A Comprehensive Study of Preferences Toward Urban Pocket Parks


 Urban parks are inseparable part of green spaces that have been recognized for developing the surrounding environment’s quality. Aesthetic comprehension in Urban Pocket Parks, as an environment which contains various stimuli, is influenced by multi-sensory perception that includes the integration of non-visual (tactile, auditory and olfactory) and visual stimuli. As the current research’s objective was to determine the impact of multi-sensory on visual preferences, Structural Equations Model results on 16 Urban Pocket Parks in Tabriz, Iran, revealed that there was a significant relation between visual preferences and Sound (Human and Natural sounds), and Smell (Human-Body and Natural smells) stimuli, however, the influence of touch stimuli on visual preferences had not identified significant. As a consequence, evaluating the public experience of Natural Sound-Smell Stimuli and the Human-body Sound-Smell Stimuli are essential in designing urban pocket parks; accordingly, city planners and landscape architects can enhance users’ visual preferences by considering these stimuli.

. On this basis, Lindstrom (2006) indicated that the experiences of multi-sensory can increase the users' preferences toward a speci c product (Lindstrom, 2006).
Visual factors form about 87 percent of our environmental perceptions (Bell, 2013), and there is a huge focus on these factor as our environment has been surrounded by them ( The multi-sensory shape representation suggested by other researchers, brings in the whole sensory system and affects the spatial con guration (Amedi, Jacobson, Hendler, Malach, & Zohary, 2002). It is assumed that in order to form the properties of amenity and the environment by an observer, the connection between all the events that occurred in the environment, the features of all the involved objects, and all sensory inputs' arrangement is used (Visell et al., 2009). Generally, the senses are not isolated in terms of their usage, and the combination of all senses or some of them work in harmony (Bundy, Lane, & Murray, 2002). The multimodal integration, gives a coherent representation of objects and appoint them meaning in the matter of perceptual experiences through including all senses. Ulrich (1993) demonstrated that our environmental perception is a multi-sensory rather than being restricted to the visual sense (Ulrich, 1993). As a matter of fact, visual and non-visual stimuli of the landscape are always interconnected in a complex way and require more consideration in all aspects.
Ultimately, Cats-Baril and Gibson (1987) described that aesthetic experience and preferences in any particular context can be affected by sensory experience (Cats- Baril & Gibson, 1987). Also it has been stated that content and space dependent functions such as spatial ability and orientation, can cause by different sensory stimuli and their convergence (Carles, Barrio, & de Lucio, 1999).

Urban Pocket Parks
Interaction with nature is assumed to be one of the vital needs of mankind and bene ts of urban green

Research Objectives
Since the evaluation of landscape aesthetic preferences relies on multiple factors such as emotions, values, cultural aspects, demographic characteristics, multi-sensory perceptions and etc., it is considered to be a complicated task, yet, a comprehensive explanation of the landscape preferences according to these senses is not available. Therefore, evaluating the relationship between non-visual stimuli, which may have an impact on public visual preferences in UPP s, is the main objective of the current study.

Study area
Tabriz as the capital of East Azerbaijan Province, located in northwest of Iran, has the semi-arid climate and regular seasons. The availability and functionality of UPP s in Iran and particularly in Tabriz, is arising as a new space in the urban area. According to Table 1, Tabriz green space Per capita is 15.5 m2, which is less than international standards (40m2). Not having adequate green space, more consideration is needed with regard to the Tabriz green spaces development.

Material and Method
The current study was conducted on 16 UPPs in Tabriz, Iran. Total number of volunteered respondents to participate in this survey was 394. The devised survey questionnaire for the research contains two parts including a textual questionnaire and photograph surrogates of actual scenes and the selection of scenes was done according to the consult of a group of experts. Meanwhile, the items that were extracted from Shahhosseini's work regarding Sound, Smell, and Touch Attributes in UPPs (Habib Shahhosseini, Kamal, & Maulan, 2014), have been differentiated the non-visual stimuli. In addition, the extracted stimuli are classi ed in the regard of the senses studding in current research (Fig 1).
With utilization of con rmatory factor analysis with application of Structural Equation Modeling (SEM) in four different models, this study aimed to identify the relationship among latent and observed variables of visual and non-visual preferences. As a matter of fact, in this approach, the correlation between the dependent variable and independent variables (individually and their correlations with each other) were analyzed.
A priori assignment of interrelation between variables is needed in order to analyze data by SEM as a con rmatory approach. As stated by Byrne (2001) with availability of some awareness regarding theories or experimental research, the relationship among underlying factors and the observed measures could statistically test by using Con rmatory Factor Analysis (CFA) (Byrne, 2001). This method can be used inductively by identifying an equivalent model to assess the values of factors. Through the identi cation of an equivalent model to assess the values of factors, this approach could be used inductively. Moreover, providing predictive validation among factors, identifying the direct and indirect relations between the latent variables, and also quantities all explained and unexplained variables on the model, can be done by this technique of analysis (Blunch, 2012;Byrne, 2001). SEM in AMOS 18 program was conducted in order to explore probability of relationship between proposed variables and to analyze the goodness-of-t of alternative models considering non-visual preferences (exogenous variables) and visual preferences (endogenous variables). Afterward, the path analysis was computed for estimating the standard regression weight for each path. In fact, the standardized regression coe cient or path coe cient was used to measure the signi cant and effect of each factor. In the meantime, for testing and estimating relations of statistical data combination in a speci ed model, general rule of thumb regarding Schermelleh-Engel's model t evaluation (Schermelleh-Engel, Moosbrugger, & Müller, 2003) was applied. It is also necessary to state that all of the models were tasted in advance, in terms of their discriminant and convergent validity.

Results And Discussion
Uni cation of all senses could help achieving a vigorous nal structural model in terms of understanding the impact of non-visual stimuli on visual preferences. Nevertheless, 4 different singular models were used in this study in order to test the interrelatedness between all senses. These singular models explained the impact of Sound, Smell and Touch stimuli on visual preference at UPPs. However, it is important to mention that as coherence had high correlation with other variables, it was removed from the all-structural models. Additionally, after testing the rate of Cronbach's Alpha, the spiritual sound was removed from nal models (Habib Shahhosseini, 2014).

CFA Model on Sound Stimuli
The regression weight estimates for the connection between sound stimuli and visual preferences is illustrated in Table 2 and it was demonstrated by the singular model that only Human Sound (P value = 0.006) and Natural Sound (P value = 0.008) had signi cant correlation with visual preferences.
There are infrequent data in regard to the relationship between sound factors and visual preferences. However, it was noted in some of the available and the most relevant studies that different factors such as demographic, behavioral, and psychological characteristics, which can be considered from both social and physical aspects, have an effect on sound preferences, therefore, sound, mainly in urban areas, is considered to be a complex system (Cain, Jennings, & Poxon, 2013;Semidor, 2006). The quality of perceived sound and its connection with visual features, could help our conception of environmental valuation (Brown, Kang, & Gjestland, 2011). In a related study, Southworth (1970) indicated that human and natural sounds, as opposed to the noisy sound of certain vehicles, were appreciated more by people (Southworth, 1970).
Likewise, Southworth (1970) analyzed the sound as a non-visual stimulus in the nature, and demonstrated that the dynamic value of the landscape can be enhanced by the relationship between image and sound (Southworth, 1970). Also its worthy to note that the availability of green space and the sound of children playing inside that, would cause social interaction and probability of increase social ties (Chiesura, 2004). Natural sounds like bird singing, wind, and water through foliage appears to have important impact on human's visual preferences by evoking pleasant feelings (Gidlöf-Gunnarsson & Öhrström, 2007). Moreover, the study conducted by Clark and Stankey (1979) indicated that users' preferences ca be diminished by the non-mechanical and mechanical sounds (Clark & Stankey, 1979).

CFA Model on Smell Stimuli
With the intention of measuring the correlation between the smell stimuli and visual preferences, another singular model was conducted.
The outcome of the test revealed that while Human-body smell (P value = 0.000) and Natural smell (P value = 0.001) had signi cant correlation with visual preferences, environmental smell had negative correlation which were not considered signi cant (Table 3). Final model, with standardized estimates of RMSEA (.039), TLI (.910), GFI (.901), CFI (.918) and p=.000 seemed to be t. Additionally, R2 value for model was .16, which was close to sound's R value (Fig. 3).
The accessibility of information about impact of smell, particularly in the landscape environment, have not been considered as much as needed. In the study conducted by Todrank et al., (1995), which peoplerelated odors (e.g. shampoos, soaps, lotions, musty, sweaty) were used as a mean to analyze visual and olfactory conditions, results indicated that public preferences can be affected by odors (

CFA Model on Touch Stimuli
The impact of touch stimuli on visual preferences were tested by nal singular model. The current research took step in order to apply the need scale for touch, concerning man-made touch (Peck & Childers, 2003). An unexpectedly outcome demonstrated that all included unful lled variables within this model with consideration of achieving a t model in regards of standardized valuation of measurement criteria, revealed that public sense of touch in the small parks was not signi cant at 0.05. In the meantime, the relationship among selected variables through full edge structural model was not supported by the model with R2 = 0.0 (Fig. 4, Table 4).
As stated before, the information about senses in the landscape eld and its relationship with Touch stimuli are insu cient. In a study conducted by Whitaker et al., (2008) regarding to the role of vision and touch assessment in the perception of texture, demonstrated that either of the senses affect the texture's perception in an independent way, but a complementary manner (Whitaker, Simões-Franklin, & Newell, 2008). Information regarding objects can be classi ed referring to their material features (such as weight and texture) or/and geometric aspects (such as shape, size, and orientation). However, objects characteristics can be represented with the contribution of tactile and visual senses. On the other hand, Ayres and Robbins (2005) stated that individual's cognitive and perception processing system can be in uenced by the integration between touch sense and relative information from tactile sensation (Ayres & Robbins, 2005). After touching or seeing an object, the human body would permit the brain to encode the information in order to make conceptual decisions for recognition, action or aesthetic judgment, which can relate to its quality or attractiveness. According to Antrop (2000) and Lee et al., (2008) since common tendency to touch various objects can be in uenced by feeling dirt, cleanness has been determined as one of the landscape characteristic (Antrop, 2000; Lee, Ellis, Kweon, & Hong, 2008). In other word, human assessment and decision making processes could be affected by cleanness as a landscape value (Grimm, Grove, Pickett, & Redman, 2000), and in this regard, the lack of public interest in terms of sense of touch in the UPPs could be logical.

The Final Structural model of the Study
The correlation between public's visual preferences and the non-visual factors (olfactory, auditory, and tactile) in the UPPs was examined through the nal structural model of the study.
The results demonstrated that out of six models' factor loadings, the Human-Body sound-smell stimuli (SSS) (standardized coe cient = .39) with the P value of .013 and the Natural SSS with the P value of .012 (standardized coe cient = .30), had positive signi cant correlations with the public's visual preferences. Meantime, the revised structural model did not support the relationship between the public's visual preferences with the Environmental Sound-Smell stimuli (standardized coe cient = -.05), the Instrumental Sound (standardized coe cient = -.19), the Furniture Touch (standardized coe cient = -.07) and the Natural Touch (standardized coe cient = -.01). Unexpectedly, all the mentioned latent variables, which did not receive any signi cant values, had a negative in uence on the visual preferences (Fig. 5, Table 5). In addition, the discriminate validities between all the six independent variables were valid.

Conclusion
The current study evaluated the relationship between non-visual stimuli and visual preferences through applying a SEM to the public users of UPP s. From the public's point of views, it is conducted that Natural sound and smell, Human-body smell, and Human sound have an important effect on visual preferences, but the impact of touch stimulus on the UPP did not consider signi cant. The outcomes provide fundamental information for policy makers to increase their attention in terms of developing all kind of stimulations. Ultimately, through giving more consideration to multisensory concept and its application on the UPPs, the result could be involved throughout the design as well. Findings of this research can help designers to enhance public visitation and attract their attention to develop successful UPPs by using relevant and appropriate combination of design features in these environments. Accordingly, attention to UPPs as part of the urban space initiative could be improved to positively contribute to national commemoration, neighborhood place making, as well as revitalization efforts, and enhance the ecological bene ts through sustainable practices. The impact of all senses' combination on visual preferences will be discussed by same authors on their upcoming research. Usage of sound levels, sound sources, and smell intensity in UPPs regarding to visual preferences would be suggested for the future studies.
The current study's Final structural model, which examined the correlation between visual preferences and the non-visual factors in the UPPs, has revealed the importance of two new-extracted constructs (the Natural SSS and the Human-body SSS). These new extracted constructs had signi cant values towards the visual landscape preferences in the UPPs. Nonetheless, the other tested constructs namely the Touch factors (Natural and Furniture Touch) and the Environmental SSS as well as the Instrumental sound did not signi cantly in uence the public's visual landscape preferences.

Declarations
Ethics approval and consent to participate Informed consent was obtained from all individual participants included in the study.

Consent for publication
The participant has consented to the submission of the case report to the journal.

Availability of data and materials
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declared no Competing interests Funding This research did not receive any speci c grant from funding agencies in the public, commercial, or notfor-pro t sectors

Declaration of con icting interests
The authors declared no potential con icts of interest with respect to the research, authorship, and/or publication of this article.

Authors' contributions
The authors contributed equally.  Classi cation of Variables The Final Structural Model