Exploring the Impact of Daytime and Nighttime Campus Lighting on Emotional Responses and Perceived Restorativeness

Abstract

The quality of campus environments plays an important role in the mental health of college students. However, the impact of nighttime lighting in campus settings has received limited attention. This study examines how different landscape lighting conditions affect emotions and the perceived restorative potential, providing a mixed-method research framework to assess nighttime landscapes. The study was conducted on a section of campus roadway under three scenarios: daytime (cloudy conditions) and two nighttime settings (landscape lights and streetlights, and streetlights only). We employed wearable biosensors, visitor-employed photography tasks, affective mapping, interviews, and self-reports to comprehensively assess the participants’ emotional responses and perceptions. Statistical analyses, including the Friedman test, Wilcoxon signed-rank test, one-way ANOVA, Getis–Ord Gi* statistic and kernel density analysis, were used to evaluate differences in emotional and restorative perceptions across lighting scenarios. The results showed that nighttime environments with well-designed landscape lighting enhance the restorative potential more compared to street lighting alone and, in some cases, even surpass daytime settings. Skin conductance data, integrated with spatial–temporal trajectories and affective mapping, revealed clear patterns of emotional responses, emphasizing the role of lighting in shaping environmental quality. These findings provide actionable insights for architects and lighting designers to create nighttime landscapes that promote emotional well-being and restoration.

1. Introduction

Many college students face serious physical and mental health challenges due to factors such as heavy academic workloads, employment competition, and interpersonal pressures [1,2]. Campus outdoor spaces not only serve aesthetic and activity-related functions but also play a role in psychological recovery and stress reduction [3,4]. Lighting, both natural and artificial, significantly shapes the quality of these spaces and influences emotional and restorative perceptions. However, research on the psychological effects of campus nightscapes remains limited.

While much research focuses on daytime environments, the nighttime—an equally important period for student activities—deserves attention. A well-designed campus nightscape can offer a safe, restorative environment, alleviating stress and supporting mental well-being. Investigating its psychological impacts under various lighting conditions is therefore essential.

Traditional methods for assessing landscape preferences, such as phenomenological interviews and self-report scales [5,6], are resource-intensive and limited by subjective biases and questionnaire quality. Advances in neuroscience and physiological sensors have introduced objective measures to complement self-reports [5,7,8,9]. However, studies examining the effects of daytime and nighttime lighting conditions on landscape perception with these techniques remain limited.

To contextualize this study, the literature review is organized around three main themes: (1) the mental health challenges faced by college students, particularly in China; (2) the influence of campus landscapes on students’ mental health, with an emphasis on the restorative effects of both daytime and nighttime environments; and (3) the methodological advancements in studying walking experiences and environmental perceptions. By synthesizing these themes, the review highlights the gaps in existing research and establishes the foundation for this study’s contributions.

1.1. College Students’ Mental Health in China

Mental health among college students is a growing global concern, particularly following the COVID-19 pandemic, as one’s college years are inherently stressful [10,11]. A 2020 meta-analysis found the prevalence of depression among Chinese university students to be 28.4% [12]. The COVID-19 pandemic worsened the situation, with higher rates of depression, anxiety, and insomnia reported during lockdowns [2]. Research shows that coronavirus-related anxiety negatively impacted students’ academic engagement and resilience [13]. While the severity of mental health challenges is widely recognized, effective intervention and prevention strategies remain a key focus of research.

1.2. Campus Scenery and Mental Health

Campus scenery significantly affects students’ mental health by providing a restorative living and learning environment. This section examines the theoretical links between built environments and mental health, campus landscapes’ impacts, and daytime–nighttime scenery differences.

1.2.1. Built Environments and Mental Health

The connection between built environments and mental health has been widely studied [14]. Attention Restoration Theory (ART) argues that exposure to nature leads to the restoration or recovery of directed attention system, hence improving the cognitive capacity of the viewer [15]. Stress Reduction Theory (SRT) underscores nature’s ability to reduce stress more effectively than urban settings [16]. Emotional models, such as Russell’s two-dimensional framework [17] and Arousal Theory [18], explore how environmental stimuli influence emotions and behavior. Evidence shows that high-quality landscapes positively impact mental health by improving attention and aiding stress recovery [19,20].

1.2.2. Campus Landscape and Students’ Mental Health

The college campus, where students live, work, and study, significantly influences their mental health. High-quality landscapes on campus can reduce mental fatigue and improve focus, academic performance [21,22,23], social connections [24,25], and psychological well-being [26,27,28] and increase engagement in physical activities [29]. During the COVID-19 pandemic, the campus environment became even more critical, as studies showed that high-quality living environments reduced students’ depression and anxiety risks [30,31].

Most studies on campus landscapes focus on daytime settings, emphasizing green spaces and water features’ effects on attention and emotions [32,33,34,35,36,37]. However, nighttime environments are less studied and primarily address safety concerns [38,39,40] rather than emotional and restorative needs [41]. Given the extended class hours at Chinese universities, addressing nighttime campus settings is crucial for students’ mental health under pressures.

1.2.3. Daytime and Nighttime Scenery

According to the temporal dimension, landscapes can be categorized into daytime and nighttime scenarios. Under different light sources, such as sunlight or artificial light, the same scene will appear completely different, triggering different perceptions [42]. Daylight enhances environmental clarity and positive emotions [43], with studies linking sunlight exposure to reduced negative emotions and fatigue [44]. At night, artificial lighting is strategically designed to highlight specific elements, creating unique nighttime landscapes. Studies have shown that bright lighting can enhance the sense of safety, excitement, and preference in urban spaces [45,46,47]. However, the sense of safety at night is not solely determined by brightness; it is also influenced by luminaire parameters such as color temperature [48,49], uniformity [50], lighting patterns, and placement [51,52].

Comparative studies of specific locations under daytime and nighttime lighting reveal how illumination affects perception [53]. Cheon found that daytime and nighttime campus views affected both verbal and nonverbal restorative perceptions and brain responses differently [54]. Similarly, Zhao showed that while daytime landscapes generally offered a higher restorative quality, the nighttime restorative potential depended heavily on the quality of lighting design [55]. Li et al. employed a light-walking approach to investigate the temporal and spatial characteristics of nightscapes in university campus outdoor spaces and the impact of temporal and spatial characteristics on lightscape perception, thereby providing optimization strategies for lighting design [56]. In another study focusing on daytime and nighttime landscapes in urban green spaces, it was proposed that lighting design could enhance the quality of nighttime environments by illuminating positive elements, concealing negative ones, and integrating landscape features with lighting characteristics [57]. Comparing daytime and nighttime scenarios is essential for understanding how lighting changes influence perception and restoration, guiding nighttime lighting design to improve environments and support psychological well-being [58].

1.3. Integrating Multiple Methodologies for the Study of Walking Experiences

Walking is a fundamental activity for experiencing urban environments and serves as an effective approach to studying dynamic perceptions of urban spaces. It has been widely utilized in urban quality research and nighttime lighting design [56,59]. Moreover, high-quality outdoor campus built environments have been shown to positively correlate with increased walking activities [60,61,62].

Researching walking experiences requires combining subjective and objective methods to capture physiological and psychological responses. Traditional approaches, such as interviews and subjective scales, are now enhanced by neuroscience and wearable biosensors, enabling real-time and location-specific data collection. These methods fall into three categories: subjective assessments, physiological measures, and behavioral observations.

Wearable biosensors offer key advantages: (i) they provide real-time and objective physiological data, reducing the reliance on potentially biased self-reports; (ii) they enable continuous and high-precision measurements over time [63,64]; and (iii) they have the ability to capture dynamic physiological responses in real-world settings, critical for ecological studies. Physiological measures often outperform subjective ones [5,8], enriching research methods. Integrating neuroscience with traditional methods enables a comprehensive understanding of pedestrian experiences, revealing how individuals perceive and respond to their environments [65].

Mixed-method studies combine biosensors with traditional methods to explore pedestrian experiences. These small-sample studies emphasize subjective perceptions over statistical generalizability, providing detailed insights into individual physiological and psychological responses (

Table 1. Summary of previous studies.

Source	Sample Size	Measure Times (per Participant)	Study Area	Participant Group	Measurement Methods	Subject
Study 1 [5]:	4	3 times, each lasting for 15 min	Campus road	Students	Phenomenological interview Psychological scales Biosensing measures (ECG, EEG, EMG, SC, skin temperature, respiration)	Affective experience during walking
Study 2 [66]:	10	15 min	Urban streets (totaling 570 m)	Older adults	Biosensing measures (HRV) Isovist analysis Subjective questionnaire	Stress during walking
Study 3 [67]:	7	40 min	Commercial streets (totaling 400 m)	Students and employees	Eye-tracking data Visitor-employed photography In-depth interview	Attention and landscape experience
Study 4 [68]:	10	2 rounds, each lasting for 10 min	A road near a residential area	Female students	Skin conductance Skin tempreture Subjective questionnaires	Body responses
Study 5 [69]:	4	2 rounds	Road in VR (totaling 140 m)	Volunteer colleagues	Skin conductance Heart rate variability Gait sensor	Walkability and well-being

). While neuroscience-based methods have been applied to campus environments, most focus on daytime conditions, with limited attention given to nighttime variations. This study evaluates these methods across daytime and nighttime scenarios to enhance their relevance and applicability.

1.4. Research Contribution

In our study, a representative road on Tongji University’s Siping Campus, recently renovated in 2021 with enhanced nightscape lighting, was selected as the research site. As the primary design team, we integrated landscape lighting with existing functional lighting to improve nighttime safety and promote emotional recovery for students. In addition to a natural daylight scenario (under overcast conditions), two nighttime artificial lighting scenarios were established: pre-renovation (street lighting only) and post-renovation (street and landscape lighting), providing an ideal experimental context.

The key contributions of this research are as follows:

(1)

An examination of the impact of different lighting scenarios on college students’ emotional and restorative perceptions.

While prior research has primarily focused on daytime campus environments, this study provides empirical evidence that well-designed nighttime lighting can significantly enhance perceived restorativeness and emotional well-being. By comparing different lighting conditions (overcast natural daylight, nighttime street lighting, and the combined street and landscape lighting), our findings emphasize the importance of landscape lighting in improving students’ nighttime experience, sometimes even surpassing daytime settings.

(2)

The development and validation of a mixed-method framework for nighttime landscape evaluation.

This study integrates biosensors with traditional tools (visitor-employed photography, phenomenological interviews, Likert scale ratings, and cognitive mapping) to create a comprehensive framework for assessing students’ emotional responses to lighting environments during daytime and nighttime walks. This framework offers a novel and practical tool for evaluating nighttime lighting design, providing actionable insights for architects and urban planners to create emotionally supportive and restorative campus environments.

2. Material and Methods

2.1. Study Site

This research was conducted on the Siping Campus of Tongji University, a representative urban campus in Shanghai that underwent renovations in 2021. The site was selected due to its diverse lighting configurations, controlled experimental environment, and high level of participant familiarity, ensuring an in-depth investigation of psychological impacts under different lighting scenarios. Additionally, the study was conducted during the COVID-19 lockdown, when the campus served as the primary connection between students and natural stimuli, minimizing external environmental interference due to restricted movement. Subsequent policy changes lifted these restrictions, creating a field study under unique circumstances.

A 700 m round-trip route was selected for its diverse spatial characteristics (see Figure 1). Based on the varying landscape types, the route was segmented into three areas (Area1, Area2, and Area3). Three fixed stopping points (Stop 1, Stop 2, and Stop 3) were established to capture participants’ static environmental perceptions. Three experimental lighting scenarios were defined: (i) daytime under overcast conditions to reduce interference from intense sunlight; (ii) nighttime scene A, with both streetlights and landscape lights (post-renovation), (iii) nighttime environment B, with only streetlights (pre-renovation). This setup addresses the gap in the literature on landscape restorativeness, which primarily focuses on daytime settings with elements like greenery, water, and forests. At night, streetlights fail to fully illuminate these features, making landscape lighting essential for nighttime restorativeness. Thus, the three scenarios were designed to investigate the specific role of landscape lighting in enhancing restorativeness.

This study did not control specific lighting parameters as variables; instead, different lighting scenarios along the same roadway section were treated as independent variables. In the nighttime conditions, streetlights refer to fixtures installed along the primary traffic routes, designed to illuminate the roadway surface for visibility and safety. In contrast, landscape lighting refers to luminaires that illuminate environmental elements other than the roadway, such as architectural structures, water features, and vegetation (see Table S2 for detailed lighting configurations and source parameters). Lighting measurements were taken at 50 m intervals along the walking route, with readings in four directions (two along the road and two perpendicular) at eye level. A calibrated Everfine Spic-200 Spectroradiometer, placed vertically at 1.6 m above the ground, was used to measure corneal illuminance. Measurements were conducted during non-rainy periods to avoid additional illumination from reflections from wet surfaces. The average eye illuminance values for daytime, nighttime Scene A, and nighttime Scene B were 4408.96 lx ± 1990.73, 1.38 lx ± 0.61, and 0.98 lx ± 0.25, respectively. The relatively large standard deviations in eye-level illuminance are primarily due to shading from trees and variations in the direction and placement of luminaires. All nighttime light sources had a color temperature ranging from 3000 K to 4200 K and a color rendering index (CRI) above 60, meeting the requirements for Class IV roads, as specified in the Chinese Standard for Lighting Design of Urban Roads [70]. The color temperature and color rendering index (CRI) of light sources under different scenarios are provided in the Supplementary Materials Figure S1, Tables S1 and S2.

2.2. Participants

This study was conducted during the COVID-19 pandemic lockdown period, during which participant recruitment was significantly constrained by evolving public health policies and restrictions. Under these limitations, we adopted a qualitative research approach with a small sample size. By systematically collecting multiple types of physiological data from participants, we were able to conduct an in-depth analysis of complex data characteristics, thereby reducing the reliance on sample sizes [71]. As discussed in Section 1.2.3, this method has been widely adopted in similar studies due to the complexity of the data and measurement indicators, which made large-sample studies impractical under the current conditions. Consequently, the small-sample approach is particularly suitable for preliminary exploratory research, providing essential support for subsequent investigations (see

Table 2. Description of participants.

Participant	Gender	Age (Years)	Body Mass Index (BMI)	Self-Rating Anxiety Scale (SAS)	Self-Rating Depression Scale (SDS)	Morningness-Eveningness Questionnaire (MEQ)	Major
1	Male	18	18.94	31	35	44	Computer Science
2	Male	18	20.05	33	40	55	Transportation
3	Male	21	22.09	31	30	51	Civil Engineering
4	Female	18	19.57	23	22	44	Public Administration
5	Female	18	18.22	30	37	41	Law
6	Female	18	23.72	41	50	49	Architecture
7	Female	18	18.37	42	42	33	Environmental Science
8	Male	20	24.30	39	46	44	Physics

Notes for the criterions: ① BMI: 18.5–24.9 kg/m² is normal, 24–27.9 kg/m² is overweight, BMI ≥ 28 kg/m² is obesity, BMI < 18.5 kg/m² is underweight; ② Self-Rating Anxiety Scale (SAS): anxiety scale scoring rules are 50 points or less for normal, 50–59 points for mild anxiety, 60–69 for moderate anxiety, and 69 points or more for severe anxiety; ③ Self-Rating Depression Scale (SDS): depression scale scoring rules are 53 points or less for normal, 53–62 points for mild depression, 63–72 points or more for moderate depression, and 73 points or more for severe depression. ④ Morningness-Eveningness Questionnaire (MEQ): Morningness-Eveningness Questionnaire scores range from 16 to 30 for absolute eveningness, 31 to 41 for moderate eveningness, 42 to 58 for intermediate eveningness, 59 to 69 for moderate early morningness, and 70 to 86 for absolute early morningness.

).

Participants were randomly recruited across the university through public announcements, with eight undergraduate students selected. Unlike graduate students, undergraduates typically live on campus and attend classes near the experimental route, making them familiar with the research area. Eligibility criteria included good health, a regular living schedule, normal vision (≥0.8), no color blindness or related eye conditions, and no alcohol abuse or smoking habits.

We collected participants’ Body Mass Index (BMI), Self-Rating Anxiety Scale (SAS), Self-Rating Depression Scale (SDS), and Morningness-Eveningness Questionnaire (MEQ) data through self-reports to ensure that both their physiological and psychological conditions were within normal ranges. This standardization guarantees that our research results are broadly applicable and reinforced by the typical health status of the participants involved. All participants fell within normal ranges for anxiety and depression, and the MEQ results classified six participants as “intermediate” and two as “moderate evening”, with no extreme morningness or eveningness types, indicating similar sleep habits.

Participants signed a consent form after being provided with a written explanation of the experiment. Each participant was compensated CNY 240 for their participation. The study was approved by the Tongji University Ethics Committee of Medical and Life Sciences (No. tjdxsr019).

2.3. Study Process

The experiment was conducted from 12 to 26 November 2022, during mild and cloudy weather conditions, without rain. Daytime experiments (8:00 am–11:00 am) were conducted at an average temperature of (17.5 ± 2.6) °C, relative humidity of (68.1 ± 8.1)%, and noise level of (60.4 ± 3.8) dB. Nighttime experiments (7:00 pm–10:00 pm) were conducted at (17.5 ± 3.0) °C, relative humidity of (74.3 ± 6.8)%, and noise level of (58.0 ± 1.4) dB. For the outdoor walking section, class periods with lower pedestrian activity were selected to minimize traffic on the roadways and reduce potential disruptions to participants.

Using a repeated measures within-subjects design, 8 participants completed 3 randomized visits, generating 24 data sets (3 scenarios × 8 participants). Each experimental round included five steps: (1) preparation (T0), (2) baseline (T1), (3) stressor (T2), (4) walking (T3), and (5) post-walking (T4). All of these steps, except for T3, which was conducted outdoors, were carried out in the laboratory.

During the preparation stage, participants were provided with an overview of the experimental background, then signed a consent form and wore biometric sensors. Subsequently, they were given ample time to familiarize themselves with the experimental software on the iPad and completed questionnaires about their personal information. Baseline data were recorded during a 3 min resting state, followed by SAM scale completion. During the stressor phase, participants engaged in a 3 min mental arithmetic task, a proven method for eliciting stress responses [72]. Immediately afterwards, the participants were taken to an outdoor location for the walking test.

In the walking phase, participants were given a designated route and asked to imagine the following situation: “Imagine yourself taking a stroll around the campus after an intense study session. Along the designated route, except for three specified locations where you are required to stop, you may walk while using an iPad to capture photos of the campus to share with your friends or family”. To collect data for affective mapping, which links visual experience, physiological responses, and GPS location, each participant was equipped with wearable physiological sensors, a GPS device (Garmin eTrex 221X GPS), and a video recorder. Two researchers accompanied the participants to collect data and for safety purposes. The accompanied researchers remained half a stride behind the participants to allow the participants to determine the pace.

After walking, the participants returned to the experimental room for a three-minute break before completing a questionnaire, an in-depth interview, and a cognitive map. During the experiment, the participants were not permitted to speak, eat, or consume energy drinks (see Figure 2 and Figure 3).

2.4. Measures

2.4.1. Biosensing Measures

According to Russell’s two-dimensional theory of emotion, human emotions can be classified into valence and arousal dimensions. Empirical research shows that heart rate reflects valence [73], while skin conductance response and respiration volume indicate arousal [74,75]. This study used the Ergo LAB device to collect data from three channels:

• The pulse wave (PPG): monitored by a sensor located at earlobe with 2048 Hz sampling rate.
• Skin conductance (SC): measured by two sensors on fingers with 256 Hz sampling rate.
• Respiration (Resp): measured at abdomen with 2048 Hz sampling rate.

Preliminary data processing revealed that walking jitter significantly affected heart rate and respiration, leading to poor data quality. Thus, only skin conductance data were analyzed. Skin conductivity, regulated by the sympathetic nerves via sweat glands, increases during emotional stimulation and decreases in relaxed states. Higher levels of skin conductivity are associated with emotions such as stress, anger, anxiety, fear, and happiness [63]; lower levels of skin electricity are associated with a more relaxed state, sadness, or relief [76].

Skin conductivity data are usually divided into two components: (i) skin conductance level (SCL), representing the tonic baseline level of skin conductivity; and (ii) skin conductance response (SCR), indicating a phasic increase in the amplitude of skin conductivity. In this ecological experiment, where stimuli were not precisely controlled, SCL was combined with GPS data to analyze participants’ real-time environmental sensations [77].

Raw signals were processed using high-pass and low-pass filters to remove noise, and tonic skin electric signals were extracted with ErgoLAB 3.17.7 To account for inter-individual variability, SCL data were standardized using min–max normalization.

$${\mathsf{\Phi }}_{\mathrm{S}\mathrm{C}\mathrm{L} \mathrm{i},\mathrm{j},\mathrm{k}}={\displaystyle \frac{{\mathrm{S}\mathrm{C}\mathrm{L}}_{\mathrm{i},\mathrm{j},\mathrm{k}}-{\mathrm{S}\mathrm{C}\mathrm{L}}_{\mathrm{i},\mathrm{k},\mathrm{m}\mathrm{i}\mathrm{n}}}{{\mathrm{S}\mathrm{C}\mathrm{L}}_{\mathrm{i},\mathrm{k},\mathrm{m}\mathrm{a}\mathrm{x}}-{\mathrm{S}\mathrm{C}\mathrm{L}}_{\mathrm{i},\mathrm{k},\mathrm{m}\mathrm{i}\mathrm{n}}}}$$

(1)

where i refers to the participants; j refers to the location points; k refers to the lighting scenarios; ${\mathsf{\Phi }}_{\mathrm{S}\mathrm{C}\mathrm{L}\mathrm{i},\mathrm{j},\mathrm{k}}$ is the standardized SCL value of participant i at location j in scenario k;${\mathrm{S}\mathrm{C}\mathrm{L}}_{\mathrm{i},\mathrm{k},\mathrm{m}\mathrm{i}\mathrm{n}}$ is the minimum SCL of participant i in the control stage (the period of T1 and T2) during scenario k;${\mathrm{S}\mathrm{C}\mathrm{L}}_{\mathrm{i},\mathrm{k},\mathrm{m}\mathrm{a}\mathrm{x}}$ represents the maximum SCL of participant i in the control stage (the period of T1 and T2) during scenario k.

Standardized SCL data were downsampled to 1 Hz to align with the GPS data. To analyze the spatial distribution of SCL values across different scenarios within the same environment, we employed a spatial clustering analysis using the Getis–Ord Gi* statistic in ArcGIS [66]. This method identifies statistically significant clusters of high values (hot spots) and low values (cold spots). A location is classified as a hot spot if the physiological response at that location is high and the physiological responses at the neighboring locations are also high, and vice versa for a cold spot. The Getis–Ord Gi*statistic is given as

$${\mathrm{G}}_{\mathrm{i}}^{\mathrm{*}}={\displaystyle \frac{\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i},\mathrm{j}}{\mathrm{x}}_{\mathrm{j}}-\overline{\mathrm{X}}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i},\mathrm{j}}}{\mathrm{S}\sqrt{\frac{[\mathrm{n}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i},\mathrm{j}}^2-{\left(\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{i},\mathrm{j}}\right)}^2]}{\mathrm{n}-1}}}}$$

(2)

where x_j is the attribute value for physiological response j, w_i,j is the spatial weight between physiological response i and j, n is equal to the total number of physiological responses, and

$$\overline{\mathrm{X}}={\displaystyle \frac{\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{j}}}{\mathrm{n}}}$$

(3)

$$\mathrm{S}=\sqrt{{\displaystyle \frac{\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{j}}^2}{\mathrm{n}}}}-{\left(\overline{\mathrm{X}}\right)}^2$$

(4)

The identified hot spots and cold spots across three lighting scenarios were compared to discern participants’ SCL in relation to location features.

2.4.2. Visitor-Employed Photography

Visitor-employed photography, which involves giving participants the power to freely take pictures of their favorite landscapes in a scene, thus facilitating the exploration of their true thoughts, is widely used in landscape, psychology, and other fields of research [78,79]. In this study, participants could stop, observe, or approach areas of interest during the walk and use an iPad to photograph objects or scenes they found appealing. After the walk, participants selected their three favorite photos and explained their choices in interviews.

The data analysis included two components:

(1)

Hot spot analysis (HS): Participants’ exploration time and number of photographs taken were statistically compared across the three scenarios. Kernel density analysis was then applied to identify Perceptually Exciting Nodes (PENs).

(2)

Content Analysis (CA): The photographs and corresponding perceptual descriptions from interviews were analyzed using text coding methods to classify the photos and associated perceptions.

2.4.3. Phenomenological Interview and Lusk’s Graphic Method

The phenomenological interview is a qualitative research method involving open-ended dialogues between researchers and participants to explore subjective experiences and gain insights into perceptual and cognitive processes [6,80]. Unlike conventional questionnaires, this method enables researchers to establish a deeper level of communication with respondents, facilitating a gradual exploration of research questions and avenues of enquiry throughout the interview process.

In this study, participants were asked to use six kinds of labels to mark the properties of different locations and their emotions, employing Lusk’s graphic communicative method [6]. Participants were then interviewed for more details (what happened, how they felt there, etc.) about those marked locations.

2.4.4. Self-Report

After each walk, participants completed a self-report, which included the Perceived Restorativeness Scale (PRS) and the Self-Assessment Manikin (SAM) scale. The PRS evaluates perceptions of restorativeness through 26 items across five subscales: Being-away, Fascination, Coherence, Compatibility, and Legibility [81]. Responses were recorded on a Likert 7-point response scale (coded 1 to 7). The SAM, a widely used tool for assessing emotions [82], measured participants’ subjective feelings before and after the walk. Responses were recorded on a Likert 9-point scale (1 to 9).

3. Results

3.1. Skin Conductance Measures

The Getis–Ord Gi* in ArcGIS was used to identify spatial clusters of high (hot spots) and low (cold spots) physiological responses triggered by environmental factors across the three lighting scenarios. The analysis results are presented in Figure 4. Overall, the spatial distribution of the clusters of hot and cold spots exhibited variations across the scenarios. In the daytime scene, there were 1110 hot spots and 917 cold spots. In nighttime scene A, 1419 hot spots and 1309 cold spots were identified. In nighttime scene B, 1025 hot spots and 985 cold spots were identified. The above results were significant at the 95% confidence level.

The hot spots showed similar spatial distributions across the scenarios, primarily located north of Areas 1 and 2. However, in nighttime scene A, Stop 2 also exhibited significantly higher hot spots. The cold spots varied more across the scenarios: (i) in the daytime scene, the cold spots were concentrated in Area 3 and at Stop 1; (ii) in nighttime scene A, the cold spots appeared scattered, primarily around Stop 3; (iii) in nighttime scene B, the cold spots were located on the south side of Area 1 and at Stop 2.

3.2. Phenomenological Interview and Lusk’s Graphic Method

Lusk’s graphic method and phenomenological interviews were used to capture participants’ perceptions across the three scenarios. Figure 5 shows the emotions of participants in different locations in the three scenarios.

In the daytime scene, participants reported negative emotions at Area 1, Stop 1, and Stop 3, mainly due to noise near the academic building and associated academic pressure. Positive emotions were linked to relaxing landscape elements like pavilions, bridges, and plants. In nighttime scene A, negative emotions were noted in Area 2, Area 3, and Stop 2, caused by pedestrian and vehicle activity as well as intense lighting. Positive feelings were attributed to landscape features such as pavilions, bridges, buildings, and seating areas. In nighttime scene B, negative emotions dominated due to the absence of landscape lighting, which made the environment dim, uninviting, and anxiety-inducing. The lack of illuminated features led to an unattractive nighttime landscape. Overall, the participants reported more positive than negative feelings in all three scenarios. The daytime scene and nighttime scene A evoked more positive responses compared to nighttime scene B.

3.3. Visitor-Employed Photography

3.3.1. Hot Spot Analysis

Figure 6 illustrates the participants’ exploration time and the spatial distribution of the photos captured across three scenarios. The Shapiro–Wilk normality test and homoscedasticity test confirmed the assumptions for the data. The one-way ANOVA revealed no significant difference in exploration time (p = 0.653 > 0.05) or the number of photos (p = 0.385 > 0.05) across three scenarios. The average exploration times were similar in the daytime scene (908.50 s) and nighttime scene A (902.50 s), with nighttime scene B being the shortest (866.00 s). The average number of photos per person decreased from the daytime scene (20.38) to nighttime scene A (18.50) and nighttime scene B (15.25).

The hot spot analysis (HS) was conducted using kernel density analysis in ArcGIS to identify the Perceptually Exciting Nodes (PENs). The spatial distribution of the photos reflected the participants’ interest levels in the environments. A high photo density was consistently observed at Stops 1, 2, and 3 across all of the scenarios, likely due to their designation as required stops. A key difference was observed in nighttime scene A, where the southeast corner of area 2 (near the academic building entrance) showed a high-density photo zone, absent in the daytime scene and nighttime scene B.

3.3.2. Content Analysis

Figure 7 depicts the photos taken across three scenarios revealing varying visual focus points. In the daytime scene, the participants captured diverse elements such as plants, pavilions, sky, roads, water features, and animals. In nighttime scene A, the participants focused on illuminated elements such as pavilions, plants, and buildings. In nighttime scene B, the participants’ photos were minimal, primarily capturing lighting fixtures and buildings.

The photo content and interviews revealed the participants’ perceptions of campus elements during day and night scenarios. The participants’ reasons for selecting their top three photos were categorized into four types using text coding. Tables S3–S5 in the Supplementary Materials provide additional information.

(1)

Comprehensive Visual Perception Photos reflect the participants’ perception of specific campus locations. The category is further divided into the following:

Color and Light Perception: The attractiveness of colors, brightness, dynamic effects, or related lighting factors.

“The yellow leaves scattered among the green ones look beautiful. The sky isn’t very blue, but I find the background color quite appealing”.

(Daytime Scene, Participant 7)

Total Environmental Perception: Preferences for the overall visual effect of the landscape.

“The lighting on the steps evokes a sense of tranquility and comfort, with the overall scene appearing bright and inviting”.

(Nighttime Scene A, Participant 1)

(2)

Emotional Response Photos capture the participants’ emotional associations with scenes that are not directly tied to the physical attributes of the scene but are analogously linked. For example, observing lush vegetation might evoke a sense of vitality and energy.

“This greenery is quite interesting; its round, plump shape is adorable, exuding vibrancy and an upward growth spirit”.

(Daytime Scene, Participant 3)

(3)

Unique Observational Photos capture distinctive elements noticed by the participants during their walks, differing significantly from their daily experiences, such as uniquely designed signs.

“This billboard is innovative, cute, and quite interesting”.

(Daytime Scene, Participant 5)

(4)

Imaginative and Reflective Photos refer to images that stimulate the participants’ imagination or evoke personal memories. For example, the warm, dim glow of light through a window at night might recall cozy family gatherings by the fireplace.

“This light makes me feel very cozy; it’s dim but not gloomy, just very serene. It reminds me of sitting by the fireplace with family gathered around”.

(Nighttime Scene B, Participant 2)

Table 3. The number of perceptual categories in each scene.

	Comprehensive Visual Perception Photos		Emotional Response Photos	Unique Observational Photos	Imaginative and Reflective Photos
	Color and Light Perception	Total Environmental Perception
Daytime	9	6	4	3	2
Nighttime scene A	13	4	1	3	3
Nighttime scene B	15	3	1	1	4

summarizes the distribution of the participants’ favorite photos across the scenarios. The Comprehensive Visual Perception Photos dominated across all of the scenarios. As the environment darkened (from the daytime scene to nighttime scene B), the number of Color and Light Perception photos increased, while the Total Environmental Perception photos decreased. Brighter scenarios (daytime and nighttime scene A) prompted more Unique Observational Photos, while darker scenarios (nighttime scene B) saw an increase in Imaginative and Reflective Photos. The day scenes had the most Emotional Response Photos.

3.4. Self-Report

3.4.1. PRS

The reliability analysis revealed that the Cronbach Alpha values for the Coherence and Legibility subscales were α = 0.621 and 0.662, indicating relatively low internal consistency. Further investigation identified that the items “It is a confusing place” (Coherence) and “It is easy to see how things are organized” (Legibility) did not align with the conceptual structure of the other items. This discrepancy was likely due to the participants’ familiarity with the test route and the clear navigation provided, which reduced their perception of confusion and organizational differences.

After removing these items, the Cronbach Alpha values for both subscales exceeded α = 0.70, indicating improved internal consistency. The overall reliability of the questionnaire was satisfactory (Cronbach’s Alpha: Total scale = 0.86, Being-away = 0.81, Fascination = 0.90, Coherence = 0.73, Compatibility = 0.87, Legibility = 0.77). The Friedman test revealed no significant differences in the participants’ perceived recovery across the three scenarios (see Supplementary Materials Table S6). However, the data trends indicated that the total perceived recovery ranked highest for nighttime scene A (4.59 ± 1.71), followed by the daytime scene (4.40 ± 1.76) and nighttime scene B (4.33 ± 1.77), suggesting that nighttime scene A had a greater recovery potential.

3.4.2. SAM Scale

The paired sample Wilcoxon signed-rank test was used to analyze the differences between T2 (pre-walking) and T4 (post-walking). In the valence dimension, both the daytime scene (p = 0.039 < 0.05) and nighttime scene B (p = 0.084 < 0.10) showed a significant increase after walking. In the arousal dimension, nighttime scene A demonstrated a significant decrease (p = 0.087 < 0.10). Furthermore, in the dominance dimension, the daytime scene exhibited a significant increase (p = 0.066 < 0.10). The results from the SAM scale indicate that all of the scenarios had positive impacts on valence (Figure 8).

The participants’ emotion changes in three dimensions between T2 and T4 were examined using the Friedman test to assess the significant differences between the three scenarios. Post hoc analyses were conducted using a paired sample Wilcoxon signed-rank test to elucidate specific differences. The results indicated significant differences in the arousal levels among the three groups (χ² (2) = 8.074, p = 0.018 < 0.05). Both nighttime scene A (p = 0.023 <.05, Cohen’s d = −1.33) and nighttime scene B (p = 0.018 < 0.05, Cohen’s d = −0.86) showed a significant decrease in arousal compared to that of the daytime scene, which exhibited a trend of increased arousal after walking. There were no significant differences observed in the valence or dominance across the three scenarios (Figure 9).

3.5. Integrated Analysis of SCL Measurement and Two Traditional Methods

According to the two-dimensional model of emotion, the spatial analysis results of the SCL reflect the emotional arousal, while Lusk’s graphic represents the emotional valence. The photo density distribution map illustrates the participants’ photo-taking behavior. By integrating spatial trajectories with multidimensional data, the participants’ overall environmental perception can be more comprehensive.

(1)

Emotional Patterns Along the Walking Path

As shown in Figure 10, the results of the SCL, Lusk’s graphic method, and photo density distribution are overlaid on the study site to provide a comprehensive analysis. Some places evoke emotions characterized by high pleasure and low arousal. For instance, in Area 3 of the daytime scene, Lusk’s graphic indicated positive emotional labels, the SCL results showed cold spots (lower stress levels), and the photo density map highlighted frequent photo-taking behavior. Combined with a visual analysis of the images, the primary stimuli contributing to these perceptions were identified as water features and trees. Conversely, some locations evoked low pleasure and high arousal. At the junction of Area 2 and Area 3 in nighttime scene B, Lusk’s graphic indicated negative emotional labels, the SCL results showed hot spots (higher stress levels), and the photo analysis revealed this area as a dimly lit crossroads.

(2)

Comparisons of Stopping Points

Based on Russell’s two-dimensional model of emotion, the emotional values at the three stopping points were compared across different lighting scenarios, as shown in Figure 11.

(1)

Emotional State Before and After Adding Landscape Lighting:

Comparing nighttime scene A and nighttime scene B showed that adding landscape lighting improved both the valence and arousal. At Stop 1, where only downlights were installed, the lighting change was minimal, and no significant differences were observed. At Stop 2 and 3, the landscape lighting increased arousal (from cold to hot spots), and improved valence (fewer negative and more positive labels).

(2)

Emotional State Between Daytime Scene and Nighttime Scene A:

At Stop 2, the landscape lighting significantly increased arousal compared to that in the daytime, while valence remained positive without significant changes. The results from in-depth interviews revealed that the emotional enhancement was attributed to the pavilion’s landscape lighting at Stop 2, which increased the scene’s attractiveness. At Stop 3, compared to the daytime, nighttime scene A showed fewer hot spots (indicating less stress) and more positive labels (reflecting improved valence).

Figure 11. Multi-source diagnostic of participants’ perceptions at three stops across different scenarios.

Figure 11. Multi-source diagnostic of participants’ perceptions at three stops across different scenarios.

4. Discussion

Most existing studies rely on questionnaires to explore participants’ perceptions of daytime and nighttime scenery. While biosensor technology has enabled scholars to investigate these perceptions in laboratories, stimuli such as images [54], videos, and VR technology often fail to replicate the visual effects of real lighting environments due to limitations like the dynamic range and color accuracy.

Our study addressed these limitations by conducting experiments in real-world settings with authentic lighting environments. One aim was to integrate traditional methods with biosensor techniques to evaluate landscape perception during both the daytime and nighttime. Another objective is to compare the effects of lighting scenarios on the participants’ emotions and recovery perceptions to quantify the improvements in emotional well-being resulting from the lighting renovation.

4.1. Relationship Between Outdoor Environmental Features and Emotional Response

Outdoor walking experiments have shown that environmental features influence psychological and physiological responses [9], as evidenced by the decreased SCL near resting spots or landscaped parks, and the increased SCL near safety risks or busy intersections [77,83]. However, these studies are limited to daytime settings. Our study extends these findings to nighttime conditions, observing similar phenomena across various lighting scenarios.

The SCL provides real-time insights into the relationship between environmental features and participants’ physiological stress responses across different lighting scenarios. In locations with visual attributes or aesthetic qualities—such as Stop 2 in the daytime scene and Stop 2 in nighttime scene B—the participants exhibited more cold spots (low-SCL clusters), indicating lower arousal and a more relaxed state. Conversely, in locations with potential safety risks—such as the crossroads in Area 2 of nighttime scene B—the participants showed more hot spots (high-SCL clusters), corresponding to higher arousal and stress. Notably, the addition of landscape lighting also evoked positive emotions with higher arousal. For instance, at Stop 2 in nighttime scene A, the participants reported more hot spots (higher arousal), but Lusk’s graphic method showed more positive labels, suggesting that the illuminated pavilion and pathway evoked positive emotions with heightened arousal. This indicates that landscape lighting can induce positive high-arousal emotions, which may enhance the restorative quality of nighttime environments. However, it is worth noting that these findings are not entirely consistent with the valence dimension measured by subjective scales. This discrepancy may be attributed to two factors: (1) subjective scales rely on retrospective evaluations of the overall walking experience, whereas physiological measures capture real-time responses at specific locations; and (2) the arousal dimension in subjective scales may not fully reflect physiological arousal.

Although SCL data are widely used in outdoor walking studies to measure emotional arousal, these data are limited to capturing only the arousal dimension and cannot represent valence. To address this limitation, our study integrated Lusk’s graphic method, which provides insights into the valence dimension, enabling a more comprehensive analysis of the relationship between environmental features and the two-dimensional emotional responses. Additionally, the in-depth interviews and visitor-employed photography further revealed the specific environmental elements and the participants’ perceptions that triggered these emotional responses, offering a deeper contextual understanding and traceability of emotional experiences. The consistent patterns observed across all three lighting scenarios demonstrate that the combined use of SCL measurements and traditional methods is a promising approach for evaluating nightscapes under different lighting conditions.

4.2. Impact of Nighttime Lighting Design on Emotion and Restorative Perception

Nighttime darkness often elicits negative emotions, such as fear and stress, which may diminish restorative perceptions [84]. Empirical studies indicate that perceptions of landscape restorability and safety are significantly lower at night than during the day in the same environment [54,55]. However, some scholars have suggested that the visibility of nighttime landscapes, shaped by artificial lighting design, plays a critical role in restorative perceptions. To enhance nighttime restorative quality, features like illuminated waterscapes [57,85], green plants [86,87], or colorful flowers [88] may serve as effective optimization strategies. Under well-designed lighting conditions, the restorative quality of nighttime landscapes may rival or even surpass their daytime counterparts.

Our study corroborates these findings, demonstrating that the lighting conditions in the same space significantly influence emotions and restorative perceptions [54,89]. Although the PRS scores did not show statistically significant differences across the three scenarios, the data trends indicated that nighttime scene A had the highest perceived recovery (4.59 ± 1.71), followed by that of the daytime scene (4.40 ± 1.76) and nighttime scene B (4.33 ± 1.77). Furthermore, the Self-Assessment Manikin (SAM) scale revealed significant differences in emotional perceptions before and after walking. Specifically, in nighttime scene A, the participants’ arousal levels significantly decreased after walking (p = 0.018 < 0.05), a trend not observed in the other two scenarios. This suggests that walking in nighttime scene A led to a calmer emotional state.

To further explore these findings, we integrated skin conductance measures (SCL) and Lusk’s graphic method with spatial analysis to map the participants’ arousal and valence across the three scenarios [5]. Combined with the in-depth interviews and photo analysis, this mixed-method approach allowed us to identify specific environmental features that influenced the emotional responses. For example, at the junction of Area 2 and Area 3 in nighttime scene B, Lusk’s graphic indicated negative emotional labels, and the SCL results showed hot spots (higher stress levels), suggesting that this dimly lit crossroads evoked negative emotions and high arousal. Interviews confirmed that the lack of landscape lighting in this area made it feel dark and unsettling. In contrast, at Stop 2 in nighttime scene A, Lusk’s graphic showed positive emotional labels, and the SCL results indicated hot spots (higher arousal levels), suggesting that the illuminated pavilion and pathway evoked positive emotions and visual interest. The interviews revealed that the participants found this area visually appealing due to the strategic use of landscape lighting. Additionally, as discussed in Section 3.5, the integration of landscape lighting at Stop 2 and Stop 3 in nighttime scene A led to more positive emotional responses and increased arousal compared to those of nighttime scene B. Although these findings do not support robust statistical tests, the mixed-method framework demonstrates promising potential for evaluating nighttime lighting environments. These results suggest that a well-designed lighting environment has the potential to enhance its restorative quality to levels comparable to or even exceeding daytime settings.

While well-designed nighttime lighting enhances restorative quality and emotional well-being, excessive artificial lighting can lead to adverse health effects and reduced landscape preference [49,90]. Thus, landscape and lighting must be co-designed to balance functionality, aesthetics, and emotional needs, ensuring restorative benefits without compromising health or sustainability [91].

4.3. Causes of Differences in Perceptions of Daytime and Nighttime Landscapes

Using visitor-employed photography tasks and in-depth interviews, we analyzed the reasons behind the perceptual differences between the daytime and nighttime landscapes. Changes in lighting conditions (natural vs. artificial light) significantly affect the visibility of landscape elements, altering individuals’ focus and shaping perception. During the day, sunlight reveals all landscape elements to the human eye, providing abundant visual stimuli. At night, artificial lighting defines the visible range, concentrating attention on illuminated objects. With only street lighting, the participants’ attention is almost exclusively drawn to streetlights or light emanating from buildings. Introducing landscape lighting diversifies the focal points, with illuminated pavilions and building facades enhancing the landscape’s attractiveness.

As the environmental brightness decreases from daytime scenes to nighttime scenes A and B, the participants’ focus shifted from diverse landscape elements to illuminated objects, and eventually to the light sources themselves. Although reduced brightness limits the visibility of landscape elements, it does not necessarily negatively impact human perception. The Content Analysis results (Section 3.3.2) indicate that low-brightness environments, such as nighttime scene B, create a mysterious atmosphere that reduces visual distractions and encourages greater imagination. This finding aligns with discussions on the aesthetic value of nightscapes [92]. This is consistent with previous studies [93]. When the participants’ focus was imbued with personal emotional memories, it led to positive cognition.

The visibility of nighttime landscape elements largely depends on lighting design, placing higher demands on designers. Beyond meeting functional requirements, designers should consider the users’ emotional and restorative needs when deciding which elements to illuminate. For example, illuminating green plants has been shown to enhance restorative perceptions and reduce fear [86].

4.4. Limitations

Our study had several limitations. First, while outdoor studies provide realistic environments, they also introduce uncontrollable variables. Measures were taken to maintain consistency across scenarios, such as conducting experiments under overcast, rain-free conditions to ensure stable daylight intensity, scheduling during quiet periods to reduce noise and traffic interference and choosing autumn for its mild thermal environment. However, as the experimental route was not closed to external disturbances, certain uncontrollable factors may have affected the results. Second, this study, constrained by a small sample size, primarily explored qualitative conclusions through multiple research methods. The small sample size was chosen due to the complexity and resource-intensive nature of integrating biosensors with mixed-method approaches in real-world settings. While this limits the generalizability of the findings, the detailed insights provide a critical foundation for campus lighting research. Third, the retrospective self-reports used in this study may have been affected by memory biases, leading to potential mismatches between subjective and physiological responses.

4.5. Future Research

This study provides a foundation for future research on the impact of landscape lighting on pedestrians’ emotional enhancement and restorative perceptions. By integrating SCL measurements, self-report tools, and Lusk’s graphic method, the effects of different lighting scenarios on participants’ emotional perceptions can be effectively captured. Future research should expand the sample size and include more diverse participant groups (e.g., gender, age, and cultural background) to achieve more statistically robust results and promote inclusive lighting design. Additionally, to isolate the effects of lighting, experiments could be conducted in controlled environments (e.g., closed experimental sections), where factors such as sound and thermal perception are regulated [94,95], while systematically examining specific lighting parameters (e.g., color temperature and illuminance levels).

5. Conclusions

This study employed biosensor technology and multiple traditional methods to assess the impact of campus lighting environments on college students’ emotional responses and restorative perceptions, while also developing a novel method for evaluating nighttime landscape lighting. The key findings and contributions of this study are summarized as follows:

(1)

An examination of the impact of different lighting scenarios on emotional and restorative perceptions:

The findings revealed that nighttime scene A (with landscape lighting) scored higher on the PRS compared to other scenarios, indicating a greater restorative potential, while the SAM scale showed a significant decrease in arousal levels after walking in nighttime scene A, suggesting a calmer emotional state. The SCL measurements further demonstrated that the addition of landscape lighting enhanced positive emotions, though it also increased arousal levels in some cases. The phenomenological interviews and Lusk’s graphic method supported these results, showing that both the daytime scene and nighttime scene A evoked more positive responses compared to nighttime scene B (with street lighting only), and in certain cases, well-designed campus landscape lighting even surpassed the restorative potential of daytime spaces.

(2)

The proposal of a novel approach to evaluate the effect of campus lighting environments combined with multi-source data:

By integrating SCL measurements, self-report tools, and Lusk’s graphic method, this study expands traditional evaluation approaches and develops a novel framework for assessing daytime and nighttime environments from an emotional perspective. This method enhances the evaluation of emotional and restorative perceptions and provides a new theoretical foundation for understanding the psychological impacts of lighting. The framework has broad potential for evaluating nighttime lighting quality, offering a practical tool for designing emotionally supportive campus environments.

(3)

Elucidation of the differences between and causes of emotional and restorative perceptions in daytime and nighttime scenarios:

Based on the mixed-method approach, this study revealed that the factors influencing the participants’ emotional perceptions of daytime and nighttime landscapes primarily included the overall environmental brightness, color and light perception, visual element exposure, and visual acuity. To enhance the positive emotional and restorative impacts of nighttime landscapes on college students, improvements can be made by illuminating positive elements (e.g., natural landscapes and architectural features) and reducing glare.

In conclusion, the findings emphasize the critical role of well-designed nighttime campus lighting in supporting students’ mental health and providing practical guidance for campus lighting design and planning.