MASTER
Fascinated by Façades
Exploring the role of naturalness and complexity in façade geometry on fascination and perceived restoration
van Rietschoten, A.I.
Award date:
2021
Link to publication
Disclaimer
This document contains a student thesis (bachelor's or master's), as authored by a student at Eindhoven University of Technology. Student theses are made available in the TU/e repository upon obtaining the required degree. The grade received is not published on the document as presented in the repository. The required complexity or quality of research of student theses may vary by program, and the required minimum study period may vary in duration.
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
• Users may download and print one copy of any publication from the public portal for the purpose of private study or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
Exploring the role of naturalness and complexity in façade geometry on fascination and perceived restoration
By A.I. (Anke) van Rietschoten 0901166
In partial fulfillment of the requirements for the degree of Master of Science
in Human-Technology Interaction
Supervisors:
Dr. K. (Kynthia) Chamilothori Prof. Y.A.W. (Yvonne) de Kort
2 Stress is prevalent in everyday life. Nature has previously shown to be a source of fascination and stress restoration. However, it is still unclear what visual characteristics are responsible for this restorative effect. Nowadays, people spend much of their time indoors and therefore an attempt was made to recreate this restorative experience indoor with the use of the visual characteristics naturalness and complexity. Façade geometry together with the corresponding light patterns has been shown to influence people’s perception of a space, and can therefore potentially be used as a means to create restorative interiors. Sixteen black-and-white patterns were created by manipulating the visual characteristics, complexity and naturalness independently. These patterns were used as façade geometry variations in a simulated office environment. Participants gave their impression of the daylit scenes in a within-subjects online experiment. The participants saw all sixteen façade variations twice and answered questions about perceived restoration and fascination in the first round, and questions about perceived naturalness, perceived complexity and pleasantness in a second round. The results showed that perceived naturalness was affected by both manipulated complexity and naturalness. Manipulated complexity had a significant and positive effect on fascination and a significant but negative effect on perceived restoration. Contrary to the expectations, manipulated naturalness in the façade design did not affect fasciation or perceived restoration. These findings show the importance of visual complexity in creating nature-like stimuli.
As well as the potential of façade design as a means to create restorative interiors.
Keywords: Façade geometry, Naturalness, Complexity, Fascination, Perceived Restoration
3
Preface
I choose to start the Bachelor of Psychology and Technology at the Eindhoven University of Technology for its perfect combination of the human and the technology, something that fits very well with me and I never doubted about proceeding to the master Human-Technology Interaction.
Now, after 6.5 years at the TU/e, it comes to an end with this thesis, the final product of my graduation project for the master program Human-Technology Interaction. I executed a large part of this project from the dinner table in my tiny studio apartment, due to lockdown caused by the coronavirus epidemic. Covid-19 brought new challenges for everyone. Therefore, I would like to thank my supervisors and friends, and family for all their support.
In particular, I would like to thank my supervisors Kynthia Chamilothori and Yvonne de Kort for their support in the many Teams meetings we had during the last couple of months. Even though I have seen you a maximum of three times in person during my graduation project I always had the feeling I could reach out if I needed a helping hand. Secondly, I would like to thank my friends and fellow graduate students who gave me the motivation to get to work every day especially when we were all working from home. In particular, the coffee breaks which took most of the time way too long, both in Atlas and online, gave a nice break during the day. Finally, many thanks to my family and boyfriend who were always there when I needed them.
Have fun reading!
Anke van Rietschoten
4
1. Introduction ... 6
2. Theoretical Background ... 8
2.1 The effects of stress ... 8
2.2 Restoration by nature ... 8
2.3 Restoration without nature ... 10
2.4 Visual characteristics of restorative environments ... 11
2.4.1 Naturalness ... 12
2.4.2 Complexity ... 12
2.5 Evaluation of indoor environments ... 13
2.6 Research aims, questions and hypotheses ... 14
3. Method ... 17
3.1 Pilot 1: Manipulating naturalness and complexity ... 17
3.1.1 Design ... 17
3.1.2 Stimuli ... 17
3.1.3 Measurements, procedure and participants ... 20
3.1.4 Results ... 21
3.1.5 Conclusions and implications ... 22
3.2 Pilot 2: Increasing naturalness ... 22
3.2.1 Design... 22
3.2.2 Stimuli ... 22
3.2.3 Measurements, procedure and participants ... 26
3.2.4 Results ... 27
3.2.5 Conclusions and implications ... 29
3.3 Final selection of experimental stimuli ... 29
3.4 Experimental study ... 31
3.4.1 Design... 31
3.4.2 Participants ... 31
3.4.3 Stimuli ... 32
3.4.4 Measurements ... 36
3.4.5 Procedure ... 37
3.4.6 Statistical Analyses ... 37
5
4. Results ... 39
4.1 Effects of manipulated naturalness and complexity on perceived naturalness, perceived complexity, and pleasantness ... 39
4.1.1 Perceived naturalness ... 39
4.1.2 Perceived complexity ... 41
4.1.3 Pleasantness ... 42
4.2 Predicting fascination ... 43
4.2.1 Effect of manipulated naturalness and complexity on fascination ... 43
4.2.2 Effect of pleasantness, perceived naturalness, and perceived complexity on fascination 45 4.2.3 Effect of the manipulation, pleasantness, perceived naturalness, and perceived complexity on fascination ... 46
4.3 Predicting perceived restoration ... 47
4.3.1 Effect of manipulated naturalness and complexity on perceived restoration... 47
4.3.2 Effect of pleasantness, perceived naturalness, and complexity on perceived restoration 49 4.3.3 Effect of pleasantness, perceived naturalness, perceived complexity, and fascination on perceived restoration ... 49
4.3.4 Predicting perceived restoration ... 50
5. Discussion... 51
5.1 Findings of the current study ... 51
5.2 Limitations and future research ... 54
5.3 Societal relevance ... 55
5.4 Conclusion... 56
6. References ... 58
7. Appendices ... 63
7.1 Additional variations of patterns ... 63
7.1.1 Increasing Naturalness with line width differences across lines ... 63
7.1.2 Increasing Naturalness with line width differences within lines ... 67
7.1.3 Corrected version for magnifying ... 69
7.2 Informed consent form ... 70
6 Stress is prevalent in everyday life. In the Netherlands, almost half of the reported work- related illness is psychological, with 75% of such illness being related to stress and burn-out (NCvB, 2018). Exposure to natural environments has widely been acknowledged to positively influence the mental well-being of people (Berto, 2014). In the field of environmental psychology, there are two influential theories explaining stress restoration from nature which are Psycho-Evolutionary Theory (PET) by Ulrich (1983) and Attention Restoration Theory (ART) by Kaplan (1995). PET explains restoration by positive affect that is elicited immediately when exposed to natural elements in a scene (Ulrich, 1983). In contrast, ART focuses on cognition and explains restoration by the depletion of the resource of directed attention which causes fatigue (Kaplan, 1995). For an environment to be restorative it should have extent, should be compatible with the wishes and needs of a person and should give the feeling of being away. Lastly, it should be fascinating, it should hold attention but without mental effort. Natural scenes are particularly suitable for this type of fascination and therefore to replenish directed attention fatigue (Kaplan, 1995).
Since nature is not always readily available, it is interesting to examine if non-natural environments can be restorative as well. There is some evidence that, for instance, rich ornamentation in architecture is a way to make an urban environment restorative (van den Berg, Joye, & Koole, 2016). Ommeren (2019) and Oosterhaven (2017) took an extra step and successfully investigated whether dynamic lighting variations could be created inducing a restorative experience in an indoor environment. These findings of stress restoration through non-natural stimuli suggest that certain visual characteristics of a scene, which may be present in natural and non-natural scenes, play a role in stress restoration. While there is substantial evidence for restoration from stress in natural scenes, it is unclear what visual characteristics of these scenes are responsible for this effect (Joye & Dewitte, 2018).
Since Ulrich (1983) focuses on natural elements in a scene and Kaplan indicates that nature is particularly suitable for restoration it seems likely that the naturalness of a stimulus plays an important role in stress restoration. Another visual characteristic that is prominent in the stress restoration literature is complexity. Both Ulrich (1983) and Kaplan & Kaplan (1989) mention complexity as a predictor for the preference of a scene. Han (2010) found a significant correlation between preference and restoration. Therefore, complexity as a visual characteristic may also have restorative qualities. Additionally, the combination of naturalness and complexity is very interesting since these concepts are very much intertwined since nature scenes generally are rich in details and are therefore intrinsically more complex than urban scenes (van den Berg et al., 2016). However, the question arises whether one of the two or both attributes are responsible for the restorative experience.
Identifying these restoration-inducing visual characteristics of nature is necessary to create a restorative experience without actual nature. In this thesis, the visual characteristics of naturalness and complexity are implemented in façade geometry to create a restorative experience indoors without direct access to nature. In previous studies, Abboushi et al. (2019) manipulated the façade geometry to create daylight patterns and found that there is a higher preference and reported visual interest for natural (fractal) shaped patterns. Similarly, Chamilothori et al. (2019) found a significant
7 influence of façade composition (varying the shape and distribution of façade openings in virtual reality) on how exciting and how interesting a daylit space was perceived, as well as on the heart rate of participants. These results show that façade geometry is an instrument to influence a person’s physiological responses and perception of a space. A next step would be to investigate whether it is also possible to create a restorative experience using façade geometry.
This thesis combines knowledge about façade geometry and restoration from nature to create restorative interiors. The aim of this thesis is twofold: (1) differentiate complexity and naturalness in a stimulus from each other and to investigate the individual and interaction effect on restoration and (2) to examine the potential of façade geometry and their light patterns as a means to create restoration-inducing interiors.
8 2.1 The effects of stress
The feeling of stress is very well known in everyday life. Prolonged feelings of stress can have severe consequences. The human’s stress response can be physiological but also psychological.
Examples of physiological responses to stress are increased heart rate or skin conductance (Lin, Lin, Lin, & Huang, 2011). Psychological responses can be a cognitive response, such as poorer performance (Glaser, Tatum, Nebeker, Sorenson, & Aiello, 1999), but can also be a change in affect like feeling sad or anxious (Lazarus & Cohen, 1977). In the long term, stress can lead to serious complications such as sleeping problems and chronic fatigue, resulting in more severe problems like depression which can lead to hospitalization (Danielsson et al., 2012; Salvagioni et al., 2017). Physical complications relating to stress include headaches, coronary heart disease, or type 2 diabetes (Salvagioni et al., 2017).
Stress in the work setting can lead to a lack of motivation and disinterest which are important causes for job dissatisfaction and absenteeism, and in worse cases, it leads to burn-out (Salvagioni et al., 2017). In the Netherlands, almost half of the reported work-related illness is psychological, with 75% of such illness being related to stress and burn-out (NCvB, 2018). These numbers show that there is a need for ways to reduce stress in the work environment. Common activities that humans engage in to restore from stress are doing sport, listening to music, walking in the forest, sleeping, or reading a book (Hansmann, Hug, & Seeland, 2007). Other ways of relaxation such as mindfulness and meditation became more popular over the years (NCvB, 2018). However, these types of activities are not fitting for stress restoration in a work setting. Therefore, it would be ideal to restore from stress in other ways, such as with the use of the physical office environment.
2.2 Restoration by nature
In the field of environmental psychology substantial research has been done to examine the restorative effect of nature. There are two dominant theories about the restorative effect of nature.
On one hand, there is the Psycho-Evolutionary Theory (PET; Ulrich, 1983) and on the other hand, Attention Restoration Theory (ART; Kaplan & Kaplan, 1989; S. Kaplan, 1995).
Ulrich (1983) explained the restorative effect of nature by positive affect. Stressed people felt better by viewing nature scenes compared to urban scenes. According to Ulrich (1983), this affective response is rapid and unconscious and is elicited immediately after seeing the scene and before any cognitive judgment is possible. This phenomenon is called preferenda (Ulrich, 1983). This rapid response is adapted from evolution because natural environments that are important for survival (e.g. containing food, water, or refuge) are preferred. This affinity with nature is also called biophilia (Ulrich, 1993).
In contrast to PET, Kaplan’s (1995) ART focuses on cognition instead of affect. Directed attention is the key component in ART. Directed attention is a form of attention that costs effort and therefore is susceptible to fatigue and is used for tasks that demand concentration. Directed attention is an important resource needed for human effectiveness. Being fatigued and unable to concentrate, even on pleasant tasks, are characteristics of directed attention fatigue. The recovery
9 of directed attention fatigue is called restoration or restorative experience. According to ART, there are four components necessary in an environment for a restorative experience: soft fascination, being away, extent, and compatibility (Kaplan, 1995). Soft fascination (sometimes called involuntary attention) is the type of attention that does not cost any mental effort and therefore it is not susceptible to fatigue like directed attention and is, therefore, an important factor in recovering from directed attention fatigue. This is a state of attention in which there is the opportunity to reflect and it holds the attention in an undramatic fashion. Natural environments accommodate this type of fascination (Kaplan & Kaplan, 1989; Kaplan, 1995).
Being away in this context is meant conceptual, not physical. Physically moving to another place can help but is not necessary to give the feeling of being away (Kaplan, 1995). Compatibility in ART means the compatibility between the user of the environment and the environment, the environment should be able to facilitate the goals and needs of the user to be restorative (Kaplan, 1995). Lastly, an environment can only be restorative if it has enough extent. There should be enough content in a scene to be called an environment. This does not necessarily mean it should be a large space since small spaces can give the feeling that they are a whole other world as well (Kaplan, 1995).
Nature is an example of an environment where all these four components (soft fascination, being away, extent, and compatibility) can be found. Nevertheless, indoor environments that fulfill these components are also suggested as successful restorative environments (Kaplan, Bardwell, & Slakter, 1993).
Kaplan (1995) and Ulrich (1983) have different viewpoints on why nature is restorative, but the main point that is agreed upon is that nature is a source for restoration (Kaplan, 1995; Ulrich, 1983). Evidence for the restorative potential of nature is found both through self-report (Hartig, Korpela, Evans, & Gärling, 1997; Herzog, Maguire, & Nebel, 2003) and physiological measurements (Berto, 2014). The restorative potential of nature has been demonstrated in a laboratory setting (e.g.
Herzog et al., 2003), and also in field experiments (Hartig, Evans, Jamner, Davis, & Gärling, 2003;
Tyrväinen et al., 2014). However, when nature is not readily available, representations of nature, for example through the presence of plants in an environment (Korpela, De Bloom, Sianoja, Pasanen, &
Kinnunen, 2017) or the use of projections (de Kort, Meijnders, Sponselee, & IJsselsteijn, 2006) or pictures (e.g. Herzog et al., 2003; Twedt et al., 2019) of natural environments, can also be a means for stress restoration. This gives reason to believe that real nature is not necessary to have a restorative experience.
Both theories contain unanswered questions and uncertainties about restoration. A recurring finding in the literature is that vegetation rich scenes are particularly effective for restoration (Joye &
Van Den Berg, 2011). However, it would be illogical in the viewpoint of PET that all vegetation rich scenes are a source of restoration since not all vegetation is as suitable for refuge or food source (Joye
& Van Den Berg, 2011). Likewise, Kaplan (1995) mentions caves, fire and wildlife as examples of restorative environments, but these types of environments are generally not included in stress restoration studies (Joye, 2007; Joye & Dewitte, 2018). Moreover, the concept of soft fascination in ART is vague and ill-defined (Joye & Dewitte, 2018). It is unclear how the range from soft to hard fascination is organized and how ‘soft’ a stimulus should be to be a mediator for restoration (Joye &
Dewitte, 2018). The lack of quantification of the term fascination has implications for research in this
10 of soft fascination explains that fascination can be anything that holds attention effortlessly it is unclear what makes a stimulus fascinating, and what visual characteristics are responsible for this (Joye & Dewitte, 2018; Valtchanov & Ellard, 2015).
2.3 Restoration without nature
In the previous paragraphs, the restorative quality of nature was discussed. However, there is a discrepancy between the environments that humans evolved in and adapted to and the modern urban environments that people are living in nowadays (Joye, 2007). A few studies show evidence that restoration is also possible in environments without nature. Examples of this are the study of Ouellette et al. (2005) where they found evidence for a restorative effect in a monastery and the study of Kaplan et al. (1993) where the restorative potential of museums was explored.
One way to tackle this discrepancy is with biophilic or nature-based architecture, which employs forms that mimic nature. This approach is suggested to improve the attractiveness of a scene and make it fascinating (Joye, 2007). A way to imitate nature in architecture is placing nature- like objects and details in or on buildings, in the form of ornaments. Buildings that are rich in detail have been shown to be fascinating and to have restorative potential (van den Berg et al., 2016).
Because people spend much of their time indoors (Evans & McCoy, 1998; Klepeis et al., 2001) it would be ideal to create interiors with restorative potential. Since nature is not immediately available in indoor spaces other ways to create restorative indoor environments have to be explored.
Ommeren (2019) and Oosterhaven (2017) successfully used dynamic lighting variations varying in complexity to create a restorative experience in an indoor environment, which show that similar visual characteristics (e.g. complexity) of light patterns are a promising means for restoration from stress, as will be discussed further in section 2.4.2 (Ommeren, 2019; Oosterhaven, 2017).
Another option for creating restorative interiors could be with the use of façade geometry.
While no studies have directly examined the restorative potential of façade geometry, current evidence suggests that the façade composition and the corresponding daylight patterns can be a source of visual interest and potentially induce fascination. Comparing variations of façade and daylight patterns, Abboushi et al. (2019) found that higher preference and reported higher visual interest for the natural (fractal) shaped patterns than horizontal stripes. Additionally, visual interest was higher for the more complex fractal patterns. Similarly, Chamilothori et al. (2019) examined the joint effect of façade geometry and daylight pattern on the appraisal of the scene in virtual reality.
The façade and daylight patterns were shown to influence the heart rate of the participants, as well as how exciting, interesting and pleasant the scene was perceived. In particular, irregular façade geometries were perceived as more exciting, more interesting, and more pleasant than regular patterns (Chamilothori et al., 2019), indicating that the spatial distribution it the patterns could potentially be a source of fascination. Facade geometry has been shown to affect both the subjective and physiological responses of participants. However, there is no knowledge yet about inducing fascination through façade geometry, or about the restorative visual characteristics of nature (e.g.
complexity) that could be reproduced in a façade design.
11 2.4 Visual characteristics of restorative environments
In order to design facades that can have a restorative effect, it is important to pinpoint which visual characteristics in nature are responsible for stress restoration. In the literature different aspects of natural and urban environments are examined for their restorative potential such as aesthetics, preference, visual interest, visual appeal, pleasantness, naturalness, openness, water features, presence of people or complexity (Abboushi et al., 2019; Han, 2010; Herzog et al., 2003; Marin &
Leder, 2013; van den Berg et al., 2016). However, there is no conclusive answer on which exact visual characteristics are responsible for fascination or restoration.
Both Ulrich (1983) and Kaplan (1995) agree that nature has certain characteristics which result in the scene being restorative. Ulrich (1983) mentions complexity, focality, depth, ground surface texture, the absence of threat, deflected vistas and water features as important characteristics of nature that are important for the preference of environments. Kaplan & Kaplan (1989) created, besides ART, a framework that describes the visual characteristics of a scene that are important for preference (Table 1). This framework is based on two important human needs, understanding and exploring, and the degree of inference required in these processes. The degree of inference concerns how readily available the information in a scene is: is all information immediately visible or does it require some more interpretation. These two dimensions result in four combinations: coherence, complexity, legibility and mystery. Coherence in a scene means that information is readily available and the scene is orderly and comprehensive. Second, complexity means that information is readily available in the scene but some exploration is required. Third, legibility in a scene means that an environment is easy to grasp and remember because it is comprehensible. However, not all information is immediately visible but it is easy to understand what would be in the hidden parts. Lastly, mystery in a scene has information or parts that are not immediately visible, and therefore exploration is necessary to figure out what it is exactly. These types of scenes draw attention and curiosity. (Kaplan & Kaplan, 1989)
Table 1. Preference matrix
Understanding Exploration
Immediate Coherence Complexity
Inferred, predicted Legibility Mystery
In environmental psychology, preference and stress restoration are usually investigated independently. However, preference and restoration are found to be highly correlated (Han, 2010;
Herzog et al., 2003; Kaplan et al., 1972; Purcell et al., 2001) and therefore visual characteristics that are found to be important for preference are possibly also effective for stress restoration. One example of this is naturalness. Natural scenes are found to be more preferred than urban scenes (van den Berg, Koole, & van der Wulp, 2003).
Nature is the central component of both ART and PET and therefore naturalness is an important characteristic to include for designing restorative interiors. Naturalness and complexity are two attributes that are intertwined in the literature; research in the field of stress restoration often employs vegetation rich scenes versus urban scenes, while these vegetation rich scenes are fairly complex compared to the urban scenes (van den Berg et al., 2016). However, it is unclear if either the
12 effect is a result of the combination of the two attributes. It is thus important to also examine the restorative potential of complexity, and attempt to differentiate this from the attribute of naturalness. In the following paragraphs both naturalness and complexity, as well as the relationship to restoration will be discussed further.
2.4.1 Naturalness
Naturalness is ill-defined in the literature. In most cases, only a distinction is made between natural or urban scenes (e.g. Van den Berg et al., 2016). In this thesis, naturalness will be approached as a continuous phenomenon rather than a dichotomous one. To this end, naturalness in this thesis will be used as how much something reminds of nature. Nature can be man-made (Ulrich, 1983) which is often the case in urban areas. Literature shows that natural scenes are more preferred (van den Berg et al., 2003) and fascination is higher for natural scenes compared to urban scenes (Berto, Massaccesi, & Pasini, 2008). Berto et al. (2010) showed that the naturalness of an image mediated fascination ratings, meaning that natural images scored higher on fascination compared to images of urban scenes.
The geometry in nature has been described as fractal geometry by Mandelbrot & Wheeler (1983). Fractals are self-similar, the fractal dimension stays the same no matter to what size the image is scaled (Mandelbrot & Wheeler, 1983). Fractal dimension is one way to quantify naturalness.
It has been suggested as a predictor of naturalness by Hagerhall (2005) because of its prevalence in nature (e.g. Abboushi et al., 2019; Hagerhall et al., 2008; Hagerhall, Purcell, & Taylor, 2004; Joye, 2007; Purcell et al., 2001; Taylor, 2006). Additionally, the fractal dimension is frequently used to describe the complexity of the fractal pattern, as is further described in section 2.4.2.
In other attempts to quantify naturalness in images, Berman et al. (2014) identified low-level features (e.g. edges, number of straight lines) in an image that can predict how natural the image is perceived. These low-level features of an image are responsible for a bottom-up fast judgment of perceived naturalness (Berman et al., 2014). In contrast, high-level features are features that carry semantic information, identifiable shapes like buildings or trees (Ibarra et al., 2017). The edge density, number of straight edges and the standard deviation of hue resulted in suitable low-level features to predict perceived naturalness (Berman et al., 2014). Natural images are generally more chaotic and have a higher number of non-straight edges compared to urban scenes which have a more organized structure (Kardan et al., 2015). Above mentioned low-level features indicating naturalness were able to predict the aesthetic preference of a scene (Kardan et al., 2015). Additionally, Ulrich (1983) explained the restorative effect as a fast pre-cognitive process, and therefore low-level image features seem a more promising indicator of naturalness than high-level features.
2.4.2 Complexity
The visual attribute of complexity is mentioned in both SRT by Ulrich (1983) and in the preference matrix of Kaplan and Kaplan (1989). Kaplan & Kaplan (1989) defined complexity by the number of elements in a scene; the richness of a scene. Berlyne, Ogilvie, & Parham, (1968) and Nadal, Munar, Marty, & Cela-Conde (2010) supported the definition stated by Kaplan & Kaplan (1989) by showing evidence that complexity was highly driven by the number of elements in the scene.
However, at the same time suggested that visual complexity concerns not only the number of objects
13 present but also the structure of these objects (Nadal et al., 2010). Berlyne (1963) showed there are many more elements that contribute to visual complexity, for example, arrangement, congruity, symmetry and distribution (Berlyne, 1963; Berlyne et al., 1968). The definition of complexity is not consistent through the literature or across different domains, which makes it difficult to compare the results (Marin & Leder, 2013; Nadal et al., 2010). Nadal et al. (2010) identified three different forms of complexity that influence the perception in different ways which fit with different definitions used in previous research. These three forms of complexity correspond to, the number and variety of the elements, the distribution or organization of the elements, and lastly the symmetry in the distribution of elements (Nadal et al., 2010).
The relationship between complexity and preference, and between complexity and restoration is highly investigated in the literature of environmental psychology but also in other disciplines (Marin & Leder, 2013). The evidence concerning the nature of the relationship between complexity and fascination, and between complexity and preference varies. There is evidence for an inverted U-shape relationship, meaning that especially medium complex scenes are preferred over very low or very high complex scenes, as suggested by Berlyne (1963). Abboushi et al. (2019) showed that especially moderate complex fractal patterns, measured in fractal dimension, were visually more interesting than less or more complex patterns. Similarly, Taylor et al. (2005) found evidence that preference is higher for fractals with a moderate fractal dimension. Additionally, with the use of the dynamic lighting variations, Ommeren (2019) and Oosterhaven (2017) found that especially medium complex scenarios were perceived as more fascinating (Ommeren, 2019; Oosterhaven, 2017). In contrast, Kaplan et al. (1972) found a positive linear relationship between complexity and preference.
However, this relationship was only linear when nature and urban cases were processed separately.
One possible reason why the evidence for the nature of the relationship is conflicting is because of the inconsistency in defining, and therefore in manipulating, complexity (Nadal et al., 2010).
Secondly, a linear relationship also could mean that not a full range of complexity was used. If only low and medium complexity scenarios are used, the outcome could be a linear relationship (Lindal &
Hartig, 2013).
As previously mentioned, the fractal dimension is used as a measure to quantify visual complexity in a fractal pattern. However, quantifying visual complexity is also explored in a separate domain using subjective judgments of complexity to verify objective measures of visual complexity.
Just as with the quantification of naturalness, both low-level features and high-level features are considered for quantifying complexity (Corchs, Ciocca, Bricolo, & Gasparini, 2016). With the use of image-processing techniques, Corchs et al. (2016) found that visual complexity dependent mostly on the number of objects, the amount of detail and the number of colors (Corchs et al., 2016). Another frequently used metric to predict visual complexity is edge detection (Machado et al., 2015; Mario, Chacón, Alma, & Corral, 2005). The number of edges is a measure for the number of elements in an image (Machado et al., 2015), which can be used to examine visual complexity as Kaplan (1995) defined it.
2.5 Evaluation of indoor environments
Conducting experiments in real-world environments requires many resources in terms of space and manpower and is therefore expensive (Chamilothori et al., 2019). An alternative to this
14 simulation in comparison to a real environment is that the conditions are the same for all experimental trials. Effects of seasons, day, time or weather can be avoided while in real environments such control is not possible.
The use of rendered images is previously shown as a tool to evaluate light conditions in an office environment (Newsham, Richardson, Blanchet, & Veitch, 2005). However, other subjective evaluations of environments, such as pleasantness or complexity, can be reproduced in a rendered image (Mahdavi & Eissa, 2002). More immersive virtual environments would increase the presence of participants in the simulated environments (de Kort et al., 2006). However, an increase in presence in a simulated environment affected the physiological response but not on self-reported affect (de Kort et al., 2006), whereas this thesis is based on subjective evaluations of perceived restorative potential and fascination. These subjective measures for fascination and perceived restoration are created and used previously in settings where participants rated images of natural and urban environments with positive results (Berto, 2014; Hartig et al., 1997), and therefore likely to be valid in a rendered image.
2.6 Research aims, questions and hypotheses
The literature shows that it is possible to alter the perception of an interior space, as well as the physiological responses of participants, by manipulating the façade geometry and creating daylight patterns. Yet, it is unknown if the façade geometry can also be used to create restorative interiors. To address this knowledge gap, this thesis aims to investigate the effect of different façade geometries on fascination and perceived restoration. As there is no definite answer on which visual characteristics of a natural scene are responsible for fascination and perceived restoration, this thesis will primarily focus on the visual characteristics of naturalness and complexity and how to manipulate naturalness and complexity independently to investigate the effect these concepts have on fascination and perceived restoration. Following this approach the following research questions are defined:
Q1. Is it possible to manipulate independently the naturalness and the complexity of façade geometry?
Q2. What is the effect of naturalness and of complexity in façade geometry on fascination?
Q3. What is the effect of naturalness and of complexity in façade geometry on perceived restoration?
Q4. Does fascination mediate the relationship between façade geometry and perceived restoration?
Q5. Does pleasantness mediate the relationship between façade geometry and fascination and perceived restoration?
By carefully manipulating complexity and naturalness it is expected that both concepts can be manipulated independently. However, based on the idea that natural environments are relatively complex compared to urban views (van den Berg et al., 2016) it can be hypothesized that for a scene to be perceived as natural, the scene needs to hold a certain level of complexity. As a result, the following hypotheses are considered: H1a. Only manipulated complexity, and not manipulated
15 naturalness, will influence perceived complexity; H1b. Both manipulated complexity and manipulated naturalness will influence perceived naturalness
Façade geometry has been shown to affect both the subjective and physiological responses of participants therefore it is suspected that both fascination and perceived restoration can be induced by façade geometry. Based on the knowledge of ART, that nature is a source for fascination (Kaplan, 1995), the second hypothesis is: H2. Manipulated naturalness in a façade design will have a positive effect on fascination.
The visual characteristic of complexity is expected to have an inverted U-shape effect on fascination, meaning that in particular, façades with moderate complexity will affect fascination, similar to the results found by Ommeren (2019). Therefore the third hypothesis is: H3. Manipulated complexity in a façade design will have an inverted U-shape effect on fascination
Similar to the hypotheses for the effect of naturalness for fascination, it is expected that naturalness will have a positive effect on perceived restoration. Based on the knowledge of PET and ART that nature is the source of stress restoration (Ulrich, 1983; Kaplan, 1995). For the visual characteristic of complexity again a similar hypothesis was made as for fascination. As a result, the following hypotheses for RQ 3 were defined: H4. Manipulated naturalness in a façade design will have a positive effect on perceived restoration; H5. Manipulated complexity in a façade design will have an inverted U-shape effect on perceived restoration
Secondary objectives in this thesis are establishing whether fascination mediates the relationship between naturalness and complexity in a façade design and perceived restoration.
Based on Kaplan’s ART (1995) which states that fascination is one of the four components needed for a restorative experience, the sixth hypothesis is: H6. Fascination mediates the relationship of naturalness and complexity in a façade design on perceived restoration.
Lastly, a mediation effect of pleasantness, perceived naturalness and perceived complexity is explored when a significant effect is found on the effect of naturalness and complexity in a façade on fascination and perceived restoration. Pleasantness has previously been shown to be affected by façade geometry and therefore it is explored if pleasantness can explain part of the effect of naturalness or complexity on fascination and perceived restoration. The last hypothesis that is considered is: H7. Pleasantness mediates (when a significant relationship is found) the relationship between manipulated naturalness and manipulated complexity on fascination and perceived restoration. All hypotheses and additional exploratory analyses are illustrated in a conceptual model, shown in Figure 1.
16
Figure 1. The conceptual model with hypotheses
17
3. Method
This part of the thesis describes how the defined research questions were answered. In multiple iterations, stimuli with an independent manipulation of naturalness and complexity were created which were subsequently used as façade geometry in the experimental study. Two pilot studies were conducted before the final experimental stimuli were selected. The first pilot study describes how this manipulation was initiated. Based on the results, a second pilot study was conducted to improve the manipulation of naturalness. These pilot studies are described in 3.1 and 3.2 respectively. Subsequently, the final selection of the façade designs took place, as described in section 3.3, and these were placed in a simulated office environment described in the final sections of the method; the experimental study (section 3.4). Which also includes the set-up, participants, and statistical analyses for the experimental study.
3.1 Pilot 1: Manipulating naturalness and complexity 3.1.1 Design
For an effective and independent manipulation of naturalness and complexity, a first pilot study was set-up. Sixteen black-and-white patterns varying in complexity and naturalness were compared in a 4 (naturalness ) by 4 (complexity) within-subjects experimental design. In a short online questionnaire in Limesurvey, the manipulation of naturalness and complexity was tested by perceived naturalness and perceived complexity, respectively.
3.1.2 Stimuli
The manipulation of naturalness and complexity in a façade was done by manipulating the spatial distribution of façade openings. Sixteen patterns varying in naturalness and complexity were created. The pattern used in this thesis was based on a real-world example of metal cladding used on an office façade in France (France Resille, n.d.), shown in Figure 2. The pattern was reconstructed in Adobe Illustrator 2018 CC and corrected for the perspective shown in Figure 3. This pattern was selected as being very high in complexity, because of the high number of elements (S. Kaplan, 1995), and low in naturalness, due to its straight lines (Berman et al., 2014; Kardan et al., 2015). A two- colored (black-white) pattern was chosen similar to the studies of Chamilothori et al. (2019) and Abboushi et al. (2019). To avoid the effect of light falling through the façade in the experimental study, the perforation ratio (= number of white pixels / total number of pixels) of the patterns was kept constant at 40% ± 2%. The number of 40% is similar to what Chamilothori et al. (2019) used. All images of patterns were created with the dimensions of 365 x 260 mm and were exported in png format with a high resolution of 300 ppi.
18
Figure 2. Metal cladding by France Resille (n.d.)
http://france-resille.com/en/facade-sunshade/ Figure 3. Base Pattern
The low-level features of an image were used to manipulate the naturalness and complexity in this pattern. In particular, edge density and the presence of (non) straight edges are taken into account, the use of color was not considered in this thesis. Berman et al. (2014) and Kardan et al.
(2015) suggested that the number of curves is important for naturalness. Therefore, the least natural pattern had only straight lines. The lines are gradually curved more to increase naturalness in the pattern, as is illustrated in Figure 4. Using the pattern with only straight lines as the starting point (N1), a new pattern was created by inserting one node in each line, dividing the line into two curved parts (N2). In order to keep the composition of the pattern as similar as possible, it was chosen to start with two curves instead of one. By adding one node, two curves were created that approximated the position of the straight line. The next level of naturalness (N3) was created by dividing the line into four parts by adding two additional nodes. The most natural variation (N4) is made by dividing the line into eight parts with the use of seven nodes. With this manipulation, the four patterns rated as ‘high’ in complexity were created which different in naturalness, which are shown in Figure 6.
N1 N2 N3 N4
Figure 4. Naturalness manipulation curves
The manipulation of complexity is based on the definition by Kaplan (1995) that complexity is about the number of elements in a scene. The image was magnified to reduce the number of elements resulting in a decrease in complexity. MATLAB 2018a was used to manipulate complexity.
19 The four highly complex (C4) patterns created by the naturalness manipulation were used as input, the complexity manipulation was done. With the use of the MATLAB Image Processing Toolbox, the optimal location to magnify the image was approximated by using the root mean square of the errors.
The size of the smaller image was determined at 25% of the previous complexity level. The error was calculated for all possible locations, by subtracting 40 (the preferred perforation) from the perforation ratio at this specific location. The errors of the same locations in the four images were combined to a single value by taking the Root Mean Square. The optimal location was the location with the smallest value. The optimized locations are presented in Figure 5. This process was repeated three times to create a 4 x 4 matrix which is shown in Figure 6. When the optimal locations to magnify the image were determined the final perforation ratio was calculated for all images by using the MATLAB Image Processing Toolbox (Table 2).
Figure 5. Zoom location Pilot 1
Naturalness
N1 N2 N3 N4
Complexity C1C2C3
20
C4
Figure 6. Pattern Matrix Pilot 1
Table 2. Perforation
N1 N2 N3 N4
C1 40% 40% 39% 38%
C2 40% 40% 40% 38%
C3 41% 41% 40% 39%
C4 38% 41% 39% 40%
To ensure that complexity in the different levels of naturalness is similar, the number of elements was measured by calculating the number of edges in the images. Following the process described by Machado et al. (2015), edge detection was conducted with MATLAB 2018a using the Canny method for edge detection (Canny, 1986). Edge density for all images in Figure 6 was calculated by dividing the number of pixels of the edges by the total number of pixels in the image.
Figure 7 shows that the four patterns representing one complexity level have similar edge densities, meaning that complexity is similar across naturalness levels.
Figure 7. Edge densities patterns for Pilot 1
3.1.3 Measurements, procedure and participants
The sixteen images presented in Figure 6 were tested in an online survey using Limesurvey.
Participants answered first demographic questions about their age, gender and country where they spent most of their lifetime before evaluating the presented stimuli. The patterns were presented in a random order to the participants to avoid bias. Participants saw one pattern at a time. The questions
21 for perceived naturalness and perceived naturalness were presented under the image: “How natural would you rate this pattern?” and “How complex would you rate this pattern?”. Participants answered on a 10-point Likert scale (1 = not at all complex/natural; 10 = very complex/natural). Backward navigation was not possible. Participants took around 5 minutes to complete the questionnaire.
Participants were recruited from the private network of the researcher and were asked to participate in person. In total fifteen participants took part in the survey of which seven were male (M=0.56, SD =0.51). Fourteen of the participants was between 23 and 28 years old, one of them was aged 55 (M=26.6; SD=7.69). All participants had lived the most time of their life in the Netherlands.
3.1.4 Results
The outcomes of the manipulation check for naturalness and complexity are illustrated in Figure 8 to Figure 11. Figure 8 presents perceived naturalness for the four levels of manipulated naturalness (N1 to N4); Figure 9 presents perceived naturalness per level of manipulated complexity (C1 to C4). Perceived naturalness increases from N1 to N4 but is not increasing linearly as expected.
Similarly, Figure 10 and 11 show the levels of perceived complexity for all four levels of manipulated complexity and manipulated naturalness respectively. The difference between N1 and N4 in perceived naturalness is relatively low compared to the differences shown for perceived complexity (Figure 10). Figure 9 shows that the naturalness levels are crossing over and are not horizontal, meaning that perceived naturalness is not similar across manipulated complexity and that manipulated complexity has an effect on perceived naturalness. However, there is a difference between N1 and N4, the lowest and highest level of manipulated naturalness. Manipulated complexity showed an increase in perceived complexity and seems to be quite linear. Figure 11 shows that the different levels of complexity were distinct and did not cross over the different levels of naturalness.
Figure 8. Perceived naturalness Figure 9. Perceived naturalness
22
Figure 10. Perceived complexity Figure 11. Perceived complexity
3.1.5 Conclusions and implications
The outcomes of Pilot 1 show that the manipulation of complexity was successful. The different levels were distinct and use a large part of the scale, from 3.5 to 8.3. In contrast, the manipulation of naturalness should be stronger to be suitable as an independent variable in the experimental study. The following sections introduce further work that was conducted to increase naturalness in the stimuli and present the outcomes of a second pilot study, Pilot 2, in which the manipulation was tested again.
3.2 Pilot 2: Increasing naturalness 3.2.1 Design
Pilot 2 was done to increase the naturalness manipulation in the stimuli. Two pattern matrices were tested in a 4 x 4 within-subjects experimental design. The two matrices were a between-subjects factor: participants only saw one of the two matrices. In an online study the patterns were randomly presented in two rounds: once to answer questions about perceived naturalness and once to answer questions about perceived complexity to avoid bias in the answers.
Which question was in round one was randomly divided.
3.2.2 Stimuli
A comment given by a participant in Pilot 1 was that none of the stimuli presented were perceived as natural. One possible reason for this could be that the patterns were too regular. The small effect of naturalness in the previous pilot could be seen as evidence for this. The manipulation for naturalness and complexity of Pilot 1 was kept as the starting point for pattern matrices used in Pilot 2. It was attempted to create more irregular patterns, which would be perceived as more natural, in two different ways: (1) by varying the line width across lines and (2) by varying the line width within the same line, and resulted in Matrix A and B, respectively. These two changes were done similarly across all naturalness levels therefore the manipulated naturalness is still solely based on the presence of curves. This was done to avoid interaction between different manipulations. Edge detection, as described in section 3.1.2, was used to verify whether complexity was similar across different naturalness levels.
23 3.2.2.1 Line width differences across lines (Matrix A)
With the use of Adobe Illustrator CC, the four high complexity (C4) patterns from Pilot 1 were recreated with differences in stroke weights across lines. Two steps were taken to come to Matrix A:
(1) the number of groups (with multiple lines) with a different stroke weight was determined and, (2) the difference in stroke weights had to be determined.
Multiple variations of the four C4 patterns were created. There was little difference between using 2, 3, 4 or 5 different stroke weights. Therefore, it is decided to proceed with the simplest version with only two different strokes present. In Pilot 1, all high complexity lines had a stroke width of 17.5 pt. For this Matrix A, half of the lines in the highly complex patterns were made thicker and half of the lines were made thinner, to ensure similar perforation ratios as in Pilot 1. To determine the upper and lower limit of the line widths different versions of the patterns were created, Appendix 7.1.1 shows all variations of the pattern matrix that were attempted. Based on these variations it was established that the deviation should not be too small, there is no change visible, and also should not be too large, resulting in the need for more groups of different line widths. Therefore, a middle ground with a deviation of 4.5pt was chosen with a lower limit of 13pt and an upper limit of 22pt.
With the four newly-created highly complex (C4) variations Matrix A was created, shown in Figure 12, using the same process as described in section 3.1.2. By using an equal increase and decrease from the mean stroke weight the perforation of the images (Table 3) stayed relatively the same and within the boundaries of 38% to 42%. Moreover, edge detection described in section 3.1.2 showed that complexity did not increase when changing the line width of the line, as illustrated in Figure 13.
Naturalness
N1 N2 N3 N4
Complexity C1C2C3
24
C4
Figure 12. Pattern Matrix A
Table 3. Perforation matrix A
N1 N2 N3 N4
C1 38% 41% 41% 41%
C2 40% 40% 40% 40%
C3 40% 40% 41% 38%
C4 40% 40% 39% 38%
Figure 13. Edge densities matrix A Figure 14. Zoom location matrix A
3.2.2.2 Line width differences within lines (Matrix B)
Secondly, the line widths within lines were changed to resemble the shape of a branch. The high complexity (C4) patterns presented in Figure 12 were used as the starting point, generating a second set of stimuli. This was done with the plugin Grasshopper 5.0 of Rhinoceros 6. Again, multiple variations were created before final settings were chosen, all variations can be found in Appendix 7.1.2.
For Matrix B, all lines were manipulated with a factor of 0.25. All lines from the patterns high in complexity from Matrix A were converted into shapes with the corresponding width of the lines used in the previous section (13pt and 22pt). The sides at the ends of the shape were then manipulated, one side became more narrow with a factor of 0.25, and the other side became wider with this same factor, as is illustrated in Figure 15.
25
Figure 15. Example lines with manipulation across and within lines from N3C4 with on the left a line with a width of 13pt and on the right a line with a line width of 22pt.
These newly created C4 images were used as input for the complexity manipulation with the magnification levels explained in section 3.1.2, resulting in Matrix B shown in Figure 16. By increasing the line width on one side of the line and decreasing the line width on the other side the perforation ratios remained relatively similar, as is shown in Table 4. However, the perforation in the low complexity images (C1) did not fall within the boundaries of 40 ±2 %. Since this was the optimal magnification location for C1 based on the perforation, it was chosen to proceed with this pattern matrix. Edge detection was used to ensure complexity would not increase over naturalness levels. No clear differences could be seen in the edge density graph (Figure 13) indicating that complexity did not increase by the changes in the naturalness manipulation.
Naturalness
N1 N2 N3 N4
Complexity C1C2
26
C3C4
Figure 16. Pattern matrix B
Table 4. Perforation matrix B
N1 N2 N3 N4
C1 36% 34% 45% 42%
C2 40% 39% 40% 40%
C3 38% 41% 40% 39%
C4 38% 41% 40% 41%
Figure 17. Edge densities matrix B Figure 18. Zoom location matrix B
3.2.3 Measurements, procedure and participants
Similar to the first pilot study, Pilot 2 was conducted online with the use of Limesurvey. After participants agreed that their data would be used for research purposes, participants answered the same three demographic questions about age, gender and country where most time was spent as in the previous pilot study. Each participant saw the images of either Matrix A or Matrix B.
The participants answered questions about perceived naturalness and perceived complexity in separate rounds; they saw each image twice. The survey took about ten minutes to complete.
27 Which concept was asked first was randomized as well as the order of the images. Only one pattern was presented at a time and backward navigation was not possible. After the participants answered all questions from one concept there was a page that clearly stated the end of this section with a comment field to make certain that the participant noticed that the next round with a different question started. For the round of questions about perceived naturalness, the definition of naturalness, how much the image reminds of something in nature, was added to make sure all participants answered these questions with the same definition in mind.
Perceived naturalness and perceived complexity were phrased similarly as in the first pilot:
“How natural would you rate this pattern?” and “How complex would you rate this pattern?”.
Answers were given on a similar 10-point Likert scale (1 = not at all complex/natural; 10 = very complex/natural) to enable the comparison of the results of Pilot 1 and 2.
In total 28 participants completed the survey and were randomly divided into groups to see either Matrix A or Matrix B (NA = 14; NB = 14). For Matrix A, eight participants were female and six male. All participants in group A were between the age of 22 and 27 (Mage,A = 24.3, SDage,A = 1.27) and all had spent most of their time in their life in the Netherlands. For Matrix B, participants’ age was ranging between 21 and 25 (Mage,B = 23.7, SDage,B = 1.14). Eleven out of the twelve participants were female. Nine participants of group B spent most of the time in their life in the Netherlands other countries that were mentioned were Germany (2), Denmark (1) and Spain (2).
3.2.4 Results 3.2.4.1 Matrix A
In Figure 19 and Figure 20 manipulated naturalness and manipulated complexity of Matrix A (Figure 12) are illustrated across perceived naturalness, respectively. Figure 20 shows that the levels of naturalness were not distinctive since the lines cross and overlap each other. The most natural and complex case (N4C4) shows a decrease in perceived naturalness compared to lower complexity levels but also compared to lower naturalness levels.
Figure 19. Perceived naturalness Matrix A-I Figure 20. Perceived naturalness Matrix A-II
The manipulation for complexity was tested by visualizing manipulated complexity across perceived complexity (Figure 21 and Figure 22). Figure 21 shows that across levels of manipulated
28 that all levels manipulated of complexity are distinct and separate from each other.
Figure 21. Perceived complexity Matrix A-I Figure 22. Perceived complexity Matrix A-II
3.2.4.2 Matrix B
Figure 23 and Figure 24 show the effect of the manipulation of naturalness and complexity of Matrix B on perceived naturalness. Both graphs show that the effect of naturalness is intertwined with manipulated complexity. Figure 23 illustrates that for each complexity level the naturalness increases with the manipulation as was intended. However, the increase is not the same across complexity levels. Figure 24 shows that the ratings of perceived naturalness are different for the manipulated groups N1 to N4. However, perceived naturalness increases for N2 to N4 with complexity. For these levels of naturalness, the higher the manipulated complexity the natural the patterns were perceived. The lowest level in naturalness (N1) does not show this increase.
Figure 23. Perceived naturalness Matrix B-I Figure 24. Perceived naturalness Matrix B-II
Figure 25 and Figure 26 show the effect of the manipulation of naturalness and complexity on perceived complexity. Figure 25 shows that the manipulated complexity levels show us an even increase in perceived complexity across different levels of naturalness. Figure 26 demonstrates that manipulated levels of complexity are different from each other and are steady across naturalness levels.
29
Figure 25. Perceived complexity Matrix B-I Figure 26. Perceived complexity Matrix B-II
3.2.5 Conclusions and implications
Similar to the results of Pilot 1, the manipulation of complexity was successful in both Matrix A and B. The different levels of complexity were distinctive and did not show any overlap indicating good manipulation.
The results Matrix A show that the different levels of manipulated naturalness did not show to be distinct in levels of perceived naturalness. This manipulation of naturalness did not yield the expected results, as the results were in a similar range as the first pilot study.
On the contrary, the manipulated naturalness in Matrix B led to evaluations of perceived naturalness that used a wider part of the scale, ranging from 2.7 to 8.5. The manipulation of naturalness was more distinct than in Pilot 1 or matrix A. Figure 23 and Figure 24 indicated that manipulated naturalness is intertwined with manipulated complexity on perceived naturalness. The lines in Figure 24 show an almost linear increase for complexity indicating a main effect of complexity for perceived naturalness. This indicates that perceived naturalness is dependent on manipulated naturalness as well as manipulated complexity.
Following these outcomes, Matrix B was selected to be used in the next part of the study as this matrix yielded the most successful results in the dimensions of perceived naturalness and perceived complexity. However, the perforation ratios of matrix B were not all within the previously set boundaries of 38% – 42%. Additionally, the changes done within the line width as in Matrix B, as described in section 3.2.2.2, were not as visible as hoped in high magnification level (C1). Therefore, in the following sections will be explained how this problem was dealt with how the final stimuli were selected.
3.3 Final selection of experimental stimuli
In order to make the manipulation of the line width within lines more evident, the same manipulation was done individually on each of the zoomed images instead of only on the four high complexity (C4) images. The approach for this set of stimuli was to use the four C4 patterns from matrix A, as input for the manipulation for complexity by magnifying the image. After the locations of magnification were determined, and the pattern matrix was made, on each individual pattern, the
30 described in section 3.2.2.2.
Since the first two steps of this approach were similar to the creation of Matrix A the locations to magnify the images were the same (Figure 14). The final patterns are shown in Figure 27 and their corresponding perforation in Table 5. By adding the manipulation within the line widths after the magnification the perforation ratios were all again within the boundaries 38% - 42%. Again, edge densities confirmed that complexity did not increase over naturalness levels, as is shown in Figure 28.
Naturalness
N1 N2 N3 N4
Complexity C1C2C3C4
Figure 27. Experimental stimuli
Table 5. Perforation
N1 N2 N3 N4
C1 41% 41% 41% 39%
C2 40% 41% 39% 40%
C3 40% 42% 41% 41%
C4 38% 41% 40% 41%
31
Figure 28. Edge densities experimental stimuli
To ensure complexity, the number of elements in the pattern, was similar the aim was to keep the composition as constant as possible across different naturalness levels. However, by magnifying the composition of the patterns were not as similar as hoped. Due to the curves in the naturalness manipulation, the lines deviated from the initial path resulting in different compositions. This is mostly visible in the lowest complexity level, C1. Edge densities of the patterns showed that the difference in composition did not have a great impact on the number of elements in the scene.
Additionally, the data from the pilots did not show clear differences when the composition differed between the four images in the lowest level of complexity. Since no clear differences were found and the location to magnify the image was chosen to optimize the perforation ratio for each pattern it was chosen not to change the composition after magnification.
3.4 Experimental study 3.4.1 Design
The experimental study was set up as a 4 x 4 within-subjects experimental design. Sixteen images of a simulated office environment, with patterns varying in manipulated naturalness and manipulated complexity as façade geometry, were compared. Renderings of daylight simulations of these scenes were shown to participants in an online questionnaire using the Limesurvey software.
The set of sixteen images was presented to the participants twice. In the first round questions about the dependent variables, perceived restoration and fascination, were included. In the second round, the questions were about perceived naturalness, complexity and pleasantness. In each round the images were randomized. The order of these rounds was fixed to ensure that the answers of possible mediators did not interfere with the answers for dependent variables. Within these two rounds, the questions are presented in random order.
3.4.2 Participants
Participants were recruited via the participant database Prolific (www.prolific.co). A total number of 40 participants were recruited via the database. The number of participants was
32 effect of f = 0.25. While the effects of naturalness and complexity on fascination and perceived restoration in the literature are found to be medium-to-large (Berman et al., 2014; Ommeren, 2019;
Twedt et al., 2019), in the present study the manipulation is expected to be more subtle, with expected effects smaller than the f = 0.35 found in Van Ommeren (2019) which manipulated both the color and the design of the visual stimulus. Six additional participants were recruited for pilot testing and correcting for potential drop out.
To be eligible to participate in this study, participants were required to be fluent in English and have a normal or corrected-to-normal vision. Therefore, these criteria were used as inclusion criteria in Prolific. Participants were asked to conduct the experiment on a laptop or desktop, not on a mobile device, so the full image was visible on the screen. Each experimental session lasted around 15 minutes and participants were compensated £1.90 for their time.
Forty-one responses were submitted in Limesurvey. Responses with a response time below 5 minutes or above 45 minutes were rejected. One response was rejected by Proflic due to the long response time but still submitted in Limesurvey after 2.5 hours. Because of the time limit, this observation was not included in the dataset. The remaining 40 participants (24 male, 15 female) included in the dataset had an average age of 25.5 (SD = 6.90, Min = 18, Max = 45). Participants were recruited from all over the world. Figure 29 shows the country of residence and the country where participants spent most of their lifetime. There were small differences between the country of residence and the country where participants spend most of their lifetime.
Figure 29. Geographical data participants
3.4.3 Stimuli
The patterns manipulated on complexity and naturalness were used as a façade geometry in a virtual room. The simulation was created from real-world meeting room 2.422 in the Atlas building on the campus of Eindhoven University of Technology (Figure 30). The dimensions of the room were
