1. Introduction
As urbanization continues at an alarming rate, it not only threatens biodiversity but also increasingly separates people from the natural environment [
1], which is directly linked to undesirable social-related health diseases such as obesity, diabetes, depression, and mental fatigue [
2,
3]. During the last decade, many studies have demonstrated a positive association between nature and individuals’ health [
4,
5,
6]. Urban green spaces (UGS), which are mainly composed of natural environments, are an important barrier between healthy and unhealthy lifestyles [
7]. Hence, the notion of creating UGS that optimize their restorative effects has become a key focus for UGS planning in densely populated areas [
8,
9].
To date, several studies have shown that UGS have restorative effects for inhabitants, such as attention recovery [
10,
11], stress, and mental fatigue restoration, mood promotion, and the prevention of depression [
10,
12,
13,
14]. Two broadly accepted theoretical frameworks that explain how green environments affect human health and well-being are the attention restoration theory [
11] and stress reduction theory [
15]. Despite the wide applications of these two theories in UGS studies, effective guidelines for designing a restorative environment are lacking [
2]. This is because the dominant characteristics of restorative environments are unclear, especially for UGS [
16]. Therefore, it is necessary to identify which structural components of the UGS are the most important driving factors for human health [
16,
17]. To address this, the present study examined the restorative effects of UGS differing in characteristics, based on vegetation structures [
18,
19,
20].
It is worth noting that whilst it is generally realized that UGS have restorative effects, there is evidence that preferences have played an important role in attracting people to restorative environments and in keeping them in such environments for a prolonged period of time; that is, individual preference may be a keystone for the restorative action of environments. Several studies have identified a positive relationship between aesthetic preference and mental restoration [
9,
21,
22,
23,
24]. However, not all specific types of UGS facilitate individual preference [
25]. Hoyle et al. and Han demonstrated that an environment with high restoration is not always a high aesthetic quality environment that garners high preference [
16,
26]. Given these inconsistent results, the relationships between recreational preferences and restoration in the different environments require further examination.
Psychophysiological responses can often be elicited by the visual perception of environmental stimuli [
27]. Most empirical studies have employed subjective rating scales to examine the environmental effects on human health, while objective measurements have rarely been used [
28,
29]. The electroencephalogram (EEG) can be used to detect an individual’s electrical activities in the brain, and is considered an objective indicator of the physiological stress response in humans [
30,
31]. The spontaneous bioelectrical signal of the brain can be divided into alpha waves (8–13 Hz), beta waves (>13 Hz), delta waves (equal to 4 Hz), theta waves (4–8 Hz), and other types according to the different frequencies of the EEG [
30]. Hankins et al. found that when stress or fatigue increased, the EEG alpha waves vanished and changed into beta waves and theta waves, which indicated that the EEG alpha waves were prominent in a state of relaxed wakefulness [
32,
33,
34]. Ulrich also showed that alpha waves resulting from the exposure to natural photos were significantly higher than those of exposure to townscape photos [
35]. Chang et al. found that the EEG alpha waves of a simulated landscape with a restorative function were increased compared to other simulated landscapes [
36]. The previous studies showed that the increase of these EEG alpha waves’ value can reflect the physiological relaxations experienced when one is exposed to natural features. Therefore, this study used the EEG alpha waves as a primary indicator of physical relaxation in the experiments.
In addition, compared with the traditional on-site survey that is often influenced by uncontrollable factors such as traffic noise, graffiti, litter, etc. [
37], virtual reality technology (VR) is a new way to maximize the perception of an environment by isolating the visual and auditory sense of the outside world, using a head-mounted display device [
38]. This guides the user in creating a sense of immersion in a three-dimensional environment, and gives them a better sense of the environment [
39]. Accordingly, this study used VR technology as visual stimulation, integrating it with ecological and social methods to explore the effects of the different types of environment on human psychological and physiological health. The specific objectives were to investigate:
What is the difference in the restorative state before and after visual stimulation by using VR devices?
What are the effects of the different types of environments on people’s physiological and psychological restoration?
Which types of environments do people prefer? How do people’s preferences relate to restorative effects of environments?
2. Materials and Methods
2.1. Stimulus of Experimental Images
The visual stimulus materials were VR panoramic photographs, which were shot by a panoramic camera (Insta360 Pro-I) in different urban environments of China. The resolution of the panoramic photographs is 7680 × 3840 (8K) pixels. Visual stimuli were displayed by VR glasses (Pico Goblin VR all-in-one) with a resolution of 2560 × 1440 pixels, and a screen refresh rate of 70 Hz (<20 ms). As shown in
Table 1, the selected visual stimuli were classified into the three types of environment according to land use, including grey space, blue space, and green space. UGS were further divided into four types based on the horizontal vegetation structures according to their vegetation patterns and plant configurations, which were open green space, partly open green space, partly closed green space, and closed green space (
Table 1). A sorting task procedure was adopted to ensure the selected experimental stimuli were matched according to a specific type of environment. Ten landscape architecture experts were invited to evaluate and sort the 240 panoramic photographs (40 photographs for each type) into six types of environment according to the classification system, and then to rank each photograph within a type based on the size, shape, location, and composition of vegetation structure (
Table 1). The evaluation resulted in the selection of the top five panoramic photographs with high scores for each type of environment, amounting to 30 photos (
Figure 1).
2.2. Participants
In this study, 120 Chinese college students were voluntarily recruited through on-site invitation and online invitation by WeChat (mean age = 20.7, SD = 2.13, 58 males and 62 females). They were healthy with myopia degrees of less than 800 degrees, which was the maximum required degree of the VR glasses. Participants were assumed to have accumulated fatigue through two hours of classroom learning immediately prior to participation. Groups of 20 participants were randomly assigned to one of the specific types of environment for visual evaluation.
2.3. Measurements
2.3.1. Physiological Stress
An EEG can be used to measure the brain’s response to external visual stimuli. This study used the NeuroSky portable brainwave device with a NeuroSky TGAM brain wave chip inside to measure human physiological responses through electrodes attached to the scalp; to obtain the EEG data of the participants, the brain wave data were sent to a computer in real time. The EEG alpha waves (8–13 Hz range) obtained can be used as a primary indicator for physiological stress [
40], as higher values of the EEG alpha waves indicate better restoration of physiological stress [
41].
2.3.2. Psychological Stress
The study measured participants’ stress through the 40-item Profile of Mood States (POMS-SF) scale, which has been most widely used in restorative research, and as a tool for evaluating the psychological stress of an individual [
42]. Participants were asked to respond on a 5-point Likert-type scale (1 = not at all; 5 = feel very strongly) of POMS-SF before and after visual stimulation. The scale reveals seven affective measures including tension, anger, fatigue, depression, vigor, esteem, and confusion, which can be classified into two broad emotional dimensions: “Negative mood” (calculated as weighted averages of anger, tension, fatigue, confusion, and depression subscales, and the minimum possible score is 6.08, the maximum possible score is 30.42) and “positive mood” (calculated as weighted average of vigor and esteem scales, and the minimum possible score is 5, the maximum possible score is 25). A higher score for each dimension corresponds to a higher emotional level.
2.3.3. Attention
The Stroop color task was used to assess the restoration of participants’ attention before and after visual stimulation. The Stroop color task is an effective and reliable selective attention measure [
43], which is often used for restorative research experiments. In this experiment, the Stroop color task was used for measuring the attention capacity of participants within 45 s, in accordance with the result of the preliminary experiment. Twenty participants were recruited to conduct the Stroop color task first, in order to ascertain an appropriate length of time for the new participants to fulfill the Stroop task successfully in the formal experiment. They were presented with 70 color words printed in incongruent colors, and asked to name the ink color and to complete the task within a specified time [
43]. For example, when the word “yellow” is presented in red ink, the participant must say “red”. Only the responses that were correctly completed in 45 s were recorded. The attentional capacity of the participants was calculated by a percentage based on the correct number of responses out of 70 items. The higher the percentage received, the better the attention capacity of the participant.
2.3.4. Preference
The participants’ preferences for each panoramic photo of the environment and their reasonings were collected through a questionnaire survey. For each panoramic photo of one specific type of the environment, the respondents were asked to indicate whether they preferred the environment while wearing VR glasses, using a “Yes/No/Hard to say” response. The latter response was added to provide some indication of how well the environment was perceived by the respondents through VR glasses, as a high percentage of “Hard to say” answers was likely to indicate that the environment demonstrated by VR glasses was less well accepted. In the subsequent statistical analysis, the “Hard to say” answers were treated as “No” in accordance with Wang, and the score assignment was thus “No/Hard to say” = 1 and “Yes” = 2, correspondingly [
44].
2.4. Design and Procedure
In order to eliminate the nervousness of the participants and to better allow them to understand the experimental process, the procedure, and the relevant equipment’s function within the experiment, were introduced for participants before the experiment began. Each set of experiments in this study was conducted in the afternoon or evening after the participants endured at least 2 h of class time, which induced an accumulation of fatigue in the participants. Each group of participants was asked to see only one type of the panoramic photos at a time.
First, participants were asked to complete the tests for baseline measures of attention and psychological stress, by using the Stroop color task within 45 s, and then completing a POMS-SF questionnaire. The portable EEG electrode was placed onto participants’ foreheads, and they were asked to sit facing the wall to temporarily exclude external visual stimuli. Then, they were asked to open and close their eyes alternately in two one-minute cycles to determine their brainwave baseline, in order to identify the baseline of psychological stress (
Figure 2a,b).
Next, the VR glasses were used to play panoramic photo slides for each participant, corresponding to his/her group assigned. To ensure that the testing time was not too long, the participants could fully experience the scene in the VR glasses. Each slide show contained 5 photos and each photo was displayed for 1 min, with the entire watching time lasting 5 min. All participants’ EEG data were collected during the watching time. Then, the VR glasses and portable EEG electrode were removed from participants. The preference questionnaire for each panoramic photo was then filled out.
Finally, a second round of measuring of attention and stress was completed, consisting of the Stroop color task and the POMS-SF questionnaire. It required approximately 35 min for the whole experimental procedure (
Figure 2c).
2.5. Data Analysis
The final number of participants was reduced to 116 due to missing brainwave data. SPSS 25.0 software (IBM, Armonk, NY, USA) was used in all statistical analyses. In order to detect differences in the restoration of people’s attentional fatigue and psychophysiological stress before and after viewing the natural environment through VR devices, one-way analysis of variance (ANOVA) was used. To determine how different types of environment affect people’s attention and mood, an analysis of covariance was used, in which pre-tests were regarded as covariates to compare with post-tests of attention and mood in groups exposed to different types of environment. ANOVA was also used to compare the effects of the different types of environment on participants’ physiological stress and recreational preferences. Correlation analysis was employed to explore the relationship between the participants’ recreational preferences and the restoration factor of the environments. The preference was determined by the average score of 116 respondents for the environment they perceived. The restoration of the environment was divided into three variables: ‘attentional fatigue’, ‘physiological stress’, and ‘psychological stress’. Psychological stress was subdivided into negative and positive mood. Each variable was judged by the average score of relevant tests of the participants.