Measuring the Benefits of Plants: Visual Exposure and Human Well-being Impacts of Lush Streetscapes¹

This paper demonstrates how green landscape elements draw attention and gaze. Studying the preconscious response to trees, shrubs, and plants helps us understand how we process our surroundings. The green industry can educate stakeholders regarding potential health impacts on healthy behavior by promoting more Lush streetscape design.

While the environmental psychology literature is rich with evidence about innate human preference for plants, few studies examine how biometric tools can improve understanding of how the unconscious human mind responds to plants and what that means for cardiovascular health and well-being.

Cardiovascular disease (CVD) is the leading cause of death globally, except in Africa, and is mainly preventable (Martin et al. 2024). While mortality rates for CVD in the United States have steadily declined for the last four decades, this decline has recently begun to plateau, and heart disease remains the leading cause of death (Wall et al. 2018). Physical inactivity is consistently identified as a modifiable CVD risk factor, and it was included in the recent presidential advisory from the American Heart Association’s “Life’s Essential 8” campaign (Lloyd-Jones et al. 2022). In a systemic review, Lacombe and Armstrong et al. (2019) found that people who reported being physically active had a lowered risk, by at least half, of experiencing a CVD event or dying from CVD when compared to those considered inactive.

Many variables impact health, and the present study further explores the relationship between horticulture, the built urban environment, and CVD risk factors of physical activity (PA), diet, and wellness. Research suggests that viewing plants can improve heart health (Song et al. 2017, Kim and Mattson 2002, Park and Mattson 2008, Chang and Chen 2005). In a longitudinal study with elders, Ward Thompson and Curl et al. (2014) found a significant increase in participants’ perception of PA and their neighborhood’s walkability after planting street trees. Of interest is that while there was a perception of an increase in PA, the data did not support an actual increase in walking. Looking at horticulture in urban environments, Clancy and Ryan (2015) and Grinde and Patil (2009) report that the inclusion of biophilic design in urban environments improves mental health and well-being (Sundling and Jakobsson 2023).

The importance of this relationship has been highlighted recently in light of how the COVID-19 pandemic dramatically altered people’s work, life, and recreation habits and, by extension, their exposure time to plants around them. In Bristowe and Heckert’s (2023) review of the use of green infrastructure during the pandemic, they found increased use of neighborhood green infrastructure and appreciation for green infrastructure and its health benefits, such as stress relief. Frank et al. (2022) found a direct association between neighborhoods categorized as walkable, with higher physical activity, and lower rates of obesity and diabetes.

In a systemic review of the hierarchy of walking needs, Paydar et al. (2021) outline how natural elements and greenery are shown to have a notable impact on stress reduction and improved mental health of pedestrians. Higher levels of green space in neighborhoods have been associated with healthier cortisol levels (Roe et al. 2013). Urban life and urban stressors are factors identified as motivating people to look for areas with more green space (Kuo and Sullivan 2001).

A growing body of literature measurably demonstrates the power of priming in shaping human behavior (Harris et al. 2009, Yi 1990, Fishbein and Yzer 2003). Less well-studied is how priming around positive images of plants can impact people’s health and well-being. Korpela and Klemettila (2002) primed subjects with either natural or built stimuli, finding that the former group responded to joy more quickly. Another study by Hietanen and Korpela (2004) showed that environmental scenes with higher restorative levels were associated with positive emotions. Marselle et al. (2020) identified street trees as an important component of urban green space. They studied their effect on mental health by examining the anti-depressant prescription rate across a scale of street trees present. Their findings suggest that unintentional daily contact with nature through street trees close to the home may reduce the risk of depression, especially for individuals with low socio-economic status.

Eye-tracking has been used in graphic design and other computer-based visuals to measure attention, fixation, or concentration on an image (Poole and Ball 2006, Duchowski 2017, Holmqvist et al. 2011). While viewing real-world objects will generate greater autonomic responses than viewing images of those objects, the ability to validly measure those responses is diminished. Therefore, by using eye-tracking software, a researcher can validly and reliably capture autonomic (preconscious) responses to still images on a screen. There has been rapid growth in these types of studies, including a number by leading horticulture scholar Professor Bridget K. Behe at Michigan State University (Behe et al. 2013, Behe et al. 2017). Rihn et al. (2016) have shown that eye-tracking is an effective tool to measure how people’s gaze, in concert with a survey of stated preferences, can capture consumer preferences for ornamental plants. More recently, Behe et al. (2020) used eye-tracking with surveys to evaluate store plant displays. Eye-tracking has allowed the study of how people perceive landscapes (Dupont et al. 2013, De Lucio et al. 1996) and urban scenes (Hollander et al. 2020a, Hollander et al. 2021).

Biometric studies provide a new language in which words like fixations and preattentive processing enter the urban planning lexicon to show the conscious and unconscious impact of street design elements on increasing physical activity (Zaccoro and Atherton 2018) and promoting walkability (Sussman and Hollander 2021). In his seminal book on urban design, Jan Gehl’s Cities for People (2013) suggests that high-quality environmental edge conditions help invite visitors and create highly desired public spaces. Also, the cost of adding natural elements can be low, making the cost-benefit higher.

The present study utilized stimuli manipulating the edge conditions of streets and sidewalks, hypothesizing that images with Lush horticultural elements will correlate with eye movements and fixations. By analyzing participants’ physical activity levels, dietary intake, and wellness scores four weeks after exposure, we examined whether a measurable relationship exists between exposure to horticultural design elements and improved cardiovascular health behaviors.

This study evaluated participants’ preconscious responses to images with either Lush (with plants or trees in view) or Barren (with little to no plants or trees) conditions. These biometric responses were captured with eye-tracking, heatmaps, and facial analysis software. Key research questions were asked:

1. How can exposure to horticultural elements impact the physiological indicators of stress as measured by facial expression?

2. How can eye-tracking advance understanding of attentiveness to plant images versus banal images? Which plant characteristics draw the most attention?

3. Does exposure to horticultural elements in streetscape images have a measurable effect on self-reported physical activity levels, nutritional intake, and wellness scores four weeks later?

4. What impact did images of plants have on participants’ reported health and gardening practices, including whether they purchased more green industry products over the prior four weeks?

5. Are there significant differences in scores on wellness questions for persons with two or more plants in their home compared to persons who report having zero or one plant in their home?

In this study, we used survey and eye-tracking methodology in a pre-post evaluation design to assess the potential impact of a slideshow intervention presenting images of Lush versus Barren streetscape scenes on self-reported physical activity (PA), dietary intake (DI), wellness scores, and plant behavior. This study was approved by our university’s institutional review board in September 2022. Data was collected between December 2022 and May 2023.

The research team collected study stimuli, images, and videos by photographing the streetscape in a neighborhood in Medford, Massachusetts (USA). All images were framed as if from the eye-level vantage point of a pedestrian, standing in the foreground of the scene. Images were all captured at a minimum of 300 dpi resolution in landscape orientation (1920 ×1080 pixels, Bit depth 32).

In total, we collected over 200 images throughout three visits. Of these, ten images were selected to be altered using Photoshop (Adobe Photoshop 2023, Adobe Inc., San Jose, CA) to create novel images that either introduced or deleted horticultural streetscape elements, creating two conditions for a slideshow (Lush and Barren). For example, horticultural elements such as trees or flowers were added, sections of the street and/or sidewalk were deleted and/or replaced with alternative materials, and colors were edited to create consistent variables such as sky/grass color across all images. See Figure 1 for a sample.

This study also presented video stimuli to participants and analyzed fixation and gaze path data using iMotions technology (described further, imotions.com). Three videos were recorded on researchers’ cellphones (Apple and Android) cameras in landscape orientation from the perspective of a walking pedestrian (04-sec duration; 1920 × 1080 frame; 30.16 frames/second). These videos were then duplicated and altered, resulting in three Lush and three Barren streetscape videos included in each condition, respectively.

Survey instruments were administered pre-intervention (baseline) and again four weeks later (post-intervention) by providing a link to a secure website to collect participants’ responses. The participants’ responses and survey instruments were constructed using the Qualtrics platform (Qualtrics, Provo, UT. 2023), ensuring the confidentiality of all participant data. Survey questions included:

• Physical activity level in the past 7 days: International Physical Activity Questionnaire, Craig et al. 2003 (referred to as IPAQ),

• Intake of fruits and vegetables in the past 7 days: National Cancer Institute screener, Thompson et al. 2000, (referred to as NCI)

• Overall health and well-being: Questions from Short Form (SF) 36 General Health Questionnaire, Brazier et al. 1992 (referred to as SFQ)

• Plant behavior activity: Questions regarding caring or shopping for plants; importance of plants in home (referred to as PBA)

In both the “baseline” and “post-intervention” stages, we used two versions of the iMotions software to capture participants' unconscious and emotional reactions. The study integrated the analysis of online recordings seamlessly into the iMotions Online version, encompassing qualitative questions and pre-recorded stimuli, such as pictures and videos. Following the recordings, iMotions 9.3. was utilized for comprehensive analysis of all data collected throughout the study. Biometric software (iMotions 9.3.1) utilizes algorithms that detect facial landmarks and apply rules based on psychological theories and statistical procedures to classify emotions (Stocki et al. 2018, iMotions 2016). The iMotions online platform allowed us to reach a diverse population unbounded by proximity or affiliation with a laboratory setting, as it could be accessed by anyone with a desk or laptop computer with a working camera.

The inclusion criteria were: 1) being a resident of the United States, 2) being 18 years or older, 3) having the ability to speak and read English, and 4) having limited plants in the home. Han and Ruan (2019) conducted a systemic review of 45 papers researching the effects of indoor plants on subjective psychological perceptions and concluded: “that indoor plants positively affect self-reported perceptions.” In order to control for the level of daily exposure to plants, respondents reporting zero or one plant in the home were included in the intervention study; those with two or more plants reported were invited to take a survey that included only the SFQ and PBA questions to compare in a separate section of the study.

The study was advertised through social media platforms (the university department’s Facebook page and newsletter advertisements, Reddit posting platform, and sharing the advertisements with local universities’ urban planning departments). A link to the recruitment and study information was provided.

Between December 20, 2022, and December 27, 2022, there was a suspicious, overwhelming response to our advertisements (N=847 times link to the Qualtrics survey was accessed), and we temporarily stopped collection to investigate. Researchers quickly realized many of the given answers were fraudulent or scam responses based on multiple duplications from identical IP addresses and locations outside the United States. Security settings were improved, and the study was reopened on January 16. It remained open until March 10, 2023, when 151 verified respondents successfully accessed/were introduced to the study survey and intervention via the iMotions link.

For participants who met inclusion criteria, additional links were presented for consent forms, the Qualtrics platform for the baseline survey, and an embedded link for the slideshow intervention via the iMotions platform. Participants were randomly assigned by iMotions software to either Barren or Lush stimuli conditions. Once participants entered the iMotions platform, instruction slides were included as outlined in iMotions’ guidelines to reinforce proper set-up, room lighting, posture, and head alignment. In addition, we used both pre- and post-calibration slides (n = 13) with instructions before each set to encourage the best performance. After pre-calibration, the intervention slideshow images were automatically presented once to each participant in a randomized order within the Barren or Lush condition (image exposure time: 5 seconds; video exposure time: 4 seconds of zooming in in the examined road) to capture preconscious biometric response. Sixty-seven (67) respondents completed the slideshow presentation. They were then presented with post-calibration slides, reconsent confirmation, and a secure link to a separate survey to provide name and email for compensation.

For biometric data, we employed these recordings’ threshold for inclusion in analysis: 5.0 eye tracking accuracy, 70% eye tracking quality, and 60% face quality⁴ (iMotions R Notebooks, 2022). Following the above quality guidelines, forty-seven (47) participants qualified for inclusion in the analysis: twenty-eight (28) respondents in the Barren condition and nineteen (19) in the Lush condition. We defined attrition by both participants drop-out and quality standards, as described in Holmqvist et al.’s (2023) comprehensive review of eye tracking methodology. Data was then processed via the iMotions algorithm software and downloaded to a secure Dell computer at the [university name removed to preserve anonymity] campus.

Four weeks after study participation, participants who completed the survey and iMotions intervention were invited via email to complete the identical survey again (called the post-intervention survey) via a secure link to the Qualtrics web platform. All participants who completed the intervention, regardless of reconsent status, permission to use images, or recorded data quality, were offered nominal compensation for their participation in the form of a $10 e-gift card, which was sent to the email address that they provided via a secure link to a separate survey thereby collecting this information apart from study data.

After all recordings were completed, the iMotions online platform was locked to prevent new recordings. Subsequently, recorded samples were filtered based on predefined quality metrics, and those meeting the criteria were exported from the online iMotions software and imported into iMotions 9.3. software for further analysis. The samples were then categorized as either Lush or Barren. Heatmaps, gaze paths, facial expressions, and emotional reactions were exported from the software for analysis. The analysis proceeded in two stages. Initially, the entire image of each stimulus was considered as the area of interest (AOI), referred to as full AOI analysis. Subsequently, fractions of the full image displaying sidewalks, streets, and green elements were individually analyzed to provide more specific metrics regarding these spatial elements (AOI polygons). The analysis extended to examining all Qualtrics responses, categorizing them, and creating tables for each category. T-test and Anova tests were performed using Prism software for results where statistical significance was indicated.

Using analysis from aggregate dynamic heatmaps and identifying the first dwell duration metrics, we see how fast an element attracted the participant's eye. In both the Lush and Barren conditions, the first and shortest fixations were related to text, labels, or other humans in the observed stimuli (Fig. 2).

Increased gaze time, as depicted with green, yellow, and red tones in aggregate heatmaps, shows the density of gaze points, indicating increased attraction to an element. In 60% of the Lush stimuli sets, plants acted as the focal points of gaze (higher gaze time shown in red in Fig. 3). The presence of plants also resulted in a larger dispersion of image observation (points shown in green cover bigger parts of the examined stimuli). More of the image details were drew more attention of participants in Lush stimuli versus Barren stimuli.

There was no significant change in gaze time for image pairs in which the cars appeared as part of the road or of the road's parking lot. Although, in general, the fixation count metric was higher for Lush images for two pairs of images, this observation wasn’t the same for all the pairs (Table 1). There was a decrease in gaze time in the above-mentioned sets⁵ in the Lush condition, where the view depth was interrupted, and there was no clear horizon line, as seen in grey in Table 1.

This research significantly contributes to our current understanding of human preconscious response to the environment. It confirms that people initially focus on elements that pose a potential danger (such as cars, streets, and words), followed by Lush environmental elements. These initial fixations are survival elements our brains prioritize for processing (Sussman and Hollander, 2021). Moreover, the broader dispersion area indicates a shift in attention towards green elements, suggesting a heightened interest in plant life within the environment.

Although participants tended to scan the whole image and more fixation counts can be recorded, the zoom-in process minimizes the objects to be observed. This analysis process enables a more realistic identification of the elements on which participants fixate.

In contrast to the finding above based on the full images, honing in on just the observation of the streets and sidewalks showed that the participants generated more fixations on streets and sidewalks for the Barren images. This aligns with research by Harten et al. (2017), which suggests that stressed participants tend to fixate on the AOIs more often. Our study's higher number of fixations further supports this, indicating an increased stress level among participants in Barren environments, as shown in Table 2.

The decreased fixation rate for streets and sidewalks in the Lush image shows the importance of horticultural elements in reducing stress. In Lush images, calmer observation is indicated by decreased fixations and more revisits. Furthermore, all the sidewalks, even those in the background, were spotted in Lush images but did not consistently draw attention in Barren images. This is consistent with the larger areas of dispersion found in our full image analysis. In both conditions, the time of first fixation was longer for streets than for sidewalks. However, participants in the Lush condition detected sidewalks more quickly than those in the Barren condition (Table 3).

A detailed analysis was conducted to evaluate the statistical significance of the observed differences in mean durations of participants' initial fixations on sidewalks across two categorically distinct sets of images. This investigation centered on each participant's first point of gaze fixation on sidewalks as depicted in both the Lush and Barren image categories. For every set of images, the duration of each participant's first fixation on the sidewalk was meticulously recorded and subjected to a T-test. This analytical approach revealed that participants' initial fixations on sidewalks within Lush images occurred significantly more rapidly, as evidenced by a statistical p-value < 0.05 for pairs 1,3,5-10. This pattern was opposed for sets 2 and 4, with p-values of those results equal to p-value < 0.0001. However, it is essential to note that deviations from this general trend were identified only in those sets. Notably, within Lush image settings in those two sets, participants' initial gaze was occasionally diverted to horticultural elements and mailboxes on the sidewalks. This resulted in a perceptible delay in the participants' first observation of the sidewalk itself, underscoring the complexity of visual engagement with urban landscapes and the strong correlation of the horticultural elements to stress reduction and observational patterns (Fig. 4).

Although the context that a plant appears can influence such comprehension, as shown in the presented heat maps, the characteristics of plants that affected the observation process varied. Plant images with more foliage drew less attention but more observation dispersion, while tree trunks generated longer fixations and observation duration, as seen in Figure 5.

In addition to the observation data of full and individual spatial elements, patterns of physical reactions were analysed. The data on facial reactions and emotional recordings provided further evidence for the connection between plants and the decrease in stress levels. Data were organized with a ranked scoring of emotions, as seen in Figure 6. Based on D.H. Lee et al. (2013), eye-widening increases the visual periphery effectiveness of participants by almost 10%. Eye widening and eye closure were present consistently in Lush images, showing a wider interest, but not in Barren images.

According to Arlin Cuncic (2023), pursed lips (stretched and/or tightened) indicate distaste from an observed stimulus. Those measures were higher for Barren than Lush. Combining these facial reaction data with the anger recordings for both categories has shown the Barren participants’ dissatisfaction with Barren images. For Barren images, anger, confusion, and contempt scores appeared consistently and were absent for Lush stimuli. All participants showed anger in all Barren images, while confusion and contempt scores appeared in almost 50% of participants when viewing Barren images (Fig. 6).

Emotions of joy, positive reaction, and surprise were consistently evident in reactions to the Lush images. We noticed a homogeneity in the participants’ reactions to Lush images except those where the horizon was obscured. In these images, there were multiple fixation points with smaller duration, as seen below, with a smaller red point and a dispersion of yellow points as participants were scanning the full AOI image looking for the horizon (Fig. 7).

Looking at images that showed a response of joy-happiness-smile facial expressions, we see in the corresponding heatmaps that gaze and attention were directed to natural elements such as trees, stones, sky, and brown and green colors. These findings are consistent with other research indicating that aesthetics (as defined by the presence of trees, views, and perceived attractiveness of buildings) are associated with subjective well-being (Kent et al. 2017).

Four weeks after study participation, no significant differences were found in diet or health and well-being scores for either group. A comprehensive summary of survey results can be found in Table 4 where diet, nutrition items (juice, fruit, etc.) and overall well-being are presented.

The difference in IPAQ scores four weeks after intervention proved significant for increased moderate physical activity in the Lush group but not for Barren group participants. Though the differences between groups for vigorous activity, walking, and total Metabolic Equivalent Task per minute (Mets/min) scores were not significant, a higher percentage of Lush participants increased their score for total Mets/min than Barren participants (61% vs. 42%). Total Mets/min scores for the Barren group decreased for 58% of participants; in the Lush group, this was a smaller decrease at 33% (Table 5).

To qualify for the study, respondents were limited to having zero or one plant in their home (N=44). Respondents who reported having more than two plants (2-3 plants, N=31; 4 or more plants, N=56) were invited to answer the overall health and well-being and plant-based activity questions. This allowed us to examine baseline data for correlations between these outcomes and the different categories of number of plants in the home (0-1, 2-3, or 4 or more).

When asked how their health compared to one year ago [ranging from much better (100) to much worse (25)], there was a significant difference across the three categories. Those with zero or one plant in the home were the only ones to rate their health as much worse than a year ago, as seen in Figure 8.

Based on our results, the preconscious attention to green elements, i.e., observing all the “green areas” of the image or not, does not seem to influence the purchasing of or interest in plants. The only other significant difference found was for the energy/fatigue domain scores. Using a scale with 100 = “All of the time” and 0 = “None of the time,” responses to questions regarding energy and fatigue were coded and averaged with higher scores indicative of a more favorable health state, i.e., higher energy (Ware and Sherbourne 1992), For those with four or more plants scores for energy/fatigue domain concentrated at 60-80 compared to 40-60 for those with 2-3 plants. For those with zero or one plant, there was a wider range of energy/fatigue scores, as seen in Figure 9.

An analysis of energy fatigue among participants categorized by the number of plants at home revealed noteworthy statistical disparities. Specifically, a comparison between those in Category 0 (0-2 plants) and Category 2 (2-4 plants) exhibited a significant difference, as indicated by a p-value of 0.0063 in the statistical test. Similarly, the comparison between Category 0 and Category 4 (4 or more plants) yielded a p-value of 0.0003 in the ANOVA test, further emphasizing the significance of the differences observed (Fig. 10). These findings underscore the pivotal role of plants in enhancing human well-being, particularly as evidenced by the notable distinctions in energy fatigue across varying levels of horticultural elements within participants' residences.

For plant behaviors, participants with 2-3 plants in the home bought more of them (80%) than received them as gifts (54%). This group also had the highest percentage of reporting shopping for plants in the past four weeks (65%) than those with four or more (55%) or those with 0-1 plants in the home (45%). For participants with four or more plants, 80% of this group bought and received them as gifts.

In conclusion, our study examined the biometric, preconscious response to Lush and Barren streetscapes. We were curious about what attracted attention and prompted emotional responses and asked how priming positive images of plants could shape behavior with the potential to improve health and well-being.

We saw that the position and characteristics of plant elements were related to eye movements and fixations. Conceptually, it is essential to recognize that humans evolved with this natural world and seek out plant images in everyday life (Sussman and Hollander 2021). Lush images attract more attention. When green elements are present, people observe the whole built environment in detail instead of focusing on just one spatial element, making calmer observations of the examined stimuli. More specifically, the research showed the expression of non-anxiety-related emotions whenever the participants observed natural elements.

Edge and horticultural elements impacted this study’s attention to streets and sidewalks. All sidewalks drew attention in the Lush images, but participants did not see some of those same sidewalks in the Barren images. The combination of results strongly suggests the existence of a relationship between the horticultural elements in the streetscape and the increase in physical exercise, which is considered one of the factors influencing well-being.

Hollander et al.’s (2020b) eye-tracking analysis showed that sidewalks presented in study image stimuli elicited a gaze sequence response, which suggested the pedestrian’s eyes may be drawn along the path, encouraging walking. Similarly, Feurerberg et al. (2014) found the presence of sidewalks to have a positive effect (4%) on the tendency for adults to take a walk.

The built environment can be further studied to support the inclusion of Lush elements that promote healthy behaviors such as walking. Further exploration of how Barren images elicit facial expressions of stress, anger, or contempt could also be studied. How do these emotional states impact behavior, stress levels, and, by extension, cardiovascular health? For example, elements of safety, such as the existence of sidewalks, buffer strips, and street trees, affected parents’ decision to allow children to walk to school (Kweon et al. 2021). Improved well-being can be promoted with a further understanding of these dynamics.

We did not find an impact of Lush images on reported gardening and health practices. However, engaging the green industry and its customers directly to compare specific design elements and create new image comparisons of before/after landscaping projects, for example, would be interesting. By considering the preconscious response to images with Lush elements, the green industry can educate designers and customers creating healthier environments and streetscapes.