Psychological restoration is the renewal of individual depleted resources caused by adaptation of ordinary demands, such as stressors and challenging tasks (Honold et al., 2016). Restoration involves short-term, mood-like outcomes related to physiological, affective, and cognitive states (Kaplan and Kaplan, 1989). A restorative environment is one whose qualities promote psychological restoration (Hartig, 2011).

The psychophysiological stress recovery theory (PSRT) and Kaplan's attention restoration theory (ART) are the two dominant theories on restorative environments. Based on the biophilia hypothesis, the PSRT suggests that human beings have a predisposition for unthreatening natural settings because an innate preference for natural environments is an advantage for early humans (Ulrich, 1983). The theory argues that exposure to nature can automatically reduce stress of people (Ulrich, 1983). Nature post-stress recovery includes psychological, affective, and cognitive changes related to the recovery from excessive physiological arousal, reduced negative emotions, and an increase of positive feelings (Ulrich et al., 1991). Psychophysiological stress recovery theory relies on successful studies where active or passive transactions with natural settings show affective results measured through self-reported mood scales (Huang et al., 2020). The ART conceptualizes restoration as a process that returns cognitively fatigued people or a person with a high negative affect to a mental state wherein these affectations are reduced (Rydstedt and Johnsen, 2019). This theory suggests that exposure to an environment with restorative attributes helps restore directed attention fatigue. Qualities linked with experiences of being away, fascination, compatibility, and extent are essential to restoration (Kaplan and Kaplan, 1989). These components quantifying the restorative potential are mediators between the physical environmental attributes and restorative outcomes (e.g., Hartig et al., 1997; Lindal and Hartig, 2013; Martínez-Soto et al., 2014b; Marselle et al., 2016). Recent approaches to evaluate the restorative capacity of urban environments have explored various experimental settings at acute and long cumulative mental fatigue or stress levels, considering brief or repeated transactions with restorative periods (Pasanen et al., 2018a; Hunter et al., 2019). These include contrasting nature vs. stressful and controlled environments (Pasanen et al., 2018a). Evaluation of the restorative capacity of urban environments has been conducted in need for restoration conditions (Jiang et al., 2014; Stigsdotter et al., 2017; Subiza-Pérez et al., 2021) and without the need for restoration (Cheon et al., 2019; Kang and Kim, 2019). The need for restoration is commonly induced by causing a psychological deficit of acute mental fatigue or psychological stress. The studies concluded that exposure to restorative built environments relates to psychophysiological changes associated with the restorative potential of a place. Namely, psychophysiological outcomes are related to an improved sense of relaxation and revitalization (Johansson et al., 2011; Van den Berg et al., 2014), a decrease in negative and clinically relevant mood states, reduction in perceived stress (Brooks et al., 2017; Hazer et al., 2018; Bielinis et al., 2021), and an increase in attentional performance (Subiza-Pérez et al., 2021).

Research trends in restorative environments group into four categories: (a) comparative studies on advantages of restoration of natural vs. urban built settings (Ulrich et al., 1991; Pasanen et al., 2018a), (b) studies on the exposure to multiple green space typologies (Matsuoka and Kaplan, 2008; Honold et al., 2016), (c) studies relating to the dichotomy nature vs. architectonic/historic built environments (Karmanov and Hamel, 2008; Scopelliti et al., 2019), and (d) studies in which architectural qualities of indoor and outdoor scenarios result in restorativeness beyond greenness (Lindal and Hartig, 2013; Subiza-Pérez et al., 2021).

Restorative built spaces include green spaces (Lorenzo et al., 2016; Stigsdotter et al., 2017), residential and non-residential streetscapes (Gidlow et al., 2016; Kabisch et al., 2021), blue spaces (Gidlow et al., 2016; Subiza-Pérez et al., 2020; Kajosaari and Pasanen, 2021), cultural and historic places (Ouellette et al., 2005; Clow and Fredhoi, 2006; Hidalgo et al., 2006; Fornara and Troffa, 2009; Herzog et al., 2010; Xu et al., 2018; Subiza-Pérez et al., 2021), and, in a lesser extent, indoor built spaces (e.g., shopping malls and coffee shops, Staats et al., 2016; Payne and Guastavino, 2018; work environments, Raanaas et al., 2011; Korpela et al., 2014). Restorative attributes of these space categories relate to their aesthetic qualities (Galindo and Hidalgo, 2005; Lindal and Hartig, 2013; Subiza-Pérez et al., 2020), the degree of naturalness, diversity of vegetation, tree cover density, functionality, and extension (Van den Berg et al., 2014; Ettema, 2016; Lorenzo et al., 2016). These attributes have been consistently studied as predictors of restorative experiences. However, other environmental qualities beyond nature traits significantly contribute to the restorative potential of a place. The levels of exteriority and visibility, which promote visual and locomotive permeability, relate to the perceived restorativeness (Hauru et al., 2012), the restoration likelihood (Lindal and Hartig, 2013), environmental preferences (Herzog, 1992), the attentional performance, and mood states (Huang et al., 2020). Psycho-environmental indexes of the enclosure are used to quantify these attributes (Subiza-Pérez et al., 2021). Walkability potentializes restorative effects (Gatersleben and Andrews, 2013; Zuniga-Teran et al., 2017; Lin et al., 2019) and is a predictor of better health (Honold et al., 2016). The presence of landmarks in urban spaces (see Bala, 2016) facilitates wayfinding, promoting restorative outcomes (Gatersleben and Andrews, 2013; Pasanen et al., 2018a). Human-made buildings that sustain various social and historic meanings could reach similar restorative potential as natural areas (Xu et al., 2018). Psychological or physiological impairment has not been observed in these spaces (Stigsdotter et al., 2017). Street proximity relates to deleterious restoration effects (Bornioli et al., 2018). People density also hinders restorative environmental outcomes (Staats and Hartig, 2004; Galindo and Hidalgo, 2005).

Non-visual modalities, such as the sense of hearing and smell, play an essential role in restorative experiences. Natural sounds have a restorative potential (Ratcliffe et al., 2016) due to induced mood and cognitive recovery effects (Benfield et al., 2014; Abbott et al., 2016; Shu and Ma, 2018). Odors directly influence the experiences of people of their surrounding environment and proportionate a rich set of localized information that enhances the sensory experience with the natural environment (Henshaw, 2014; Shaw et al., 2015). Olfactory pleasantness from natural smells enhances positive moods and reduces negative feelings (Jo et al., 2013) and stress (Hedblom et al., 2019). Odors found in urban environments (e.g., caused by air pollution) relate to sustained stress (Hedblom et al., 2019). Therefore, olfactory sensory inputs are relevant when creating environments that reduce stress and promote restoration (Henshaw, 2014; Sowndhararajan and Kim, 2016; Truong et al., 2019).

Wilson (1984) borrowed the term “biophilia” from Erich Fromm (1984) to suggest that human beings possess an innate tendency to seek connections with nature and other forms of life. Such tendency refers to “focus upon life and lifelike forms, and, in some instances, to affiliate with them emotionally” (Wilson, 2002, p. 134). The concept of affiliation is equivalent to a trait measure related to the experiential sense of oneness with the natural world (Wilson, 1993). These sets of experiences are used to describe the deep sensations of exploring the natural and the cognitive and affective manifestations that arise while inhabiting such environments (Appleton, 1975). From the early conceptions of biophilia to the biophilia hypothesis, how greenery in built environments affects the human connection with nature has been the main topic of interest (Söderlund and Newman, 2015). Biophilic design is an approach that attempts to connect people with nature (Kellert, 2008; Browning et al., 2014). Several elements and attributes of biophilic design have been elaborated through the intuition of the designer (for a review of such attributes, see Kellert, 2008, 2018; Browning et al., 2014; Ryan et al., 2014; Sturgeon, 2017), thus supporting the human needs of affiliation with nature through the design of indoor and outdoor built spaces (Söderlund, 2019). Some psycho-environmental measures of biophilic design are included in the biophilic quality index by Berto and Barbiero (2017). Large-scale application of biophilia can also be found in the biophilic cities movement (e.g., Beatley, 2016) and in the field of cultural ecosystem services (Chang et al., 2020). Biophilic designers incorporate biophilic qualities (e.g., the imitation and use of patterns and shapes that relate to natural elements; Söderlund and Newman, 2015) to promote similar psychological benefits to those obtained with authentic natural experiences (Salingaros, 2007). Recent work on biophilic design has investigated the psychological impact of nature-like features in built environments (Gillis and Gatersleben, 2015). These include mood and cognitive functioning improvement, increased positive emotions and creativity, and life satisfaction (Ortegón-Cortázar and Royo-Vela, 2019; Roskams and Haynes, 2020). Experimental studies have also observed low skin conductance and decreased blood pressure and heart rate (Yin et al., 2018; Shen et al., 2019). These findings indicate that an individual connection with nature could determine the ability of people to positively evaluate the restorative potential of a natural environment (Berto et al., 2018). The latter findings agree with the notion that while the biophilic affinity is focused on individual preferences linked with connectivity with nature, the restorativeness of the place is an environmental quality (Berto et al., 2018). Our research points out the role of several environmental biophilic qualities in the promotion of the cognitive, emotional, and physiological components of the recovery and restoration processes (Rydstedt and Johnsen, 2019). The latter suggests that PSRT and ART support the biophilic design practice (McSweeney et al., 2019; Shen et al., 2019; Yin et al., 2020). However, despite advances in understanding the physical and physiological effects of biophilic features in built environments, several researchers agree that further evidence is required to characterize such effects quantitatively. Existing studies have focused on measurements of the impact of biophilic qualities on the restorative responses and the overall benefits of biophilic spaces (Yin et al., 2018, 2020; Coburn et al., 2019; Shen et al., 2019).

This article identifies the biophilic design features and other environmental qualities in a sample of 65 interior and exterior spaces in a city of a middle-income country (Leon Guanajuato, Mexico). We examine the restorative influence of a set of environmental qualities that cross the barriers between the ecological, spatial, urban, architectural, psychosocial, and sensory aspects of places. A further step of this study is to establish an appropriate characterization of such scenarios according to these qualities. Along with the latter attributes, the study includes the evaluation of the places regarding their restorative quality, preference, mood states, and situational stress. A structural equation model summarizes the statistical links between the included covariates.

The research questions of this study are as follows:

As for Q3, we have proposed a conceptual model of the restorative influence of the environmental qualities of the places on mood and stress (see Figure 1). Our model departs from a social ecology theory (Stokols, 1996), which considers the categorical and dynamical inquiry about the people-environment relations based on three elements: the physical setting, the psychological process, and the psychological outcomes (Stokols, 1987; Werner et al., 2002). From a theoretical point of view, the model considers the basic assumptions of ART (Kaplan and Kaplan, 1989) and PSRT (Ulrich, 1983) to explain the restorative effects of the proposed environmental qualities on mood states and psychological stress related to the exposure with indoor and outdoor built settings. The model considers several environmental qualities, such as vegetation proportion, landmark salience, and the biophilic qualities of the built elements. Olfactory pleasantness was included as an observed dimension of the environmental qualities. The model considers the components associated with the restorative potential of a place as those that indirectly influence the relations between the environmental qualities and psychological outcomes. This consideration is supported by evidence, suggesting that the link between the environmental attributes and the restorative effects is mediated by the perception of the restorative qualities of the environment (Hartig et al., 1997; Lindal and Hartig, 2013; Martínez-Soto et al., 2014b; Marselle et al., 2016; McAllister et al., 2017; Masoudinejad and Hartig, 2020). We also hypothesized that the restorative influence of environmental qualities on moods and stress was related to decreased negative moods and perceived stress, and with higher feelings of positive mood states (McSweeney et al., 2019; Huang et al., 2020). Our proposed model implemented structural equation modeling (SEM) to evaluate how the model fits empirical data and infer the nature and characteristics of the causal associations (Figure 1).

With around 1.6 million inhabitants (Instituto Nacional de Estadística y Geografía, 2015), Leon is the most populous city in the state of Guanajuato, Mexico, and it is the seventh most populous metropolitan zone of the country (Secretaría de Desarrollo Agrario Territorial y Urbano, Consejo Nacional de Población, and Instituto Nacional de Estadística y Geografía, 2018). The town was founded in 1575 (Arias Padilla, 2020). Looking for the representativeness of the settings (Winkel et al., 2009), data were collected in 65 exterior and interior spaces in Leon. These spaces were selected to include interior and exterior places, whether contemporary or historic, with varying proportions of vegetation, different uses, and located in quite distinct areas of the city. The spaces were selected using the methodology described in Medina (2019). Most of the chosen sites are commonly used by people for leisure, shopping, and social interaction, but the study also included spaces like waiting rooms in hospitals and bus stations.

The present study subjects freely visited the place, answered different psychological scales, and took photographs of their surroundings. Pictures of the spaces taken by the participant volunteers enabled a panel of expert architects and psychologists to rate a set of 12 environmental qualities (three levels for each dimension). A study in which the raters evaluated specific psycho-environmental characteristics of places in situ (e.g., tree density, orientation, enclosure) is San Juan et al. (2017). The environmental qualities were evaluated based on the multiple photographs of the places taken by the participants from standing or sitting points of view (Figure 2). Google Street View and aerial photographs were used when the photographs of the participants did not allow for assessing any of the 12 dimensions of a place. The selected 12 dimensions measure the physical and social qualities of places that may have restorative potential and a positive impact on the health and well-being of people. These dimensions are defined in the following paragraphs.

Three outcome measures were considered in our study. These are the Mexican adaptation to the revised perceived restorativeness scale (RPRS) (Hartig et al., 1996; Martínez-Soto and Montero, 2008), the Spanish version of the profile of mood states (POMS) (Arce et al., 2000), and the stress and arousal adjectives checklist (SAACH) (King et al., 1983). We assessed restorative potential of the place, using RPRS. The scale components are derived from the theory of Kaplan of restorative environments (Kaplan and Kaplan, 1989). The RPRS is a 25-item scale that contains five subscales: being away (five items; e.g., “This place is a refuge from unwanted distractions”; α = 0.76), fascination (five items; e.g., “This place is fascinating”; α = 0.81), compatibility (five items; e.g., “It is easy to do what I want here”; α = 0.75), coherence (four items; e.g., “It is easy to see how things here are organized”; α = 0.75), and scope (four items; e.g., “I experience this place as very spacious”; α = 0.81) and two indicators of environmental preference (e.g., “I like this place”). The participants responded on an 11-point scale from 0 (Nothing, does not apply to the experience described) to 10 (Completely, it does apply to the experience). Ratings for 23 items (excluding environmental preference) were averaged to measure the perceived restorativeness of an individual (denominated as global restorative potential, GRP; Martínez-Soto et al., 2014a). Previous studies have shown that the RPRS is sensitive at the subscale level (Korpela and Hartig, 1996; Tenngart Ivarsson and Hagerhall, 2008). Similar to other studies, our research considers scalar scores of 0–4 (low), 5–7 (moderated), and 8–10 (high) to describe the restorative potential of the place (Berto, 2005; Martínez-Soto et al., 2014a). The mood states were measured, using the Spanish version of POMS (Arce et al., 2000), for which the participants rated 63 adjectives expressing mood states on a five-point Likert scale, ranging from 0 (not at all) to 4 (very much). The POMS consist of seven subscales, including vigor (eight items; α = 0.85), friendliness (seven items; α = 0.70), depression (14 items; α = 0.95), anger (12 items; α = 0.87), fatigue (seven items; α = 0.90), tension (eight items; α = 0.82), and confusion (seven items; α = 0.70). There are two options for assessing mood: the past week or now. In this study, the participants were asked to choose the right-now option. The SAACH instrument was used to measure situational stress. The Spanish version adapted by Ortega (2002) in Mexican population is an 18-item scale referring to the constructs of stress (seven items; α = 0.84), arousal (four items; α = 0.77), and exhaustion (four items; α = 0.72). The three subscales are rated, using a four-option response format: 1 (yes definitely) to 4 (definitely not). The odors of the city environment were measured through the sensory component of olfactory pleasantness and were rated using a single item “How pleasurable is the odor in this place?” 1 (very unpleasant) to 3 (very pleasant) (Baron and Kalsher, 1998; Hedblom et al., 2019). Finally, noise annoyance was evaluated through the item “How annoying is the noise in this place?” with the response options 1 (nothing) to 3 (very much). A description of the psychological measures used in our study and the corresponding rating scales are listed in Table 1.

A mobile crowdsourcing platform (Ruiz-Correa et al., 2017), which includes a mobile application (called “UrBis”), enabled the administration of psychometric instruments in the field. UrBis allows app users to collect geolocalized multimedia data (photos, videos, audios, and text). UrBis users had to read and agree to the consent form to activate the app. The GPS location where each user completed the psychometric instruments was the only personal information collected. All mobile data collected remain in a Tier 3 data center at the National Supercomputer Center, sponsored by the Consejo Nacional de Ciencia y Tecnología (CONACYT). Data storing procedures followed a strict protocol that complies with the Mexican law concerning the security and privacy of contributions of users. Data minimization procedures enabled quantitative analysis requiring the safe use of GPS data. Minimization procedures include the use of GPS data with limited accuracy to prevent personal information disclosure. The UrBis app was developed for the Android platform (Figure 3). At the beginning of the survey, the participants were asked to complete a brief battery of demographic information (i.e., sex, age, the highest level of completed education, zip code for the location where they have lived for the longest, years of residence in the city, etc.). Correct completion of these data was required to proceed to the administration of the psychological scales.

Two trained researchers visited the 65 spaces. Upon arriving at the places, they approached the site visitors, explained the study objectives, and invited them to participate. Two eligibility criteria were considered: (a) participants older than 16 years, and (b) city residents and visitants that are frequent users of the place (tourists were excluded from the study). All the participants first signed the informed consent and received instructions in collecting data, using UrBis mobile app. Then the investigator asked the participants to respond to a survey on sociodemographic information. Before answering the scales, the participants were asked to realize a 3-min free exploratory itinerary of the place to allow them to observe and appreciate the place without any constraint (de la Fuente Suárez, 2020, in press). During this experience, the participants were instructed to take photographs of the place. Then the participants were asked to answer the digitized versions of the RPRS, SAACH, and POMS to report the olfactory pleasantness of the place, noise annoyance, restorative potential, situational stress, and mood states they experienced when physically present in the explored setting (Jiang et al., 2019). The average response time was 20 min. Data were collected between January 2018 and July 2019, either during the week or the weekend and mainly on sunny days (M = 24.6°C; SD = 3.66). The surveys were conducted between 8–13 h and 18–20:30 h.

A multidimensional scaling (MDS) technique was used to categorize the 65 spaces based on 12 environmental qualities. The MDS is an exploratory data analysis that has been used in several studies relating environmental qualities and perceptions (Green, 1999; Coburn et al., 2019; de la Fuente Suárez, in press). In the present study, the MDS was realized in SPSS (PROXSCAL). The Euclidean distances between distinct places were used as dissimilarities.

Data from 470 collected surveys were included for analysis. Cronbach's alpha was computed for each scale and measure to verify their reliability. Statistical associations between the categories of the place and the psychological measures (restorative potential, moods, and stress) were conducted through non-parametric (Kruskal–Wallis ANOVA) analysis of variance due to deviations of sampled data from the Gaussian distribution. Post-hoc pairwise comparisons, using the Dunn–Bonferroni approach, were automatically conducted for any dependent variables for which the Kruskal–Wallis test was significant. The significance level was set to α = 0.05.

We conducted a confirmatory factor analysis and SEM (Rosseel, 2012; Li, 2016) to infer statistical associations among our measured environmental and psychological variables and five latent constructs: environmental qualities, restorative potential, positive and negative mood states, and situational stress (Figure 1). We used off-the-shelf R programming tools to implement SEM analyses (Rosseel, 2012). The following fit statistics were reported for the model estimation (Hoyle and Panter, 1995): the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR) (Hu and Bentler, 1999). Because chi-square is sensitive to sample size (Kline, 2011), with a high probability of being statistically significant with modest sample sizes (Iacobucci, 2010), we did not use it as a model fit measure. Instead, we followed the suggestions of Hu and Bentler (1999) and a set of cut-off criteria for a good fit (a CFI close to 0.95 or higher), RMSEA (close to 0.06 or lower), and SRMR (close to 0.08 or lower). Likewise, the percentages of variance explained (R²) in the outcomes latent variables were evaluated (Kline, 2011).

By performing a MDS analysis (Figure 4), the 65 locations were grouped in eight categories of places, considering their similarity regarding the 12 environmental dimensions evaluated by the group of experts (Euclidean distances between data of the locations). The circles representing places in Figure 4 are closer to each other when they present more similarities regarding the 12 environmental dimensions. The resulting categories are (a) historic squares with biophilic architecture, (b) historic interiors and courtyards with biophilic architecture, (c) large parks, (d) small parks and other green areas, (e) street scenes, (f) exteriors with non-biophilic architecture and vegetation, (g) exteriors with non-biophilic architecture without vegetation, and (h) interiors with non-biophilic architecture. The utilization of MDS analysis allowed grouping the places in categories that emerged from the diagram on the basis of a set of environmental dimensions. The latter prevented that the categories of places were defined a priori and allowed for the discovery of categories that were more related to the particular nature of the studied places. In Table 2 may be noticed that six of the eight categories of places mainly correspond to exterior spaces (high level of exteriority). Large parks have the highest means of visibility and walkable surfaces of all types of places. This type of place also has the highest proportion of vegetation. The category of small parks and other green areas includes pocket parks, neighborhood parks, and median strips with vegetation. Compared with large parks, small parks are closer to streets. Regarding the street scenes, they are the places with the highest visual access to streets. Historic plazas with biophilic architecture possess the highest landmark salience, biophilic shapes, biophilic materials, and people density of all places. Meanwhile, the exteriors with non-biophilic architecture without vegetation present non-biophilic materials and non-biophilic shapes in their highest degree. Historic plazas and historic interiors are the types of places with the highest values of age of built elements. In opposition to the latter, the exteriors with non-biophilic architecture and vegetation, and the interiors with non-biophilic architecture are the environments with the newest built elements of all.

A Kruskal–Wallis test showed that the category of place significantly affects the GRP [(H7) = 122.86, p < 0.00; η² = 0.25]. The Bonferroni post-hoc method, following the Kruskal–Wallis test, found that the eight categories of places differed in their GRP. Historic squares (mean rank = 219.74) had a higher restorative potential than the small parks (mean rank = 215.76; p < 0.05) and the interiors with non-biophilic architecture (mean rank = 115.82; p < 0.00), which, in turn, differ significantly from each other (p < 0.05). In overall, large parks (mean rank = 340.93) had a higher GRP than exteriors with non-biophilic architecture and vegetation (mean rank = 268.90; p < 0.00), small parks (mean rank = 215.76; p < 0.00), street scenes (mean rank = 180.43; p < 0.00), interiors with non-biophilic architecture (mean rank = 115.82; p < 0.05), and exteriors with non-biophilic architecture without vegetation (mean rank = 136.35; p < 0.05). Both non-biophilic interiors and exteriors with non-biophilic architecture without vegetation differ, respectively, with exteriors with non-biophilic architecture and vegetation (mean rank = 268.90; p < 0.05). Finally, historic interiors (mean rank = 285.47) had a higher GRP than interiors with non-biophilic architecture (mean rank = 115.82; p < 0.05) (Tables 3, 4).

The Kruskal–Wallis H-test revealed that there were statistically significant differences between the mood state of fatigue and the type of environment [(H7) = 20.75, p < 0.00; η² = 0.03]. Furthermore, the post-hoc analysis shows higher scores of fatigue in interiors with non-biophilic architecture (mean rank = 272.48) compared with large parks (mean rank = 209.07) and street scenes (mean rank = 206.65) (adjusted p < 0.05, respectively). For tension, significant differences emerged with respect to the type of environment [(H7) = 19.25, p < 0.00; η² = 0.02]. Post-hoc analysis revealed higher scores of tension for exteriors with non-biophilic architecture without vegetation (mean rank = 324) vs. large parks (mean rank = 209.11) (adjusted p < 0.05, respectively). In confusion, the mean scores were found to have statistically significant differences between the environments according to the Kruskal–Wallis H-test [(H7) = 20.77, p < 0.00; η² = 0.03]. Post-hoc pairwise comparisons revealed statistically differences between interiors with non-biophilic architecture (mean rank = 270.03) and street scenes (mean rank = 194.83) (adjusted p < 0.05). Finally, no statistical differences were observed across the eight categories of settings with respect to the mood states of vigor [(H7) = 14.43, p = 0.04; η² = 0.01], friendliness [(H7) = 8.68, p = 0.27; η² = 0.00], depression [(H7) = 9.18, p = 0.24; η² = 0.00], and anger [(H7) = 8.93, p = 0.25; η² = 0.00].

The Kruskal–Wallis ANOVA-test revealed that the type of environment significantly affects the stress measures [(H7) = 35.82, p < 0.00; η² = 0.06]. Furthermore, the post-hoc analysis shows that large parks (mean rank = 195.48) were experienced with lower levels of stress compared with interiors with non-biophilic architecture (mean rank = 281.52; adjusted p < 0.00) and exteriors with non-biophilic architecture without vegetation (mean rank = 332.46; adjusted p < 0.05). Likewise, small parks (mean rank = 217.17) and interiors with non-biophilic architecture (mean rank = 281.52) were perceived with lower scores of stress than exteriors with non-biophilic architecture without vegetation (mean rank = 332.46) (adjusted p < 0.05). Concerning exhaustion, the non-parametric test shows statistically significant differences between the type of environment [(H7) = 21.82, p < 0.00; η² = 0.03]. The post-hoc analysis revealed that compared with large parks (mean rank = 215.66) and historic squares (mean rank = 221.03), interiors with non-biophilic architecture were perceived with higher scores of exhaustion (mean rank = 294.99) (adjusted p < 0.05, respectively). No statistical differences were observed across the eight categories of settings with respect to the arousal dimension [(H7) = 8.31, p = 0.30; η² = 0.03].

The mean score of the perceived environmental quality of olfactory pleasantness measured by the item “How pleasurable is the odor in this place?” was found to have statistical significant differences between the groups of environments [(H7) = 95.57, p < 0.00; η² = 0.19]. Subsequent post-hoc pairwise comparisons (Table 5) revealed that, overall, large parks were perceived with higher levels of olfactory pleasantness (mean rank = 312.29) than street scenes (mean rank = 173.46; adjusted p < 0.05), exteriors with non-biophilic architecture without vegetation (mean rank = 180.88; adjusted p < 0.05), interiors with non-biophilic architecture (mean rank = 184.96; adjusted p < 0.00) and historic squares (mean rank = 184.4; adjusted p < 0.00). Likewise, exteriors with non-biophilic architecture and vegetation (mean rank = 283. 49) were perceived with higher scores of olfactory pleasantness compared with interiors with non-biophilic architecture (mean rank = 184.96; adjusted p < 0.00), historic squares (mean rank = 184.4; adjusted p < 0.00), and street scenes (mean rank = 173.46; adjusted p < 0.00). In the same way, street scenes were perceived with lower scores of olfactory pleasantness compared with small parks (mean rank = 253.49; adjusted p < 0.00). Finally, small parks were perceived with higher values of olfactory pleasantness than historic squares (mean rank = 184.4; adjusted p < 0.05).

Mean scores related to noise annoyance (“How annoying is the noise in this place?”), with the response options 1 (nothing) to 3 (very much), were statistically different across the environmental categories according to the non-parametric Kruskall–Wallis H-test results [(H7) = 80.17, p < 0.00; η² = 0.15]. The post-hoc analysis revealed (Table 6) that large parks (mean rank = 159) were self-evaluated with lower noise than exteriors with non-biophilic architecture (with and without vegetation) (mean rank = 217.17 and 327.5; adjusted p < 0.05, respectively). In the same manner, large parks were perceived as less noisy than the historic squares (mean rank = 259.31; adjusted p < 0.00) and interiors with non-biophilic architecture (mean rank = 296.38; adjusted p < 0.00). After large parks, small parks were perceived with lower levels of noise annoyance compared to exteriors with non-biophilic architecture without vegetation (mean rank = 327.35; adjusted p < 0.05), historic interiors and courtyards (mean rank = 326.78; adjusted p < 0.05), and interiors with non-biophilic architecture (mean rank = 296.38; adjusted p < 0.05). Finally, both historic interiors and interiors with non-biophilic architecture were perceived with higher scores of noise annoyance compared to exteriors with non-biophilic architecture and vegetation (adjusted p < 0.05, respectively) (Table 6).

Lavaan software package (Rosseel, 2012) enabled statistical inference of our structural equation model (Figure 5). The package reports a CFI of 0.96, an SRMR value of 0.07, and an RMSEA equal to 0.05 (with p < 0.001) with a confidence interval of (0.42, 0.53), suggesting a moderately good model fit. For our data set, the package uses the maximum likelihood estimation with robust standard errors and a mean and variance adjusted test statistics. Table 7 summarizes the descriptive statistics of the model, while the covariance matrix of the latent variables is shown in Table 8. The structural equation allowed us to characterize five latent variables (environmental qualities, restorative potential, situational stress, positive mood states, and the negative mood states) that link the psychological and environmental variables considered in the study. The environmental qualities variable considers all the physical, architectural, and sensory aspects that result in restoration. Such a variable is linked to the following five dimensions: vegetation, spatial extension, landmark salience, biophilic architecture, and olfactory pleasantness. Spatial extension groups the exteriority, visibility, and walkable surface variables. These three variables are significantly correlated. The spatial extension variable is related to the depth dimension proposed by Ulrich (1977, p. 5), which is motivated by the observation that “[.] if a man has a predilection for exploring and actively seeking information, then spaciousness or depth should be favored. The presence of space signals opportunities for entering a scene and garnering new information…”

The biophilic architecture variable is a general score of a place that reaches its maximum value when the age of the built elements and the biophilic materials and shapes are in the high levels, while the non-biophilic materials and shapes are in the low levels. Therefore, biophilic architecture corresponds to the sum of the values of age of the built elements, biophilic materials, biophilic shapes, non-biophilic materials (inverted), and non-biophilic shapes (inverted). All dimensions included in this composite variable are statistically correlated. The inclusion of people density, street visual access, noise annoyance, and arousal variables hindered the model fit results. The final model did not consider these variables (Hatcher, 1994).

Our structural model shows that both situational stress and the positive and negative mood states indirectly depend on the latent environmental quality variable through the restorative potential variable. The directional associations are illustrated through single-headed straight arrows with standardized and significant path coefficients, indicating the strength of associations. Size effects are described according to an absolute standardized direct result where a value smaller than 0.1 is considered as an indication of a small effect, a value around 0.30 is regarded as an indicator of a typical or medium impact, and a value greater than 0.5 is considered as an indicator of a more significant effect (Bollen, 1989). Referred R squares (R²) of the endogenous variables represent the percentage of the variance of each endogenous variable (Kline, 2011).

The stress variable has the most significant negative and direct path coefficient with the restorative potential variable, with a large-sized effect (−0.63) and 39% of explained variance. The negative mood states variable has a moderate impact (−0.34), with 12% explained variance. The positive mood states variable has an average effect (0.24) and 6% of the explained variance with a moderate effect size. The environmental qualities latent variable directly enhances (with a path coefficient of 0.73) the restorative potential (Figure 5), which, in turn, has negative path coefficients with stress (−0.63), negative mood states (−0.34), and a positive path coefficient with positive mood states (0.24). Figure 5 shows an indirect negative association between the environmental quality variable, the situational stress (−0.46), and negative mood states (−0.25) through restorative potential. Likewise, the SEM also shows that the positive mood states involve the indirect linear effects of environmental qualities through restorative potential (0.18) (see Table 8). In summary, an increase in the environmental qualities values increases the restorative potential and the positive mood states (with a combined path coefficient of 0.17). It decreases the situational stress and negative mood states.

A MDS technique was used to categorize the 65 spaces based on 12 environmental qualities rated by a panel of experts. The use of MDS analysis based on such evaluations allowed for generating specific categories of real places found in a Mexican city. The present study focused on the significant differences regarding the restorative qualities of the categories of place having different magnitudes of the environmental dimensions as they naturally occur. The holistic selection of the categories of place played a role in the significant differences between the categories regarding their restorative potential. Other environmental dimensions, e.g., related to the use or function of the spaces, should be inquired in future studies.

As expected, the environmental qualities had an indirect path through the restorative potential. According to the model in Figure 5, there was a direct and positive effect between environmental qualities and restorative potential. As observed, the construct environmental qualities conformed by the dimensions of (a) vegetation, (b) spatial extension, (c) landmark salience, (d) biophilic architecture, and (e) olfactory pleasantness share a significant percentage of explained variance with the restorative potential variable, suggesting a relative capability to predict the restorative potential of the place. Our results confirm that the experience of spending time in environments with restorative qualities reduces the perception of psychological stress and negative moods and improves positive mood states, with moderate to high size effects. The relation between affective and stress states reported are congruent with both field and experimental studies on built restorative environments (Johansson et al., 2011; Lindal and Hartig, 2013; Stigsdotter et al., 2017; Hazer et al., 2018; Pasanen et al., 2018b; Hedblom et al., 2019; Hunter et al., 2019; Truong et al., 2019; Kajosaari and Pasanen, 2021; Subiza-Pérez et al., 2021) and related studies on restorative influences of indoor and outdoor biophilic environments (Hoyle et al., 2017; Yin et al., 2018, 2020; Ortegón-Cortázar and Royo-Vela, 2019; Roskams and Haynes, 2020). Further studies of replication with direct and indirect measures of the studied environmental qualities are needed to confirm its restorative influences, considering more robust measures of restorative outcomes (e.g., cognitive and physiological restoration indexes).

Planning a space by paying attention to the spatial extension, biophilic materials and shapes, vegetation, and pleasant smells guarantees the restoration. Following results of this study, it may seem that architects and urban planners should confront the unsolvable problem of designing built environments for people that inherently prefer the natural and despise the non-biophilic built environments. Not all environments are in contact with nature or permit the maintenance of vegetation. There resides the importance of the inclusion of architectural elements being designed with biophilic principles since they solve the aversion to the non-biophilic built elements and allow environmental restoration. It is important to remark that, in this study, the places with higher proportions of biophilic materials and shapes are those with old-style architecture (higher age of the built elements). Other studies could consider the inclusion of historic places without biophilic architecture (e.g., simple antique buildings) and contemporary places with biophilic architecture (e.g., parametric architecture works). In that manner, it would be clearer if what makes a place restorative is its historic value, independently of its biophilic qualities, or if the biophilic materials and shapes are what make a place restorative, independently of its historic value. Four dimensions related to biophilic shapes and materials were presented. More research is needed in which even more specific dimensions are inquired. The dimension of biophilic shapes could be divided into naturalness, curvilinearity, and complexity. Meanwhile, the dimension of biophilic materials could be subdivided into the naturalness of the material and the distinct tactile qualities of the texture.