Multimodal Quantitative Research on the Emotional Attachment Characteristics between People and the Built Environment Based on the Immersive VR Eye-Tracking Experiment

Abstract

The campus landscape contributes a lot to students’ mental and physical health. Students’ emotional attachment to landscape space is an important scientific basis for landscape design. This study used immersive virtual reality eye tracking supported by HTC Vivo Pro and an emotional attachment scale to investigate the relationship between different landscape elements and students’ visual behavior and emotional attachment. ErgoLab and SPSS were used to analyze the indicators. The results showed that: (1) Artificial elements were more likely to attract students’ visual attention and continuously enhance their interest in the landscape. (2) The waterscape space was more likely to attract students’ visual attention, while the attractiveness of arbors and shrubs was related to their color and spatial location. (3) The characteristics related to nature were generally conducive to the establishment of students’ emotional attachment, including both the natural elements and artificial structures that could reflect the natural texture and time traces. (4) Three-dimensional spatial sequence design of landscape elements significantly affected students’ visual focus and emotional experience. The results further contribute to providing a clearer understanding of how students’ preference for specific landscape elements can be obtained and used in decision making for the planning and management during campus renewal and design.

1. Introduction

Campus landscapes are important determinants of students’ well-being, as the design and construction of a natural landscape environment has been proven to have important value for students’ physical, emotional and mental health [1,2,3,4,5,6,7]. During the pandemic of COVID-19 in the past three years, the landscape space in campuses has become an important place for students to engage in outdoor activities and receive emotional healing especially during campus lockdown [8,9,10]. As most Chinese first-tier cities such as Beijing have entered a stage of stock renewal [11,12], in the campuses located there that cover a large area attention has also begun to be paid to students’ emotional experience in landscape improvement. Under these circumstances, evaluating campus landscape quality facilitates more efficient use of limited space [13,14,15]. Meanwhile, as public landscape on campus always serves as space for students’ extracurricular activities, rest and socialization, its quality has been proven to enhance students’ emotional attachment and satisfaction with the campus environment through perceptions. As is known, visual perception accounts for over 80% of all surrounding information among the different types of human perceptions [16], and this has been widely recognized for its value in healing people’s physical and mental health, creating positive emotions, and promoting more humanistic landscape environment design [17,18]. In addition, the importance of campus landscape evaluation in improving students’ life quality and formulating more reasonable university environment design and management strategies has become increasingly prominent [19,20].

That visual perception has been widely used to study people’s preferences to landscape and related evaluation paradigms has been established by multiple scholars, such as the prospect refuge theory proposed by Appleton [21,22] and the well-known preference assessment matrix proposed by Kaplan [23,24]. Based on this, perception-oriented methods including the scenic beauty estimation method [25] and semantic differential method [26] have been further applied to study the relationship between people’s emotional status and the aesthetics of the vegetation [19,27,28]. In the existing research, the methods of exploring users’ emotional demands of the landscape are mostly qualitative, and the evaluation of their emotional experience has mostly been achieved through questionnaires, interviews, observations and ethnographic analysis [19,29]. Though new techniques such as machine learning and data mining have been used to simulate the landscape’s aesthetic quality and evaluate people’s preferences in recent years in a more objective way [30], the internal mechanism by which people perceive landscapes through vision and gain emotional experience still remains to be addressed [29,31]. Meanwhile, human emotion is still difficult to measure in a quantitative way and this always calls for more innovative, holistic and scientific methods [32,33].

In recent years, the development and improvement of ergonomic technology has provided scientific means to explore human emotions through objective physiological channels [34,35,36,37]. For example, a variety of portable wearable devices for measuring physiological signals such as electroencephalography (EEG), galvanic skin response (GSR), heart rate variability (HRV) and eye trackers have begun to be widely used in research concerning human perception and emotion and have expanded from laboratory research in the medical field to multidisciplinary research in the real world [32,38,39,40,41]. These techniques provide researchers and designers with a better means to measure people’s perception and emotions. Among them, the application of eye tracking is expanding in the field of environmental design-related research, as it provides relevant researchers with more convenient ways to detect and track people’s interaction with landscape features through vision, supplementing the lack of scientific means in traditional design disciplines to explore the relationship between people’s inner emotions and the outside world.

The visual system is the most important channel for human beings to obtain external information, and the study of eye movement is considered to be the most effective means in the study of visual information processing [42,43]. During the Middle Ages, people began to use instruments to observe and experiment with eye movements [44]. In 1901, Dodge and Cline developed the first accurate, unforced eye-tracking device [45]. The concept of eye tracking was first developed in the 1930s, and in the early days it was mainly used in psychology, aerospace-related dynamic analysis and commercials. In 2000, Christopher D Wickens et al. applied eye-tracking technology to visual attention allocation for aircraft traffic monitoring and avoidance [46]. In 2001, Hirotaka Aoki and Kenji Itoh used eye-tracking technology to analyze the impact of auditory information on viewers’ visual perception during TV commercials [47]. Nowadays, with the development of optical sensor technology and the improvement in computer information-processing capabilities, eye-tracking technology is widely used in multidisciplinary research related to visual perception, such as neuroscience, cognitive psychology, sociology, marketing, geography, industrial design, urban planning, architecture and landscape design and evaluation [18,48,49,50,51,52,53,54].

Eye-tracking technology quantifies the visual attention and cognitive processes of the human eye by measuring various eye-movement indicators. Compared with the questionnaires, interviews and other methods in traditional built-environment research, the data is more objective and has the advantages of quantitative research and direct evaluation [32,55]. Researchers can infer the psychological state of participants by recording and analyzing eye-tracking data, so as to obtain more reliable results. As it is useful to reflect people’s interest in the research object through objective indicators such as fixation point, fixation duration, etc., most researchers use 2D images of the studied environment as stimuli to explore the scene attributes on evaluation. For example, it has been proven that different components of a landscape image would influence its ranking during the evaluation [54,56,57,58]. Images that contain more greenery tend to rank higher and are more likely to be favored [59]. Buildings, animals and other artificial objects may also affect how people feel about the landscape [29,60,61]. Some studies explore the influence of color proportions and openness and heterogeneity of space on visual evaluation of landscapes [62]. The in-depth study and expansion of eye tracking in landscape research has also promoted the further understanding of the aesthetic, practical and healing effects of landscape.

Although existing studies have proposed that different landscape components affect people’s preferences and evaluation results, there is a lack of discussion on how specific landscape elements influence people’s perception [31,62]. Meanwhile, the evaluation of landscape mostly focuses on its aesthetic characteristics, and a series of measuring methods quantify people’s subjective perception of landscape through rating scales. However, few studies focus exclusively on people’s emotional experience of different landscape elements [33]. In addition, most current eye-tracking studies pay attention to mountains, forest, urban green spaces and blue-green spaces, but few studies research the emotional attachment of students to campus landscape spaces. Furthermore, the existing eye-tracking studies mostly use two-dimensional pictures played on computer screens as stimuli objects, which is quite different from the experience of observing the landscape in real scenes. The visual dimension of the image is limited, and it is especially difficult to verify the effect of spatial-depth changes on eye movement. The maturity of immersive VR glasses technology such as HTC Vivo Pro provides the possibility of conducting research using three-dimensional panoramic pictures [42,63,64].

This study focuses on campus landscape space and takes a newly built multi-functional campus public landscape space called the Heart of Forest in Beijing Forestry University as the research object. GoPro was used to take three-dimensional panoramic images, combined with HTC Vive Pro VR glasses to collect eye-movement indicators. The ErgoLab platform was used to perform visual clustering and analyze the above indicators. Emotional attachment scales to measure the degree of positive and negative emotions, the degree of place attachment and the emotional attachment characteristics of specific landscape elements were constructed and used to study students’ emotional attachment to the landscape [32,33,65]. With the help of 3D eye-tracking methods, the relationships between the different type and combination of natural and artificial elements with students’ preferences were studied, and their emotional attachment to specific landscape elements was explored to further corroborate and supplement the eye-tracking results. The paper seeks to analyze the following issues:

(1)

The characteristics and differences of participants’ eye-movement behaviors according to different natural and artificial campus landscape elements and features.

(2)

The emotional attachment characteristics and differences according to different natural and artificial campus landscape elements and features.

(3)

The relationship between eye-movement behavior characteristics and emotional attachment.

The study aimed to reveal differences in students’ visual perceptions of and emotional attachment to different characteristics of campus landscape through a promoted 3D eye-tracking paradigm, and to provide design guidance for the selection and combination of natural and artificial landscape elements from a perspective that truly cares about the students’ emotional experience.

2. Materials and Methods

2.1. Research Area and Stimuli

The research area of this study is the public landscape space “Heart of Forest” of Beijing Forestry University. It is located in the central area of Beijing Forestry University and was completed and put into use in September 2020 (Figure 1).

Heart of the Forest is a landscape renewal project to improve the quality of the campus environment after the demolition of the former school hospital. This landscape space not only contains rich natural landscape elements, such as water areas, plant combinations of different heights, etc., but also contains various types of artificial landscape elements, such as rest pavilions, wooden seats, stone paths, and water canals made of rusty steel plates. In addition, the designer set up a variety of interactive landscape installations, such as signs with commemorative words and a forest museum consisting of a series of plant specimens to highlight characteristics of Beijing Forestry University. Since its completion, this landscape has become a popular public space on campus and became an important place for emotional healing during the three-year COVID pandemic period of campus closure. This makes it a perfect object for this study to explore students’ emotional attachment to the campus landscape.

Spatial photographs of the landscape have been proven to be an effective substitute for on-site research of the environment, and the laboratory environment is also better to avoid the influence of outdoor uncertainties on research results [55,66,67,68]. In addition, the maturity of head-mounted VR glasses can provide people with an immersive spatial visual experience, which has been proven to be able to effectively simulate the perception experience of visual depth and distance in space in the real environment [69,70]. In accordance with the image input format requirements of HTC Vivo Pro eye-tracking technology in VR environments, the study used GoPro to take panoramic pictures according to the spatial experience flow line of the research area in September, 2022. Ten panoramic pictures were taken to cover the main landscape space of the Heart of the Forest, which basically covered the main natural, artificial and interactive elements that make up the landscape of the research area. Since the shooting time was autumn, the characteristics of the natural elements were affected by the seasons. Adobe Photoshop was used to adjust color tolerance and image size of the 10 panoramic pictures. These pictures were then used as experimental materials in the immersive virtual reality eye-tracking experiment (Figure 2).

2.2. Participants and Experimental Process

Previous studies have proven that student data are representative in eye-tracking research [55,71,72]. Meanwhile, as the object of this study is the campus landscape environment, students are its main user group. Based on this, 98 students of different majors from Beijing Forestry University were recruited as participants. They all had normal corrected vision and had no ocular or neurological diseases. Students from both landscape design-related majors and other majors were included, as the previous studies showed that students with professional design training seemed to view landscape photos differently and have different aesthetic preferences compared with other major students. The 10 spatial panoramic pictures were played to participants in the sequence of a real experience of walking in the “Heart of Forest” landscape space, and each panoramic image was played for 30 s. Participants were allowed to view each image freely through HTC VR glasses without viewing guidance, and participants could freely rotate their heads and bodies 360 degrees for a more realistic simulation of the real landscape viewing experience.

The experimental equipment included a set of HTC VIVE Pro eye devices consisting of a pair of headset VR glasses, data cable, and power cable (HTC Corporation, Taiwan, China) and a laptop computer with a 17-inch color monitor of 1920 × 1080-pixel resolution (Figure 3). The HTC VIVE Pro eye here was a modified version of Tobii Pro VR technology that seamlessly integrates Tobii eye-tracking technology by hiding the eye-tracking sensor behind the lens. As a result, the HTC VIVE Pro eye was able to capture gaze data from the vast majority of people at a sampling rate of 120 Hz without affecting the VR experience. Eye-tracking data were transmitted via standard HTC Vive cables, eliminating the need for any external cables. Eye-tracking data, on the other hand, were processed using the Tobii Eye Chip to minimize CPU load. Steam VR platform was used to drive VR programs. All eye-tracking data were recorded and analyzed through the ErgoLab 3.0 platform from Beijing Kingfar Technology Co., Ltd. (Beijing, China).

The experimental procedure consisted of three parts: resting data collection, free viewing of VR panoramic pictures, and filling out emotional attachment scales (Figure 4). The total duration was about 13 min per person. The resting process required participants to sit still with their eyes closed for 3 min to restore a relatively calm emotional status at the beginning of the experiment as much as possible and relieve their visual fatigue. After resting, 10 panoramic pictures began to be played sequentially, each picture played for 30 s, and the participants could stand and turn their bodies slightly to immerse themselves freely in the virtual reality pictures in VR glasses. After the second stage of viewing, the participants took off their VR glasses and immediately filled out the emotional attachment scale.

2.3. Eye-Tracking Index Selection and Emotional Attachment Scale Construction

Eye tracking refers to the process of automatic positioning of the pupil center and fixation point of the eye with the help of related instruments. Three types of eye-movement characteristics are commonly recorded including fixation, saccade, and pursuit, and its specific description indicators are composed of two dimensions of time and space, such as first fixation duration, mean fixation duration, total fixation time, fixation sequence, number of saccades, etc. Since vision can reflect people’s attention and emotional tendencies and is even regarded as a way of thinking, and eye-tracking signals can provide a natural and effective way to observe user’s visual behavior, it is conducive to exploring the landscape characteristics of the research area that trigger people’s emotional experience in the process of interacting with the landscape more accurately and objectively. Based on the environmental characteristics of the “Heart of Forest” landscape space, this study collected and analyzed five indicators in the eye-tracking data, including time to first fixation (TTFF), fixation count (FC), mean fixation duration (MFD), visit count (VC), and mean pupil diameter (MPD), which were proven in the previous studies to be closely related to the user’s emotional experience. Their definition and corresponding emotional representations are shown in

Table 1. Definition and corresponding emotional representations of chosen eye-tracking indicators.

Eye-Tracking Indicators	Definition	Corresponding Emotional Representations
TTFF (time to first fixation)	The amount of time that it takes a participant to look at a specific AOI from stimulus onset.	TTFF can represent both bottom-up stimulus-driven and top-down attention-driven searches. The shorter the TTFF, the stronger the attraction of the object to the participant, which is more conducive to the elicitation of emotions.
FC (fixation count)	The total number of fixations generated by participants when viewing each AOI.	A higher FC indicates a stronger interest in the corresponding AOI, which may correspond to a stronger emotional attachment of participants.
MFD (mean fixation duration)	The average length of fixation generated by participants when viewing each AOI.	The longer the MFD, the higher the participant’s attention to landscape elements or spaces, possibly indicating greater interest and emotional attachment.
VC (visit count)	The times a participant returned their gaze to a particular spot, defined by an AOI.	The VC indicates the landscape element or space which repeatedly attracted the participant (for better or worse). A higher VC indicates that the AOIs were more attractive to participants, corresponding to a stronger emotional attachment experience.
MPD (mean pupil diameter)	The average value of the change in pupil size when participants viewed the 10 landscape panoramic pictures.	Changes in MPD are directly associated with changes in participants’ emotions, but do not necessarily correspond to positive or negative emotions.

.

The eye-tracking process was recorded using HTC VR VIVE PRO Eye (head-mounted VR glasses with Tobii eye-tracking technology, from HTC Corporation, Taiwan, China), and the experimental data were recorded and exported after individual calibration of the viewpoints using ErgoLab 3.0 platform (multimodal data collection and analysis platform developed by Beijing KingFar International Inc., Beijing, China). Areas of interest (AOI) were delineated by element type. Based on the purpose of this study, the AOI that described the natural elements of the landscape included water, arbors, shrubs and lawns, and stones. AOI describing the artificial elements of the landscape included natural material paving (wood pavement, slate and stone paving), artificial material paving (cement pavement), trace material (rusty steel plates), architectural structures (pavilions and chairs), symbols (logo, picture, herbarium, etc.). The AOIs of the 10 panoramic pictures were drawn and their representative pictures of scenes are shown in Figure 5. According to the research purpose to study students’ emotional attachment as well as the basic visual process from initial gaze attraction to scanning and back to gaze, the indicators chosen to record included time to first fixation (TFF), fixation count (FC), mean fixation duration (MFD), visit count (VC) and mean pupil diameter (MPD), which were supposed to be closely related to people’s interests and emotional attachment to the environment.

In addition, a series of emotional attachment scales including place attachment scale (PA) [73,74], positive and negative analysis scale (PANAS) [75], and specific landscape characteristics emotional attachment scale were used to explore the degree as well as dimension of students’ emotion when interacting with different landscape elements. The items of the three scales and how they are measured are shown in

Table 2. PANAS Scale (Participants were asked to indicate to what extent they felt different emotions at the exact moment when they experienced the panoramic pictures of Heart of Forest in VR. The Likert scale was used as measurement method. 1: Very slightly or not at all; 5: Extremely).

	Very Slightly or Not at All				Extremely
1. interested	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
2. distressed	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
3. excited	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
4. upset	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
5. strong	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
6. guilty	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
7. scared	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
8. hostile	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
9. enthusiastic	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
10. proud	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
11. irritable	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
12. alert	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
13. ashamed	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
14. inspired	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
15. nervous	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
16. determined	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
17.attentive	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
18. jittery	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
19. active	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]
20. afraid	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]

,

Table 3. PA Scale (From Scannell and Gifford version. Participants were asked to indicate to what extent they felt different emotions at the exact moment when they experienced the panoramic pictures of Heart of Forest in VR. The Likert scale was used as measurement method. 1: Very slightly or not at all; 7: Extremely).

	Very Slightly or Not at All						Extremely
Everything about this place is a reflection of me.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
This place says very little about who I am.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I feel relaxed when I’m in this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I feel happiest when I’m in this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
This place is my favorite place to be.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I really miss this place when I’m away from it for too long.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I feel that I can really be myself in this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
This place is the best place for doing the things I enjoy most.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
For doing the things that I enjoy most, no other place can compare to this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
This place is not a good place to do the things I most like to do.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
This place reflects the type of person I am.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
As far as I am concerned, there are better places to be than in this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
The spiritual nature of the area ties me to this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I feel that this place is my home.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I intend to continue staying in or around this place for the next few years.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I have the feeling that this place constitutes a security base for me.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I feel a connection to the visual landscape of the area.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
This place is an important part of my life.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I feel proud of this place.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
I am totally involved and committed to my school, classmates and neighborhood.	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]

and

Table 4. Specific landscape characteristics emotional attachment scale (Participants were asked to indicate what extent of attachment they felt at the exact moment when they experienced the panoramic pictures of Heart of Forest in VR. The Likert scale was used as measurement method. 1: Very slightly or not at all; 7: Extremely).

	Very Slightly or Not at All						Extremely
1. material	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
2. color	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
3. natural-related feature	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
4. form and structure	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
5. privacy	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
6. diversity	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
7. sociability	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
8. regionality	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
9. playability	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
10. uniqueness	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]
11. changeability	1 [ ]	2 [ ]	3 [ ]	4 [ ]	5 [ ]	6 [ ]	7 [ ]

.

The Chinese version of the PA scale and PANAS scale was revised and used to avoid cultural gaps [33,76]. The landscape characteristics attachment scale was constructed based on the research purpose as well as the specific characteristics of the Heart of Forest, which consisted of indicators describing not only natural and artificial characteristics but also students’ social and interactive needs for campus landscape, including material, color, natural elements, form and structure, privacy, diversity, sociability, territoriality, playability, uniqueness, and changeability (Table 4). The results of the scales could be further coupled with the data of the above eye-tracking experiment to explore the students’ emotional attachment to the landscape space more comprehensively. Previous studies had proven the validity and reliability of the scales [32,33,65,74,76].

2.4. Data Analysis

The VR eye-tracking data and attachment scale data were analyzed with ErgoLab and SPSS v26.0 software (IBM, New York, NY, USA). The area of interest (AOI) had been drawn using ErgoLab before data analysis to help identify and convert the raw data of eye tracking. For example, when the same kinds of elements appeared in different parts of the panoramic image (i.e., stone 1, stone 2, etc.), they would be combined into one variable (i.e., stone). According to previous studies [57], the image areas that did not represent elements of the landscape, such as the buildings and sky outside the Heart of Forest, were excluded from the AOI drawing and analysis. The chosen indices of TFF, FC, MFD and VC were calculated for all participants for each landscape element, while the MPD were calculated for all participants for the landscape as a whole. The eye-tracking data were processed by ErgoLab and visualized as heat maps and trajectory maps of the 10 panoramic images. Specifically, the procedures of eye-tracking analysis and attachment-scale analysis were as follows:

The overall data of both eye-tracking experiment and emotional attachment scale were tested using SPSS to explore whether they were normally distributed, in order to choose the appropriate methods for analysis.

According to the number of valid samples in the experiment (excluding 8 samples with missing or damaged data, a total of 90 valid samples were left for analysis), the normality test of eye-tracking data and scale data was carried out combining the Kolmogorov–Sminov test (K–S test), descriptive method test and histogram. The data of TFF, FC, RN, MFD, MPD and scale data had an approximately normal distribution. Due to the uneven variance of the data, the Welch test was selected in ANOVA for analysis. Cronbach’s alpha coefficient was calculated to confirm the reliability of attachment-scale data.

ANOVA (with post hoc comparisons) was used to determine differences in the four indicators representing visual behavior of natural and artificial landscape elements. Mean and standard deviation of visual behavior scores was used to identify the specific natural and artificial element types that had relatively significant influence on students’ preference. The results were then coupled with above intuitive eye-tracking heatmaps to further analyze the influence of landscape elements on human attraction to real scenes.

Spearman’s rho analysis was used to determine the correlations between visual behaviors and attachment to specific landscape characteristics. Meanwhile, it was also used specifically for scale data analysis to explore the correlations between place attachment, positive effect, negative effect and attachment to detailed landscape characteristics.

The above content constructs the framework of this article (Figure 6).

3. Results

A total of 98 participants including both undergraduates and postgraduates from Beijing Forestry University had been recruited in the study. Their ages ranged from 17 to 24. The final sample with valid data consisted of 90 students, including 15 male students and 75 female students.

3.1. Influence of Different Types of Landscape Elements on Participants’ Visual Behavior

3.1.1. Differences in the Eye-Movement Indicators for Natural and Artificial Landscape Elements

ANOVA in SPSS was used to analyze the chosen four eye-movement indicators closely related to the participants’ emotional experience of ten specific elements in the two major types of natural and artificial landscape. The results showed that there were differences in the eye-movement indicators for the 10 specific landscape elements that were subordinate to natural and artificial landscape (p < 0.05). Among them, different landscape elements showed different degrees of difference, indicating that different individuals had different visual reflections, attention and preferences for natural and artificial elements (

Table 5. Differences in eye-movement indicators for different types of landscape elements.

		TTFF (s)	MFD (s)	FC (n)	VC (n)
Overall Difference		0.000 **	0.000 **	0.000 **	0.000 **
Difference among specific landscape elements
Natural elements	Natural waters (a)	81.31 (b**, c**, d**, e**, f**, g**, h**, j**)	0.86 (b**, c**, d**, e**, g**, h**, i**, j**)	35.70 (b**, c**, d**, e**, f**, g**, h*, i**)	16.17 (b**, c**, d**, e**, f**, g**, h**, i**)
	Arbors (b)	194.90 (a**, c**, d**, e**, f**, g**, h**, i**, j**)	3.79 (a**, d**, e**, f**, g**, h**, i**, j**)	187.97 (a**, c**, d**, e**, f**, g**, h**, i**, j**)	62.78 (a**, c**, d**, e**, f**, g**, h**, i**, j**)
	Shrubs and lawns (c)	340.24 (a**, b**, d**, e**, f**, g**, h**, i**, j**)	3.68 (a**, d**, e**, f**, g**, h**, i**, j**)	85.22 (a**, b**, d**, e**, f**, g**, h**, i**, j**)	42.46 (a**, b**, d**, e**, f**, g**, h**, i**, j**)
	Stones (d)	112.29 (a**, b**, c**, e**, f**, g**, h**, j**)	1.38 (a**, b**, c**, e**, g**, h**, i**, j**)	17.90 (a**, b**, c**, g**, h**, i**, j**)	9.72 (a**, b**, c**, e**, g**, h**, i**, j**)
Artificial elements	Wood pavement (e)	24.41 (a**, b**, c**, d**, f**, h**, i**)	0.57 (a**, b**, c**, d**, f**, i**, j**)	15.60 (a**, b**, c**, f*, h**, i**, j**)	6.09 (a**, b**, c**, d**, f**, g**, i**, j**)
	Slate and stone pavement (f)	34.37 (a**, b**, c**, d**, e**, h**, i**, j**)	0.83 (b**, c**, d**, e**, g**, h**, i**, j**)	22.69 (a**, b**, c**, e*, g**, h**, i**, j**)	11.33 (a**, b**, c**, e**, g**, h**, i**, j*)
	Cement pavement (g)	26.72 (a**, b**, c**, d**, f**, h**, i**)	0.54 (a**, b**, c**, d**, f**, i**, j**)	7.90 (a**, b**, c**, d**, e**, f**, h**, i**, j**)	3.87 (a**, b**, c**, d**, e**, f**, h**, i**, j**)
	Rusty steel plate (h)	12.79 (a**, b**, c**, d**, e**, f**, g**, i**, j**)	0.40 (a**, b**, c**, d**, f**, i**, j**)	48.26 (a*, b**, c**, d**, e**, f**, g**)	6.91 (a**, b**, c**, d**, f**, g**, i**, j**)
	Pavilions and chairs (i)	93.60 (b**, c**, e**, f**, g**, h**, j**)	2.09 (a**, b**, c**, d**, e**, f**, g**, h**)	50.96 (a**, b**, c**, d**, e**, f**, g**)	22.59 (a**, b**, c**, d**, e**, f**, g**, h**, j**)
	Symbols (logo, picture, herbarium, etc) (j)	21.47 (a**, b**, c**, d**, f**, h**, i**)	1.87 (a**, b**, c**, d**, e**, f**, g**, h**)	44.84 (b**, c**, d**, e**, f**, g**)	14.01 (b**, c**, d**, e**, f*, g**, h**, i**)
n		90	90	90	90

* Significant difference with p < 0.05. ** Significant difference with p < 0.01.

).

Specifically, with regards to the result of the mean time to first fixation, natural elements overall took longer than artificial elements, which indicated that participants’ attention was more easily attracted by artificial elements. Among them, the TTFF of the rusty steel plates was the shortest (12.79 s), indicating that they were the most attractive visual focus. Symbols, wood pavements and cement pavements also had a relatively shorter TTFF (21.47 s, 24.41 s, 26.72 s). Interestingly, the TTFF of the seats and pavilions in the Heart of Forest was longer than the aforementioned artificial paving material (93.60 s), indicating that their attraction to people was weaker than that of the rusty steel plates, wooden paving and other materials with unique textures and colors. On the one hand, this might be related to the relatively larger area of artificial paving material used in the Heart of Forest, and on the other hand, it also showed that the texture and color of the material were conducive to arousing people’s interest in the landscape space experience. Meanwhile, as for natural elements, the shrubs and lawns had the longest TTFF (340.24 s), indicating that their visual attraction to participants was the weakest in the Heart of Forest, which might be caused by their lower spatial distribution. The TTFF of the pond was the shortest among the natural elements, and even slightly smaller than the pavilions and seats in the artificial elements, indicating that the water area had an important value in arousing the interest of users in the construction of the landscape. Based on the specific differences between different elements in multiple comparisons, it could be seen that for artificial elements, except for the rusty steel plates, the participants’ interest in the wood pavement, cement pavement, slate and stone pavement and symbols showed little difference (p > 0.05). As for natural elements, the participants’ interests were all significantly different.

Secondly, with regards to the overall result of the mean fixation duration, there were significant differences between different landscape elements. In general, artificial features were more easily recognized through visual behavior than natural features. The MFD of four of the artificial elements was less than 1 (0.57 s, 0.83 s, 0.54 s, 0.40 s), among which that for the rusty steel plates was the shortest, indicating that the visual cognition processing time required was short, which means they were easier to understand and recognize. On the contrary, the MFD of the arbors and shrubs and lawns was much longer (3.79 s, 3.68 s), which might relate to a longer cognition processing time. This means that participants might spend more time dealing with cognitive processes related to plants. On the one hand, this might be due to the rich visual information of the plants themselves; on the other hand, this might also be because most of the participants were landscape architecture students, and they paid more attention to the characteristics of plants in the landscape space. Based on the specific differences between different elements in multiple comparisons, wood pavement, cement pavement and symbols showed little difference (p > 0.05), indicating that there was no clear difference in their ability to attract participants’ attention. However, objectively speaking, the visual information richness of the symbols was much higher than that of the pavement, so the smaller difference might also be due to the larger material area of the pavement and the visual focus was more likely to fall on it in the eye-tracking experiment.

Thirdly, with regards to the result of the mean fixation counts, there were also significant differences between artificial and natural landscape elements. Natural elements had a higher fixation count compared to the artificial ones, among which the arbors had the highest fixation count (187.97), indicating the hardest cognitive effort made by participants when looking at them. Combined with the results of the aforementioned MFD, among the natural elements, the MFD for the pond was shorter than that for stone, but the FC was higher than that for stone, indicating that although the water area in the landscape space was easy to identify, it was able to stimulate a deeper visual cognitive processing, which might indicate that the participants were more interested in the natural water area. Similarly, among the artificial elements, although the MFD for the rusty steel plates was the shortest, their FC was higher, which also indicated that their texture, color and other characteristics had aroused more interest from the participants. In addition, the results of multiple comparisons showed that there were few differences between the symbols, rusty steel plates and pavilions and chairs, indicating that participants paid similar attention to these elements.

Last but not least, with regards to the result of the mean visit counts, there were significant differences between artificial and natural landscape elements, and the natural elements had overall a higher VC than the artificial ones. The arbors had the highest VC among all the elements (62.78), indicating that participants visited them most often, which might prove that they were the most interesting to the participants. The VC of the shrubs and lawns ranked second (42.46), which proves that participants’ interest in them might not have been influenced by their inconspicuous spatial location. As for artificial elements, pavilions and chairs had the highest VC among the artificial elements, but the VC of the rusty steel plates was relatively low. Combined with the results of the aforementioned indicators, it showed that the rusty steel plates might have easily attracted the attention of participants quickly, but they might not necessarily have provided sustained interest.

3.1.2. Differences in Fixation Characteristics in Different Landscape Spaces according to AOI Heat Map

The AOI heat maps were used to visualize the position and distribution of fixation points in the ten spatial sequences of the landscape space in the Heart of Forest. The results showed that participants’ attention degree and area differed for different kinds of landscape elements and different spatial compositions (Figure 7).

First, with regard to the fixation distribution, almost all 10 panoramas were concentrated in the central area of the height of the human viewpoint. The spatial vanishing point with deep depth especially attracted the visual focus of the participants. At the same time, the distance from the observation object and the change in light and shadow also affected participants’ fixation distribution. This might be related to the three-dimensional perception of stimuli by participants in immersive VR experiences. The way in which participants’ visual focus converged on elements of the built environment was much closer to the real-world spatial experience.

Specifically, with regard to participants’ interest in different kinds of landscape elements, in open spaces containing pavilions and symbols, the fixation was most focused on these elements that stand out from the natural landscape or are more readable and interactive (Figure 7(1,2,6)). For example, even in a low or marginal spatial position, text symbols were still able to attract more focus. In semi-enclosed or enclosed spaces where interactive elements were not visible, the fixation distribution was more concentrated on water spaces (Figure 7(3)).

Meanwhile, with regard to participants’ AOI position as well as the landscape elements distribution, the visual information was mainly concentrated in the upper part of the panoramic picture, the interactive and unique elements and the extinction point of space. This not only indicates that landmark buildings and interactive landscape elements generally form the visual focus in a scene, but also reveals that the perspective of distribution of landscape elements in a space affects their appeal to people.

3.2. Influence of Different Landscape Characteristics on Participants’ Emotional Attachment

3.2.1. The Overall Features of Participants’ Emotional Attachment to the Heart of Forest

The descriptive statistical results help reveal the overall features of participants’ emotional attachment (

Table 6. Descriptive statistical results of the overall features of emotional attachment.

	α	Mean	SD
Place Attachment	0.851	4.39	0.77
Positive affect	0.775	2.54	0.59
Negative affect	0.790	1.27	0.33
Overall attachment to landscape characteristics	0.843	4.96	0.83

). As is shown, the Cronbach’s α of all 4 scales was greater than the fundamental value of the reliability coefficient of 0.7, which identified their high internal consistency. The results showed that participants generally had a strong place attachment to the Heart of Forest (MPA = 4.39 > 3.5). Specifically, the promotion of positive emotional experiences by the space was more pronounced than negative emotional experiences (Mpositive = 2.54 > Mnegative = 1.27). But the average degree of positive effect only exceeded its median value, indicating that participants’ positive feelings were not very strong. Meanwhile, the specific landscape characteristics had contributed greatly to the establishment of the emotional attachment, as the mean attachment exceeded its median value significantly (MOATLC = 4.96 > 3.5).

In addition, the degree of attachment to landscape space and specific characteristics were quite different among different participants (SD = 0.83, SD = 0.77), which indicated that in addition to the built-environment components, participants’ attachment to landscape might also be affected by more complex factors such as different individual experiences and cognition.

3.2.2. The Correlation between Place Attachment, Positive and Negative Effect and Attachment to Specific Landscape Characteristics

According to the correlation analysis results (

Table 7. Correlation analysis results between emotional attachment and different characteristics of landscape.

Variable	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. positive effect	1.000
2. negative effect	0.059	1.000
3. place attachment	0.507 **	−0.104	1.000
4. material	0.276 **	−0.036	0.312 **	1.000
5. color	0.192	−0.171	0.465 **	0.401 **	1.000
6. natural-related feature	0.155	−0.155	0.363 **	0.494 **	0.630 **	1.000
7. form and structure	0.318 **	−0.193	0.417 **	0.468 **	0.366 **	0.491 **	1.000
8. privacy	0.309 **	−0.152	0.403 **	0.040	0.352 **	0.064	0.237 *	1.000
9. diversity	0.246 *	−0.321 **	0.488 **	0.313 **	0.409 **	0.448 **	0.485 **	0.361 **	1.000
10. sociability	0.307 **	−0.083	0.548 **	0.311 **	0.323 **	0.240 *	0.392 **	0.122	0.573 **	1.000
11. regionality	0.311 **	−0.315 **	0.370 **	0.236 *	0.216 *	0.269 *	0.361 **	0.256 *	0.414 **	0.314 **	1.000
12. playability	0.281 **	−0.107	0.338 **	0.278 **	0.122	0.221 *	0.319 **	0.407 **	0.518 **	0.248 *	0.300 **	1.000
13. uniqueness	0.225 *	−0.302 **	0.446 **	0.316 **	0.433 **	0.451 **	0.382 **	0.251 *	0.498 **	0.407 **	0.491 **	0.465 **	1.000
14. changeability	0.113	−0.299 **	0.164	0.073	0.303 **	0.184	0.196	0.299 **	0.349 **	0.265 *	0.304 **	0.362 **	0.469 **	1.000
15. Overall attachment to landscape characteristics	0.362 **	−0.288 **	0.596 **	0.534 **	0.637 **	0.606 **	0.660 **	0.494 **	0.775 **	0.613 **	0.599 **	0.613 **	0.729 **	0.553 **	1.000

* p < 0.05 (2-tailed), ** p < 0.01 (2-tailed).

), the overall attachment of participants to the Heart of Forest was significantly positively correlated with positive effects (0.507 **), negatively correlated with negative effects (−0.104), and significantly positively correlated with the specific landscape characteristics (0.596 **), indicating that the landscape space itself effectively promoted the establishment of participants’ place attachment and their formation of positive emotional experiences. This reveals the emotional attachment value of the landscape design of the Heart of Forest. With regard to the specific landscape characteristics, the color, diversity and sociability of the Heart of Forest were significantly correlated with participants’ place attachment.

Among them, the color of the landscape showed a significant positive correlation with nature-related features (0.630 **), which further indicated that it was the color from nature that promoted the formation of the aforementioned place attachment. Meanwhile, any two of the three characteristics of material, natural-related features and form and structure were significantly positively related to each other (0.494 **, 0.468 **, 0.491 **), indicating that the pavilions and seats that were built from natural materials or reflecting the structural mechanical characteristics of natural materials were conducive to the formation of people’s emotional attachment to the landscape space. The form and structure were also positively related with the diversity of the Heart of Forest (0.485 **), indicating that diverse structural forms of multiple pavilions might also be one of the reasons for promoting emotional attachment. In addition, the natural-related features showed a significant positive correlation with the uniqueness of the landscape space (0.451 **), proving that the natural landscape design, plant configuration and water space design promote its uniqueness.

3.3. Relationship between Participants’ Emotional Attachment and Visual Behavior Concerning Different Types of Landscape Elements

3.3.1. Relationship between Visual Behavior and Emotional Attachment Indexes

According to the results of correlation analysis between participants’ visual behavior and emotional attachment indexes (

Table 8. Correlation analysis results between visual behavior and emotional attachment.

	FC	TTFF	MFD	VC	MPD
Positive Effect	−0.065	0.142	0.008	0.033	−0.317 **
Negative Effect	0.143	0.031	0.016	−0.067	0.139
Place Attachment	−0.032	0.042	0.020	0.176	−0.225 *
Overall Attachment to Landscape Characteristics	−0.085	−0.039	−0.108	0.255 *	−0.117

* p < 0.05 (2-tailed), ** p < 0.01 (2-tailed).

), it can be seen that the positive effect on the participants and their overall place attachment are significantly correlated with MPD. Specifically, pupil diameter showed a significant negative correlation with positive affect and place attachment, that is, the larger the pupil, the weaker the degree of positive emotional experience and place attachment. Psychological studies had shown that emotional arousal activates the autonomic nervous system, which in turn causes pupil changes, and compared with positive emotional stimuli, people were more sensitive to negative emotional stimuli, and were more likely to produce pupil dilation and expansion for a longer time. This further proves that viewing various landscape spaces in the Heart of Forest brought participants a strong positive emotional experience and promoted the formation of their place attachment.

Meanwhile, the results also showed that the participants’ emotional attachment was unrelated to fixation count, time to first fixation, mean fixation duration and visit count. This was consistent with the previous studies that people’s psychological cognition and preferences were not correlated with fixation duration and fixation count [55,71,77].

3.3.2. Relationship between Visual Behavior and Emotional Attachment Evaluation Factors of Specific Landscape Characteristics

In order to find the correspondence between different types of landscape elements in the eye-tracking experiment that attracted participant’s attention and their emotional attachment degree to different landscape characteristics derived from the results of the scale, the higher fixation-rate elements in the above heat map were matched with participants’ responses in the emotional attachment scale (

Table 9. Participant’s attention and attachment factors in the Heart of Forest.

Space Sequence in VR Experience	Landscape Elements on Which Gaze Was Focused (in Descending Order)	Relating Landscape Characteristics in Emotional Scale
1. Water space with pavilion	Pavilion, water, shrubs, stones, arbors	Form and structure, natural-related features, material, color
2. Water space by pavilion	Water, pavilion, chair, stone, arbors	Natural-related features, form and structure, sociability
3. Water space inside pavilion	Arbor in the center, shrubs, pavements, water, stone	Natural-related features
4. Linear waterfront space with plants	Water, shrubs	Natural-related features
5. Linear waterfront space with symbols	Symbols, water, shrubs	Regionality, uniqueness, natural-related features
6. Open lawn space with pavilion and symbols (far from pavilion)	Pavilion, water, symbols, arbor (ginkgo biloba with yellow leaves)	Form and structure, natural-related features, regionality, uniqueness
7. Open lawn space with pavilion and symbols (close to pavilion)	Symbols, pavilion, arbors (ginkgo biloba with yellow leaves)	Uniqueness, diversity, regionality, form and structure, natural-related features, color
8. Pavilion inside space with symbols (specimen exhibition wall)	Symbols of specimen exhibition wall, symbols on the lawn	Uniqueness, regionality, sociability, playability
9. Semi-enclosed pavilion space with maple tree and symbols	Symbols (interactive), arbors in distance, arbors nearby (the maple)	Regionality, uniqueness, natural-related features
10. Pavilion inside space with chairs	Symbols (outside pavilion), pavilion (inside structure and furniture)	Sociability, uniqueness, form and structure

).

The panoramic images of Sequences 1–3 (Figure 2) simulate the visual experience of the participants entering the Forest Pavilion by the Forest Marsh from the outside. The results of the heat map show that spatial perspective significantly affected the visual focus distribution of the participants, and the artificial elements were more attractive than the natural ones. However, the visual focus does not directly correspond to the strongest emotional experience; for example, the results of the scale showed that nature-related elements and their colors promoted place attachment more strongly than the formal structure of artificial elements, and the form and structure of artificial elements more effectively promoted the formation of a positive effect on the participants.

The panoramic images of Sequences 4–5 (Figure 2) simulate the visual experience of walking along the stream of the Rain Garden from the Forest Marsh to the Museum in the Forest pavilion. The heat map showed that the water and the symbols drew the visual attention of participants among the natural and artificial elements of the two sequences, respectively. This was consistent with the results of the emotional attachment scale. In addition to the emotional eliciting value of the natural-related features as in the above sequences, the results indicate that artificial symbols promote people’s attachment by arousing people’s emotional resonance through their regionality and uniqueness; for example, the monument of the original site of the school hospital in sequence 5 turned out to draw the main visual attention of the participants.

The panoramic images of Sequences 6–7 (Figure 2) simulate the visual experience when walking from the Rain Garden to the Museum in the Forest pavilion. The participants’ attention was mainly focused on artificial elements as was revealed by the heat map, such as the pavilion and symbols with words. This was consistent with the result of the emotional attachment scale showing that the form and structure of the artificial elements, as well as their territoriality and uniqueness, were the main promoting characteristics of emotional attachment. In addition, the heat map of Sequence 7 showed that the yellow leaves of the ginkgo biloba tree in autumn attracted more attention than other natural elements, corresponding to the emotional promotion value of color associated with natural-related features in the attachment scale.

The panoramic images of Sequences 8–10 (Figure 2) simulate the visual experience inside the Museum in the Forest pavilion and from the interactive installation of Whisper in the Forest. The artificial elements were the main visual focus of the participants in these sequences. Specifically, when the wall of plant specimens was in the field of vision, the specimens attracted the most visual attention of the participants. When the artificial elements in the field of vision were mostly chairs, lamps and other furniture, the participants’ visual attention tended to concentrate more on the structures and plants outside the pavilion. This was also related to the effect of the spatial depth on the visual focus brought by the VR experience. The results of the scale further indicate that the participants preferred these elements because they represent the uniqueness of Beijing Forestry University. The sociability and playability of this public space also contributes to the emotional experience it provides, while in the Whisper in the Forest, the participants’ visual attention was mainly focused on the interactive installation and surrounding notes hanging on the maple tree. However, the results of the scale showed that the changeability of the installation failed to promote participants’ emotional attachment, indicating that the interactivity had not been realized and was consistent with the current situation that the installation is not functioning properly in reality.

4. Discussion

4.1. Different Artificial and Natural Landscape Elements Have Differences in Observation Mode

Previous studies have shown that people have different visual behaviors according to different types of landscape elements [29,55,57], which was also found in the results of this study. Based on this, this study further reveals the difference between the observation modes of artificial elements and natural elements, and the influence of spatial perspective on the visual behavior of landscape elements which is highlighted in the virtual reality experimental environment supported by VR glasses. This is something that previous eye-tracking experiments using two-dimensional images as stimuli have not been able to show.

In general, the TTFF, MFD and FC of artificial elements were smaller than those of natural elements, indicating that their visual attractiveness and recognizability were higher. On the one hand, it may be related to the greater proportion of natural elements in the landscape environment, and on the other hand, it may also be related to the fact that the participants are more students of landscape architecture and are more interested in the setting ratio of natural plants. This shows that the setting of artificial elements in the landscape environment is very important for the arousal of people’s interest, but it will also be affected by individual experience or preferences.

Specifically, with regard to the different artificial elements in the Heart of Forest, it can be found that materials with rich textures such as rusty steel and wooden pavement and symbols such as text logos and plant specimens in the Heart of Forest are most likely to arouse people’s attention and interest. For the former, on the one hand, it shows that the use of textured materials in campus landscape design is conducive to attracting students’ attention, and on the other hand, it may also be related to the large area occupied by these materials in the visual scope. For the latter, it shows that the setting of landscape structures that represent the cultural and regional characteristics of the campus might have promoted students’ interest in the landscape by improving the interactivity of the landscape.

With regard to the different natural elements in the Heart of Forest, water area was the most appealing element to draw participants’ visual attention in almost all the space sequences. Interestingly, the reflection of plants in the waters attracted even more attention than the plants themself. The stones and slabs set in the landscape space were also more likely to arouse people’s interest quickly when they first looked at the landscape, but this interest did not last long. The former may be related to the uniqueness of the stone in terms of color, spatial location, and type in the landscape space, while the latter may be due to the stone having no more characteristics that could attract people’s further observation. The arbors and shrubs were generally similar in terms of visual attraction to people, but the arbors were superior. This may be due to the fact that some of the ginkgo biloba, maple trees, etc., appear with more diverse and striking colors in autumn, which can also be seen in the AOI heat map. On the other hand, arbors occupy a larger proportion of the area in the panoramic pictures, and their positions are more layered in the field of vision, which more easily attract people’s attention in VR eye-tracking experiments that are closer to the real-world visual experience. This not only confirms once again that the setting of plants in landscape design should consider the form, color and other states of their whole life cycle, but also emphasizes that as the most important landscape elements, plants themselves also have a three-dimensional spatial volume. So, the spatial position of plants in the landscape also has a significant impact on people’s feelings, for example, so considering the viewing distance and scale when choosing and designing plants is conducive to enhancing their attractiveness.

4.2. Different Artificial and Natural Landscape Elements Trigger Different Levels and Degree of Emotional Attachment

By analyzing the degree and direction of emotional attachment between participants and the landscape space of the Heart of Forest and its specific characteristics, this study confirms the effectiveness and importance of campus landscape space in the construction of students’ emotional experience and place attachment. In general, place attachment was significantly positively correlated with positive effects and negatively correlated with negative effects, and the physical, interactive and social characteristics of the Heart of Forest all significantly promoted the formation of students’ attachment to the space. These are consistent with previous studies. Besides, the correlation analysis of the emotional attachment scale also revealed the following three parts:

(1)

The spatial composition and temporal characteristics of natural elements in the landscape significantly affect people’s emotional experience.

The results of the scale showed that the color characteristics associated with plants had a significant contribution to the participants’ attachment to place and the formation of positive emotions, and the shooting time of the panoramic pictures coincided with the autumn when the plants were the most colorful, which further highlighted the value of diverse colors in plant design of school landscapes. This is also corroborated by the results of the eye-tracking results above.

In addition, the significant correlation between natural features and structure and forms proves the role of spatial composition of natural elements in promoting emotional attachment, indicating that the spatial composition of natural elements is also a factor affecting participants’ emotional perception, which is consistent with the above VR eye-tracking results. This finding has not been emphasized in previous studies, probably because most studies use two-dimensional pictures instead of real landscape space for research, but the process of human observation of space is three-dimensional; for example, the visual vanishing point in space significantly affects people’s perception of space.

(2)

The diversity of artificial elements and the communicable public space they provide are main factors affecting the degree of students’ emotional attachment.

This highlights the functional value of the campus landscape and shows that the campus landscape space should not only provide emotional healing through visual interaction, but also provide a space for college students to socialize freely.

(3)

The use of regional and unique artificial elements can significantly enhance people’s emotional attachment to landscape space.

The results of the scale revealed that the use of regional and unique artificial elements in the campus landscape space could strengthen students’ emotional attachment to it. The memorial construction that responds to the history of the campus can be designed to trigger the emotion of memory or commemoration, such as the monument of the university hospital in the landscape sequence of the Heart of Forest. And the sense of belonging of students can also be enhanced by designing landscape nodes with campus characteristics, such as the plant specimen wall with the specific characteristics of Beijing Forestry University in the Heart of Forest.

(4)

Whether the interactivity of artificial elements is conducive to emotional experience depends on their later maintenance.

Interactive artificial elements in landscape spaces are often used to enhance people’s interest and positive emotional experiences. However, whether it can be a positive promotion is closely related to the later maintenance. For example, the uniquely designed “Secret Whispers in the Forest” space in the Heart of Forest is equipped with devices that can store “secret” and “memory” voices. The original intention was to provide interesting interactions to deepen students’ emotional attachment to the campus, but because it has not been well maintained in the later stage, the equipment could not function normally, and instead brought a negative emotional experience to students.

4.3. Connection and Interaction between Visual Behavior and Emotional Attachment to Landscape Space

In previous studies on natural landscapes, cultural tourism scenes and public spaces, there is no unified conclusion on whether the relationship between eye-movement indicators and emotional and cognitive evaluation of landscape is significantly correlated. For example, some studies have shown that fixation count and mean fixation duration are significantly correlated with evaluation results, while there are also studies that have not clearly found a correlation between them. The mean pupil diameter, on the other hand, has been significantly correlated with emotional and cognitive indicators in most previous studies. The results of this study show that the fixation count, time to first fixation, mean fixation duration and visit count have no correlation with different dimensions and degrees of emotional attachment, but the mean pupil diameter significantly correlated with positive effect and place attachment. However, although the former does not show a significant correlation at the level of data indicators, and the eye-movement index and the emotional attachment scale are systems of different dimensions, based on interdisciplinary theories from psychology and neuroscience, we believe that there is a certain degree of mutual influence and constraints between them. Therefore, the relationship between visual behavior and emotional attachment to the landscape space and its characteristics is analyzed using experimental results to explore their connection further.

(1)

The connection between different natural and artificial visual landscape elements and students’ emotional attachment to landscape characteristics.

As is shown in the previous study, artificial elements in landscape are more attractive to people’s eyes, even though they occupy a small part of the space. For example, some research [55,67,68] has shown that buildings in the natural environment always attract more attention. The same results have been revealed in this study. Based on this, this study further found that the artificial elements focused on by the human eye are related to the emotional attachment established, but the degree and dimension depend on the specific characteristics of the artificial elements. For example, the more cultural, regional and unique artificial elements of the site are not only the main fixation objects of human eyes, but also significantly promote the positive emotional experience and place attachment. Meanwhile, the interactive and playful artificial elements, such as the specially set recording device in the Heart of Forest, are also the core visual focus, but their role in promoting emotional attachment is weakened by a lack of maintenance and inability to realize their original function.

As for natural elements in the landscape space, the waterscape turned out to be the core fixation object for all human eyes in this research, which is consistent with the results by Zhou [29] as well as Gao [55] who suggest that water, especially dynamic water spaces, is always the most attractive part of a landscape. This study also found that the reflection in the water attracted more attention than the original object, which may be due to the fact that it provided a more dynamic and unique visual experience, while promoting the establishment of emotional attachment. In addition, it is interesting to note that elements that are located around the water more easily became the fixation focus along with the water space, such as the wooden and stone pavements near the water that are more attractive than when they exist alone, which is consistent with the previous findings of Gao et al. This is consistent with the significant correlation between the structure, form, and color of natural-related features in the promotion of emotional attachment showed by the results of the scale. It reveals that the combination of different landscape elements may have the potential to be more attractive in the landscape and provides people with a more significant emotional experience.

Through virtual reality eye-tracking experiments supported by VR glasses, this study also found that both natural and artificial elements are more likely to attract people’s attention when they are at the vanishing point of space at the height of the human viewpoint, and their ability to attract attention even exceeds the characteristics of the elements themselves. This has not been shown in previous eye-tracking studies based on 2D images, further suggesting that the spatial organization of landscape elements may have a greater impact on people than the elements themselves. This further shows that perceiving and appreciating landscape space has a temporal and spatial rhythm, and this rhythm will greatly affect people’s perception and emotional experience. Here, the importance of the spatial sequence of landscape elements for emotional attachment is emphasized.

(2)

The interaction between students’ visual behavior and emotional attachment to campus landscape space.

Previous studies have found that people’s emotional experience of space through visual interaction always begins with unconscious perception or curiosity without a clear purpose. This is because human visual observation is a process in which human cognitive and emotional experiences are influenced by stimuli that change over time and space.

Combined with the VR experiment, it can be speculated that the formation of different degrees and dimension of emotional attachment of participants in this process may be the result of people’s observation of the dynamic accumulation of landscape space. As the old saying goes, the eyes are the windows of the soul, and people’s emotions are also the psychological reactions after the eye’s reception and processing of different information in the landscape space. Therefore, the spatiotemporal nature of vision leads to the fact that the outcome of each eye movement has an impact on the next eye movement, as well as the formation of emotional attachment. Based on this, it can be inferred that landscape elements, especially their spatial composition, have a direct impact on this process, especially when people experience the real three-dimensional built environment.

But in the process of establishing and changing emotional attachment, it still remains unknown to what extent each eye movement affects it, and vice versa. In the future, it will be possible to conduct more in-depth research and analysis on these topics by combining multimodal emotion research techniques such as physiological measurements, EEG measurements, etc. This may be helpful to understand the process of people’s cognitive and emotional experience of landscape space, so as to help designers make better use of landscape elements and design spaces that are more conducive to people’s emotional attachment.

4.4. Limitations and Future Research

Until now, the study has made great effort to explore the effects of campus landscape elements on students’ visual behavior and emotional attachment and their connection. Though the results have provided some interesting perspectives, several limitations should also be considered.

First, there are limitations in the respondents. The data of this study were mainly collected from students of various majors in Beijing Forestry University under 24 years old. Though previous studies have shown that students’ data can be representative for landscape visual behavior research, neither gender nor nationality will influence people’s eye movement [78], and perception of as well as attachment to specific landscape elements is more likely to be driven by biological mechanisms, but students’ educational experience and background will affect their preferences for different landscape elements [79]. For example, most students in the experiment major in landscape architecture and urban planning, and they are familiar with the designer of the Heart of Forest, which may influence their emotional attachment. Meanwhile, students with a professional background in landscape and botany design may pay more attention to different natural plants, which will affect their visual behavior. In addition, all the participants in the study are young people, whereas the users of the campus landscape space also include teachers, various types of staff and their families. In terms of demographic characteristics, it is worthwhile to further study the cognitive, behavioral, and emotional attachment characteristics of landscape space for people of different ages and professional backgrounds.

Second, there are limitations in the experimental materials. The study was conducted based on the panoramic image of the Heart of Forest. With the support of HTC VR glasses, although this study was carried out using virtual reality immersive VR eye-movement experiments to explore students’ visual behavior, compared with previous studies using two-dimensional pictures as experimental materials, it is closer to the human experience in the real landscape space, but it still cannot cover the physical factors such as sound, temperature, light changes, and smell in the real environment. Therefore, the characteristics of people’s visual behaviors on site and their differences to that of photo-based experiments are worthy of further analysis and discussion. In addition, this study selected one of the representative landscape spaces in Beijing Forestry University as an experimental stimulus sample to collect data and explore the related issues. But the deviation in landscape space such as its background, seasonal factors, location, and even its designer may also affect people’s visual behavior and emotional attachment. Therefore, it is necessary to select more campus landscapes of different scales, different locations, and different seasons for more in-depth discussion in future research.

The visual quality and emotional healing effect of landscape are of great value in the design and renovation of campus landscape space, and are an important part of creating a good environment for students’ daily learning and life. A large number of empirical studies in the field of landscape architecture have proven that the application of eye-tracking experiments can reveal the visual perception process of the public and identify key landscape elements. Overall, this study analyzed the characteristics and differences of students’ visual behavior and emotional attachment when viewing different artificial and natural campus landscape characteristics through VR eye-tracking experiments and emotional attachment evaluation, and to explore the factors affecting students’ emotional experience when observing landscape spaces. This study not only proves the value of previous eye-tracking research results, but also improves the authenticity of the environmental visual experience in classical eye-tracking research with the help of new virtual reality technology. The coupling study combined with the emotional attachment scale further reveals the emotional appeal to people of landscape space from a multimodal perspective. In the future, combined with different physiological measurement technologies and machine learning methods, it will be possible to explore the landscape space elements and their spatial structure that are conducive to promoting people’s positive emotional experience in a more complete and in-depth way, and establish the corresponding evaluation system and model. This could also inform further development of emotion-oriented spatial design theories and methods.

5. Conclusions

Taking the campus landscape of the Heart of Forest in Beijing Forestry University as an example, this study explored the differences and relationships between visual behavior and emotional attachment to different natural and artificial landscape elements with the help of VR eye-tracking technology. The results of the two parts of the data not only reveal people’s subconscious and conscious preferences for landscape space elements and characteristics from different levels, but also corroborate and supplement each other. The main findings can be summarized as follows:

(1)

The results of eye-tracking indicators show that artificial elements are more likely to quickly attract people’s visual attention and continuously enhance their interest in the landscape. Combined with the results of the scale, it further shows that artificial elements with regionality, uniqueness and diversity are conducive to significantly promoting people’s emotional attachment, and whether playability and changeability can promote positive emotional experience are closely related to the maintenance of artificial facilities after completion.

(2)

For natural elements, a waterscape space composed of water and its surrounding elements is more likely to attract people’s visual attention, while the attractiveness of arbors and shrubs is related to their color and spatial location. Plants with yellow and red color changes in autumn and in the spatial position of the vanishing point of human eyes are more likely to be noticed. Furthermore, the characteristics related to nature are generally conducive to the establishment of students’ emotional attachment, which is not only limited to the natural elements in the landscape, but also includes artificial materials and structures that can reflect the natural texture, time traces and structural logic, such as pavilions that can extract the texture and mechanical characteristics of wood, and rusty steel plate landscape structures that can reflect the texture of materials.

(3)

Another important conclusion of this study is that the three-dimensional spatial structure and spatial sequence design of landscape elements will significantly affect people’s visual focus and emotional experience. This emphasizes the difference between experiencing real landscape space and two-dimensional landscape pictures and illustrates the importance of considering the spatial hierarchy of landscape elements and the spatial-temporal behavior of people experiencing the landscape in landscape space design.

(4)

The research provides the following possible references for the design theory and management of public landscape space in colleges in the future: for natural landscape elements, the design methods such as configuring plants with unique seasonal characteristics, creating more waterfront rest spaces, and arranging the elements in the space and time sequence of human perception may exert a significant impact on students’ emotional experience. For artificial landscape elements, the design and construction of structures using natural materials and structures, and the setting of regional, cultural or commemorative installations, can effectively promote the construction of students’ emotional attachment. For the interactive landscape elements designed in combination with new technologies, it is necessary to rely on regular maintenance and management after completion to ensure their normal use, otherwise it is easy to trigger negative emotions in students.

In addition, a corresponding relationship has been revealed between artificial and natural landscape elements in eye-tracking experiments and landscape characteristics explored in an emotional attachment scale. That is, people’s visual behavior and emotional attachment are closely related, and will change and interact with their experience of time and space in a certain landscape. The fixation behavior affects the emotional experience it produces, and the change in emotional experience will then affect the focus of vision.

The results show that when designing a campus landscape space to provide students with a good emotional experience, it is not only necessary to consider the selection and design of each natural and artificial landscape element, but designers also need to consider the spatial–temporal sequence of people’ viewing and experiencing landscape spaces and make a reasonable spatial combination of these elements. At the same time, it is also important to strengthen the maintenance of the landscape after its construction to promote its positive emotional experience.