2.1. Experimental Water Scape
To understand the role of reflecting water on spatial perception, it was essential to use a typical sample in this study’s setting. For this reason, we used the famous classic Chinese garden Wang Shi Yuan (The Garden of the Master of the Fishing Nets) in Suzhou, China. Wang Shi Yuan is considered one of the finest gardens in southern China; in fact, it is one of UNESCO’s World Heritage Sites. The garden is divided into eastern and western sections. The eastern part consists of residential quarters, whereas the garden area is approximately 5,400 m2 wide in the western part. The entire experimental scenery area was approximately 740 m2, and the water body of this area was a pond (approximately 447 m2) located in the centre, surrounded by buildings, walls, and vegetation. Thus, the scenery is relatively enclosed. Since many studies have considered this layout in Wang Shi Yuan a typical example of using water to expand the perceived spatial dimension and relieve the feeling of enclosure, this scenery is suitable for this study (3,11,12). The plan of the pond and surrounding landscape elements in Wang Shi Yuan is shown in Fig. 1, where the main element is the pond set in the centre, and the main viewpoint is on the east bank of the pond, beside a pavilion. The pond is spread in front of the pavilion, and beyond (west bank) the pond is another pavilion and a long corridor. An additional pavilion and natural stones are set on the south bank of the pond. There is vegetation on the northern bank of the pond (Fig. 1). The reflection features two pavilions, a long corridor, vegetation, and the sky.
2.2. Experimental Setting
The experimental picture was taken using an iPhone12 Pro max camera (4032 × 3024 px) from the main viewpoint beside the pavilion, with the camera facing the pond and beyond (Fig. 1). Photographs were captured during spring (April 2019) on a moderately sunny day with occasional clouds to reduce the influence of weather, season, and equipment. To maintain a clear reflection on the water surface, the photograph was taken on a day with no wind, and no visitors were around, at 7:00 a.m. The viewpoint height on the horizontal line of sight was 1.20 m by eye level (sitting position).
For comparison, the image was modified by an expert using Adobe Photoshop (Adobe Photoshop 2023 (24.6.0) running on Windows, Adobe Inc.). In the modified picture, the pond was replaced by a pavement with a traditional ordinary pattern in a classical Chinese garden, this kind of pavement is common in Chinese garden, which can prevent participants from feel the modified picture unrealistic. Participants saw both versions of the image in a randomized order. Each participant was asked to report whether they could sense the picture that presented the original scenery after the experiment. None of the participants claimed that they could distinguish which picture presented the original scenery during the entire study; thus, the validity and reliability of the objects in the present study were ensured.
The pictures were then replayed on a Windows computer with a 1920×1080 px, 16:9, 22-inch monitor in the experimental room (5 m in length, 3 m in width, and 3.5 m in height) (Fig. 2). The overall environment of the room was sound-absorbing, with the ventilation and lighting systems designed to be soundproof. A sound test scanner was used to check the ambient sound before the test to maintain it within 15–30 dB and to ensure that there was no interference from other ambient sounds. An air conditioner and humidifier were used to maintain the temperature within the range of 16–18◦C which would keep the same temperature between experiment room and the outside, and humidity was set at 50–60%. In addition to the experimental room, a control space was also present from where the researchers monitored the participants. Researchers could observe the experimental room and communicate with them.
2.3. Participants
Sixty-four students from the Suzhou University of Science and Technology (13 males and 51 females, aged between 20–25) participated in this study. All participants were compensated with approximately USD 4.5 (CNY 30), and we strictly guaranteed anonymity and confidentiality of participation. They were all physically healthy with good eyesight (naked eye or soft contact lens > 0.7) and normal hearing. The expectation of space, based on memory, is an important factor in spatial perception. Only those who had never visited and have limited information of the garden (Wang Shi Yuan) were included in the experiment. Thus, we focused on the effect of the visual stimulation of water reflection. All the participants were informed that they could withdraw from the experiment at any time. The Ethics Committee of the Suzhou University of Science and Technology approved the human research protocols for this study (IRB 190703). Previous studies have shown that sex does not play an important role in eye movement or psychological results (18). Thus, the difference in the numbers of male and female participants should not influence the validity of the present results. Participants provided brief demographic information such as age, health status, and ethnicity. All participants were of Han nationality and lived in southeastern China. Using the statistical program G∗Power 3.1, we estimated that a minimum sample size of 39 was necessary for a paired–t-test (given α = 0.05, two tails, effect size dz = 0.6, noncentrality parameter sigma σ = 3.74, critical t = 2.02). For Our sample of 64 was sufficiently powered, with actual power being always greater than 0.8.
2.4. Measurements
2.4.1. Eye movement
Eye tracking and fixation were recorded using an eye-mark recorder (se Pro Eye Tracking system, 7 Invensun Technology Ltd., Beijing, China). A previous study indicated that saccadic movements take 20–50 ms, while fixations last for approximately 200 ms when reading text and longer when looking at images (19). Moreover, previous studies have focused on the relationship between visual attention to scenery and defined fixations as reaching approximately 200 ms (15–17,20). Thus, the eye fixation duration was identified as being of at least 200 ms.
When viewing a scene, eye movements reflect the patterns of visual exploration (21). In previous landscape scene studies combining environmental psychology and landscape design, free viewing was used as a normal operation for participants without a goal-oriented mechanism (22). In an experiment that used an eye-tracking device to record eye movements, people had fewer fixations on natural landscapes than on urban landscapes. Fewer fixations were recorded for plants with leaves than those without leaves (23). Many studies have demonstrated a significant relationship between fixation duration and interest (24,25). Another study mentioned that more fixations in the same observation time increased the observer’s capacity to recognise and memorise what is represented (26). Full identification of an object in a scene requires foveal inspection (27). As an indicator, fixation represents the amount of attention focused on a scene (28). Greater eye fixation is thought to reflect a scene being viewed with greater effort (29). Furthermore, the majority of fixations occurred on the most salient object in the scene, meaning that fixation predicted the object’s level of saliency in a scene (30).
2.4.2. Subjective evaluation
To assess the overall subjective feelings associated with the picture (with a pond vs. with a pavement), the participants were asked to fill out a questionnaire using the semantic differential (SD) method (31,32). Based on previous studies (15,33–35), 17 pairs of adjectives (emotional: gentle–stiff, pleasing–annoying, relaxed–tense, bright–dark, clear-obscure, warm–cold, rhythmic-arrhythmic, harmonious-discordant, calming–upsetting, three-dimensional-two-dimensional, variety-monotonous, imaginative-unimaginative, large-small, broad–narrow, open-closed, secure-insecure, and simple-complex) were used in the questionnaire. For each pair of adjectives, the participants evaluated the picture on a scale ranging from − 2 to 2. For example, if the pair of adjectives is “gentle–stiff”, then − 2 (very gentle), − 1 (a little gentle), 0 (neither gentle nor stiff), 1 (a little stiff), and 2 (very stiff). To verify whether the water body could expand the perceived spatial size, each participant was asked to estimate the size of the entire site by typing a value in meters squared.
2.5. Experiment Procedure
When the participants arrived in the waiting room beside the laboratory on the morning of the experiment, they were fully informed of the procedures and how the instruments would be used. After receiving a description of the experiment, participants signed a consent form for participation. The experiment was conducted from 15–27 April, 2022. The time of the experiment was determined based on the weather forecast, avoiding rainy days so that participants would not feel the gap between the picture content and the outside weather. Participants were instructed to turn off their mobile phones. In the experimental room, the participants sat on a chair in the most comfortable position, 50 cm away from the monitor. The eye-tracker device was placed in front of the participants to record their eye movements accurately. Calibration of the eye movements was performed. Participants were instructed to relax, close their eyes, and not move their heads. While the participants rested with their eyes closed, the first picture was presented on the monitor. Simultaneously, the examiner asked participants to open their eyes and view the scene freely. According to previous studies, Eye movements were recorded for 60s (15,16,20,35). The participants were asked to close their eyes again. After completing the eye-movement measurements, the participants were asked to fill out the SD questionnaire for the most recent picture and to estimate the area of the entire site. The entire process was repeated using another picture. The order of the pictures was randomised across participants (OP and MP) to avoid habituation effects. Finally, participants were asked to report whether they could identify the original scenery (not modified by Photoshop). The total procedure lasted approximately 15 minutes for each participant (Table 1).
Table 1
Time schedule of experimental procedure
| Step1: Preparation (5 min) | Step 2: Viewing (1 min) | Step 3: questionnaire (2 min) | Step 4: Repeat from step2-step3 (3 min) | Step 5: Remove Device (4 min) |
Device fixing | ● | | | | |
Condition check | ● | | | | |
Eye movement | | ● | | ● | |
SD questionnaire | | | ● | ● | |
Area estimation | | | ● | ● | |
Remove device | | | | | ● |
2.6. Analysis
2.6.1. Picture Analysis
Many studies have quantified the landscape elements of experimental photographic stimuli to distinguish the differences between landscape elements (36–38). Regarding eye movement data and the characterization of visual attention, the whole scene was divided to four areas of interest (AOI): Water surface/ Pavement (AOI1), Building (AOI2), sky (AOI3), and vegetation (AOI4). These four AOIs completed the visual field in this experimental analysis (Fig. 3). Adobe Photoshop 2023 was used to label the AOIs by colouring each picture, and the ratio of the AOI area in the picture was calculated. The formula is as follows:
Ratio of AOI area = (Number of AOI pixels) / (Total pixels) × 100
2.6.2. Data Analysis
A paired t-test was used to analyse eye-movement data (i.e. average fixation duration, total fixation duration, number of fixations, mean pupil diameter, saccade velocity, and saccade amplitude) for the two pictures (OP vs. MP). A previous study found that the ratio of the AOI area was positively correlated with the total fixation duration and number of fixations (26,39). To compare the eye-movement data of the AOIs, this study adopted two parameters to eliminate bias driven by area differences between the AOIs of the two scenes.
1. Density of Visual Information (DVI) was calculated by following formula:
DVI = (fixation duration in the area of interest/total fixation duration of the whole scene) / Ratio of AOI area.
2. Power of Visual attractiveness (PVA) was calculated by following formula:
PVA = (fixation number in the area of interest/total number of whole scenes) / Ratio of AOI area.
In a previous study, DVI was used to measure how long the viewer needed to comprehend the object (40). This study adopted this as a parameter to explore whether water bodies and their reflections can increase the information content of scenery. Previous studies revealed that the number of fixations indicates how viewers tend to view objects (7,41). In this study, the PVA was used as a parameter to measure how often the viewer consciously or unconsciously focused on an object. A paired t-test was used to compare the same AOIs between the two pictures.
Data from the SD questionnaire regarding participants’ feelings were subjected to factor analysis to extract factors (maximum-likelihood method, Promax rotation, eigenvalues > 1). The factor scores and estimated areas of the SD questionnaires were analysed using a paired t-test.
In all cases, the significance level was set at p < 0.05, and marginal significance was set at p < 0.1. The statistical analyses were performed using the Statistical Product and Service Solutions (SPSS) version 29.0 software (IBM Corporation, Armonk, NY, USA). All data are shown as mean ± SE (standard error) and 95% CI. All statistical analyses were performed using the SPSS version 29.0 software (IBM Corporation, Chicago, IL, United States).