], which is highly consistent with the distribution of visual hotspots in Beihai Park (a typical traditional garden) in this study—core areas of water features and decorative details of buildings (such as window lattices and eaves) are all high-frequency fixation points. However, this study further reveals that there is gender differentiation in the perception of these elements; women pay more attention to the interactive relationship between plants and buildings (core–radiation pattern), while men focus on the overall structure of buildings. This adds a gender perspective to the conclusion in the literature that “mid-to-high-rise vegetation and water features enhance aesthetics” [55], suggesting that landscape design in gardens needs to take into account the differentiated interpretation of landscape elements by different genders.

It is worth noting that the conclusion in the literature that “areas without a horizon and completely open spaces are related to low aesthetic perception” [50] is reflected in this study as a balance of gender differences; men’s need for visual control over open spaces (such as squares) and women’s preference for sheltered spaces together constitute the gendered expression of preference for “semi-open environments” [52,53]. That is, the optimal choice at the group level may be the neutralization of the needs of different genders, which provides a theoretical basis for inclusive park design.

4.2. Gender Differences in Gaze Distribution Patterns

In this study, females’ gaze preferences for architectural reflections and aquatic plants in water landscapes, together with the hotspot distribution extending from core plants to peripheral structures in the plant landscape, are consistent with Lucio et al.’s proposition of “systematic visual exploration strategies adopted by females.” [57]. Ma et al. found that “females spend longer gazing at shrubs and flowers”, and this study further corroborates that females exhibit stronger visual sensitivity to plant density and micro-elements (e.g., plant morphological contrasts) [18].

Tarashkar’s study explored the relationship between the landscape spatial quality index (LSQI) and women’s preference for landscape elements (LC) through a photo questionnaire survey of 178 women. It was found that women had the highest preference for shelter scenes, such as shrubs, trees, and flowers. This is consistent with the longer duration of the female gaze towards vegetation landscapes in this study [58]. This discrepancy may stem from definition differences in “natural areas”; the “natural landscapes” in this study (e.g., waterfronts, plant patches) more closely resemble urban green spaces, whereas the literature may refer to wilderness-type natural environments, indicating that females’ perception of natural landscapes varies across scales. Khaleghimoghaddam found that women perceive the built environment and its visual elements in more detail than men, but this study revealed that men fixate significantly longer on classical architectural decorations (such as red lanterns and plaques) than women [59]. This may be attributed to cultural disparities in research contexts; the architectural symbols in Beihai Park carry historical narratives, to which males are more inclined to attend for their structural and semiotic meanings, whereas gender differences may be attenuated in modern urban parks.

4.3. Gender Differences in the Eye-Tracking Index

In this study, there is a significant correlation between total fixation duration (TFD) and questionnaire scores—women showed longer TFDs on water features, which were associated with higher subjective ratings. This validates Tim Holmes’ conclusion that “cumulative fixation duration is a better predictor of preference during long presentations (5000 milliseconds)” [60]. The research by Glaholt et al. further reveals that fixation duration is an effective indicator of preference choices [61]. Consistently, Wang et al. found a positive correlation between fixation duration and subjective interest, which is consistent with the results of this study [62].

What is worth further discussion is that the correlation in this study of “the longer the fixation time, the higher the preference degree” may be closely related to the characteristics of the experimental design and the attributes of the research subjects. This study adopted a free viewing task, which means that participants’ fixation behavior is not subject to forced guidance by external task instructions, but instead, they spontaneously allocate visual resources based on their own inherent attentional tendencies. In such an unconstrained context, individuals are more likely to keep their gaze on objects that they are subjectively more interested in and prefer.

Meanwhile, the research scene is park landscapes. As a typical recreational outdoor space, the park environment itself generally has the attribute of “being pleasant” and mostly aims to satisfy people’s positive emotional experiences. When there are no obvious aversive or irrelevant elements in the environment, differences in fixation time are more likely to directly indicate differences in preference degree—that is, the stronger the sense of pleasure and the higher the preference degree for a certain landscape element (such as a waterscape), the more one tends to prolong the fixation time to obtain a more adequate positive experience.

Women had a significantly higher number of fixations (NF) on water features, while men showed a higher NF on architectural landscapes. This aligns with Ma et al.’s conclusion that “women fixate on detailed elements more frequently” [18]. This may reflect gender-based differences in cognitive styles; women tend to explore details from multiple perspectives, whereas men focus their attention on key functional elements (such as building entrances).

4.4. Questionnaire Results and Gender Preferences

In this study, females gave higher questionnaire ratings for water and plant landscapes, while males showed stronger preferences for architectural landscapes, consistent with Ma et al.’s conclusion that “females are more sensitive to plant density” [18]. Polko and Kimic noted that “females value park environmental details”, and this study further confirms females’ emotional preferences for natural landscapes through subjective ratings, whereas males focus more on the functionality and structure of artificial landscapes [30].

Güngör and Akyüz contended that “there are no significant gender differences in design standards”, yet this study’s questionnaire results revealed pronounced gender divergence in preferences for architectural and natural landscapes [63]. Potential explanations for this discrepancy include the following: (1) differences in research methods, as this study integrated objective eye-tracking data, whereas existing research relies on subjective questionnaires; (2) variations in landscape cultural contexts, where architectural symbols in classical gardens (e.g., the White Pagoda) may exert stronger stimulation on males’ historical and cultural cognition. Westover proposed that “females exhibit more concentrated park visiting times”, but this study did not involve behavioral data, focusing solely on static landscape perception [64]. Future research should integrate eye-tracking with actual usage behavior to validate whether visual preferences translate into activity choices in real-world settings (e.g., whether females’ preference for water landscapes leads to more frequent stays in waterfront areas).

4.5. Research Limitations

Currently, this study is subject to three primary limitations. First, the participants were predominantly college students majoring in design-related fields. This demographic composition hinders the study from reflecting the design cognition, aesthetic preferences, and usage needs of a broader social spectrum, thereby limiting the generalizability of the research findings. The specificity of this sample introduces potential biases in the interpretation of landscape perception, and the results may not accurately represent the characteristics of more diverse social groups. Second, the experimental design relies on static photos to simulate landscape environments. While this method enables strict control over variables, it cannot fully replicate the dynamic visual experiences of real-world settings, where factors such as movement, perspective, and temporal changes influence perception. Additionally, heat map analyses show that central regions of all images consistently emerged as visual hotspots; however, whether this centric bias reflects innate visual preferences or merely photographic composition effects (e.g., how camera angles alter focal points) remains unexamined. In the collection and analysis of eye-tracking data, the central fixation pattern is prone to two major biases: first, the theoretical presupposition in fixation points localization sets the central area as the core region of interest, marginalizing data from peripheral areas and overlooking non-central fixation phenomena. Second, the simplification tendency in trajectory interpretation reduces eye movement trajectories to a one-sided narrative of “diffusion from the center to the periphery”, missing complex actual patterns such as cyclic fixations and multi-focal shifts, which undermines a comprehensive understanding of visual behavior. It is unclear if adjusting shooting perspectives would shift attention to specific landscape elements or maintain the observed central fixation tendency.

4.6. Future Research: Interactive Effects of Age, Cultural Background, and Other Variables

Age, gender, and residential experience give rise to variations in preference ratings, with preferences evolving across the life cycle. Young children exhibit the highest preference scores, whereas older adults show the lowest. Adolescence marks the divergence of preferences between males and females, as well as between urban and rural residents [65]. Cultural background influences visual attention and cognitive load during landscape observation [66], while educational level and environmental experience also play pivotal roles in shaping preferences [67].

Future investigations could address these limitations by expanding participant groups to include diverse age cohorts, occupations, and cultural backgrounds, thereby enhancing data representativeness. In situ eye-tracking experiments in actual landscape environments would capture more ecologically valid visual behaviors, illuminating how dynamic real-world settings shape attentional patterns. Additionally, integrating multisensory stimuli (e.g., natural sounds, environmental odors) would facilitate a more comprehensive understanding of how gender influences integrated perceptual experiences, bridging the gap between isolated visual analyses and complex human–environment interactions.

5. Conclusions

5.1. Gender Differences in Spatial Distribution Patterns of Visual Hotspots

Spatial disparities in visual hotspot distribution reveal that females exhibit a low-entropy, radiation-based spatial cognitive schema, with visual hotspots covering broader areas compared to males. Their gaze patterns show distinct centripetal-radiation characteristics—diffusing from central to peripheral regions—indicating comprehensive environmental scanning. Conversely, males’ direct visual attention is toward distant vistas, whereas females prioritize mid-to-near visual fields. Males’ hotspots concentrate in functional identification zones (e.g., architectural entrances, pathway extensions). Females, however, form a “core–radiation” gaze pattern that integrates fine-grained details (e.g., architectural decorations, plant community structures) with environmental contexts (e.g., interactions between plants and hardscapes), reflecting aesthetic preferences for multi-element coordination.

5.2. Gender Differences in Physiological Indices of Eye-Tracking Experiments

During the observation of all landscape types, females showed marginally higher values in the average pupil diameter (APD), number of fixations (NF), and average fixation duration (AFD) than males—reflecting distinctions in cognitive engagement depth, attentional allocation patterns, and intensities of emotional and information processing between genders. Females consistently exhibited higher emotional fluctuations, cognitive investment, more meticulous attentional distribution, and deeper information processing when viewing landscape images. Males, conversely, demonstrated a significant advantage in total fixation duration (TFD), indicating more focused and sustained attention, as well as a higher requirement for in-depth processing of specific information during landscape observation.

5.3. Gender Differences in Visual Evaluation Scores

Males gave significantly higher ratings for architectural and square landscapes, indicating their rational orientation toward functionality and spatial structure. In contrast, females showed notably higher ratings than males for water, plant, and square landscapes—reflecting an emotional orientation toward natural details and affective connections. These disparities suggest that males prioritize functional attributes and structural clarity in landscapes, while females emphasize natural elements and the emotional resonance derived from environmental interactions.

5.4. Correlations Between Physiological Indices and Psychological Evaluations

Overall, the total questionnaire scores were significantly correlated with both the total fixation duration (TFD) and average fixation duration (AFD), indicating that these fixation metrics exert a notable impact on subjective landscape evaluations. Separate analyses of the five landscape types showed significant correlations between the questionnaire scores and TFD. When combined with the previous finding of significant TFD differences across the five landscape types, this validates the consistency between subjective landscape preferences and objective eye-tracking data, highlighting the prominent influence of the TFD on landscape preference.

Notably, subjective visual evaluations of water, architectural, and path landscapes were also significantly correlated with the average fixation duration (AFD). This suggests that while the AFD was not significantly associated with gender (indicating that gender is not a primary determinant of this metric), it is correlated with questionnaire scores, playing a unique role in subjective evaluations of specific landscape types.

Informed by findings from gender-differentiated landscape perception research, park design should adopt a “functional–emotional” dual-path optimization strategy. For male users, designs should emphasize spatial core area identification and standardized visual guidance systems, highlighting structural logic—such as architectural facade hierarchies and pathway axial extensions—and employ high-saturation color contrasts to match their wide-area scanning and efficient cognitive patterns. For female users, the focus should shift to micro-aesthetic details and emotional color tones (e.g., soft hues and gradient combinations) in natural landscapes, integrated with delicate decorative elements to enhance detail-driven appeal. To meet common needs, pathway landscape design must balance linear functionality with adjacent environmental aesthetics, ensuring navigation efficiency without compromising visual appeal. Architectural spaces should integrate macro-structural clarity with craftsmanship details, using color schemes that reconcile pragmatic cues (for males) with emotional associations (for females). Psychological experience design can leverage flexible spatial boundaries, multi-scalar social zones, and logically coherent interactive interfaces to accommodate males’ preference for structural stability and females’ inclination toward sensory diversity, thereby enhancing inclusive landscape perception.