RUNNING HEAD: Thinking Processes During Garden Design

RUNNING HEAD: Thinking Processes during Garden Design

Unearthing the creative thinking process: fresh insights from a think aloud study of garden design

Andrew Pringle, Paul T. Sowden

Department of Psychology, University of Surrey, United Kingdom

In press Psychology of Aesthetics, Creativity & the Arts

Author note:

Andrew Pringle, School of Psychology, University of Surrey, Guildford, United Kingdom and the Insight Centre for Data Analytics, University College Dublin. Paul Sowden, School of Psychology, University of Surrey, Guildford, United Kingdom. Both authors contributed to writing this paper with the first draft produced by the first author based on their doctoral dissertation at the University of Surrey. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. We wish to thank Matthew Peacock for his dedicated work as the second coder of verbal protocols and his advice and expertise. We also wish to thank Adrian Banks and Ken Gilhooly for their advice and expertise on the use of think-aloud protocols and Markov chain models. Correspondence concerning this article should be addressed to Andrew Pringle, Insight Centre for Data Analytics, University College Dublin, O’Brien Centre for Science. Belfield, Dublin 4, Dublin, Ireland. Tel.: +353 017162313. E-mail address: .

Abstract

A number of theories of creativity have converged on the idea that creative thinking entails shifting between different processes. We attempt to build on recent theoretical developments through empirical work to examine creativity in the everyday environment of a garden designer. We asked designers with different levels of expertise, a matched group of fine artists and a non-designer, non-artist control group to work on a garden design. We asked them to ‘think aloud’ as they designed and we recorded audio and video. We coded resultant verbal segments as indicating the operation of different types of underlying thinking process identified in recent theoretical work. We then mapped these segments to the video of the designs and conducted Markov chain analysis to explore how thinking processes shifted as the design evolved. Finally, we examined the extent to which different types of thinking process shifts predicted the creativity of the final garden designs as determined by experts. We found that shifts between associative and analytic thinking processes predicted design creativity, but only when the operation of these two processes were tightly coupled in time. The positive association between shifting and creativity was strongest when analytic thinking processed affective content. These types of shifting were also elevated at times when a subset of participants switched between working on different designs; a strategy that positively predicted design creativity. Findings suggest expansion of mode shifting theories of creative thinking to include the importance of close coupling between different modes of thinking and of an analytic mode processing affective content.

Keywords: Mode Shifting, Creative Thinking, Design, Think-Aloud Method, Markov Chain Models

A number of theories of creativity have converged on the idea that creative thinking entails shifting between different processes (e.g. Basadur, Graen & Green, 1982; Basadur, 1995; Dietrich, 2004; Finke, Ward Smith, 1992; Gabora & Ranjan, 2013; Howard-Jones, 2002; Nijstad, De Dreu, Rietzschel & Baas, 2010). These processes resemble aspects of broader dual-process theories of cognition (Evans & Stanovich, 2013; Frankish, 2010; Stanovich & Toplak, 2012) and recent reviews have critically examined this similarity and the implications for our understanding of creativity (Allen & Thomas, 2011; Sowden, Pringle & Gabora, 2015). Most theories of the creative thinking process propose that creativity requires the generation of ideas that are then evaluated and/or honed for their intended purpose, with a growing emphasis that creativity hinges on the ability to shift between different modes of thinking supporting generative and evaluative activities (Gabora & Ranjan, 2013; Howard-Jones, 2002; Kaufman, 2011). In fact Kaufman (2011) has argued, “the highest levels of creativity require…the flexibility to switch modes of thought throughout the creative process” (p. 458). Further, computational work has developed shifting algorithms that model the human creative process (Veloz, Gabora, Eyjolfson & Aerts, 2017) and human cultural evolution (Gabora, Chi & Firouzi, 2013). In addition, laboratory studies (see Vartanian, 2009; Vartanian, Martindale & Kwiatowksi, 2007; Dorfman, Martindale, Gassimova & Vartanian, 2008; Beaty, Silvia, Nusbaum, Jauk & Benedek, 2014) and our own recent psychometric work (Pringle & Sowden, 2017) have provided empirical support for the positive association between creativity and mode shifting. However, empirical work has yet to look ‘under the hood’ at an ecologically valid example of human mode shifting to determine if mode shifting observed in a real-life creative process is linked to the creativity of the product produced at its conclusion. Thus, the key goal of the present study was to explore this important issue regarding the ecological validity of ideas about mode shifting. To do so we examined the creative process in garden design.

Looking ‘under the hood’ at the creative process during garden design

We chose garden design as the everyday environment within which to examine the creative process for a number of reasons. First, design is a recognized area of creative endeavor requiring mode shifting to generate and evaluate ideas (Cross, 2011; Dorst Cross, 2011). Second, designing a garden is a task that can be engaged in by those without specialist knowledge but where significant expertise and skill can be developed with training. This was crucial for the present study as both expert and non-expert groups of participants were included. Third, professional garden designers are capable of sketching garden designs in a short time period (e.g. within forty-five minutes) and often have to do so for clients (Fischer-Tomlin, A, personal communication, 2013). This was important as the design task had to be short enough that it could be completed within a single session to make it manageable for participants and those coding the data.

The data were video footage and verbal protocols generated by a ‘think-aloud’ process as participants worked on designing a garden. Further, the finished creative product produced at the end of the process; that is the individual’s final sketch of their garden design, was rated by expert judges for its creativity, design quality and for how closely it met the design brief. This method closely resembles that used in previous work in this journal examining creativity in visual art (Fayena-Tawil, Kozbelt & Sitaras, 2011) and related work in the journal Design Studies that commonly makes use of the ‘think-aloud’ method and ‘protocol analysis’ to examine the design process (e.g. Atman, Chimka, Bursic & Nachtmann, 1999), which we elaborate next.

The ‘Think-aloud’ method and ‘Protocol analysis’

The ‘Think-aloud’ method involves participants continually thinking-aloud their thoughts as they work on a task, in this case designing a garden. In general, the ‘think-aloud’ method is found not to effect task performance (Ericsson & Simon, 1993). It has been used previously to examine components of creativity, namely divergent thinking (Gilhooly, Fiortou, Anthony & Wynn, 2007) and insight problem solving (Fleck & Weisberg, 2004), with no differences in task performance found between groups completing a task while thinking-aloud compared to a control group completing a task silently. In the present case, all of a participant’s utterances were transcribed resulting in a verbal protocol for the entire creative process. The visuals from the video data were recorded alongside the audio of the verbal protocol. This verbal-visual protocol was then divided into segments, with a segment defined as words, phrases or sentences of any length that made up one distinct statement about something such as an idea or topic (Suwa & Tversky, 1997; Atman et al., 1999; Gilhooly et al., 2007). Short segments (typically 5 to 10 sec’s) were used to allow a fine-grained analysis of the timing of shifting between modes of thought. Importantly, a key feature of the present work was its use of a detailed theoretical framework of mode shifting to inform the coding scheme that was developed as described next.

Theoretical framework of mode shifting and coding scheme

A crucial feature of the present study’s approach is that it allows a critical test between multiple theories of mode shifting at once to determine which best explains the empirical data. Theories of mode shifting differ with respect to (1) the number of different components between which shifts occur (2) the role of affective processing (3) the degree to which the different component processes are coupled, reflected by how closely together in time they occur (4) whether the frequency of mode shifting is important and (5) whether the timing of shifts are important.

In general, dual-process theories of creativity and dual-process theories of cognition include an associative mode of thinking and an analytic mode (Sowden et al., 2015). Based on this commonality, the decision was made to pool attributes of thinking processes across models of creativity (Howard-Jones, 2002; Gabora, 2005) and dual process models of cognition (Evans & Stanovich, 2013; Frankish, 2010; Kaufman, 2011) in order to identify the operation of associative and analytic modes of thinking in protocol segments.

Further, most models of creativity that incorporate different modes of thinking only differentiate between the two modes based on their cognitive characteristics (see Howard-Jones, 2002; Gabora, 2010; Gabora & Ranjan, 2013; Vartanian, 2009). However, recent neuroimaging work suggests that affective processes, supported by default and limbic brain regions, are involved in the evaluation of ideas during a creative task (Ellamil, Dobson, Beeman & Christoff, 2012) and Dietrich (2004) has proposed a model of the interplay between different modes of thinking in creativity that does emphasize the need to consider affective processing. Thus, the present work further coded thinking as with and without affective content in the analyses of mode shifting. Consequently, coding first identified segments as associative or analytic before assessing whether each segment contained affective content or not resulting in four overarching codes for coding the verbal protocols: analytic-cognitive, analytic-affective, associative-cognitive, associative-affective.

The final coding scheme, with attributes of the different modes of thinking pooled across the theoretical models discussed (Dietrich, 2004; Evans & Stanovich, 2013; Frankish, 2010; Gabora, 2005; Howard-Jones, 2002; Kaufman, 2011) is shown in table 1 with attributes of each mode of thought shown in the ‘segment code’ column and the theoretical models that attributes are taken from indicated in the ‘source model(s)’ column. The ‘explanation’ column explains what each attribute is and the ‘example’ column gives an example of this attribute as it appears within verbal protocols. This approach allows a comparison of the two component models with the model that additionally separates out affective processes in analytic thinking derived from Dietrich (2004).

2

RUNNING HEAD: Thinking Processes during Garden Design

Table 1. Displaying the Coding Scheme used to code segments of visual-verbal protocols with either ‘associative’ or ‘analytic’ as well as indicating the further separation of segments into those with affective content or not.

Associative mode

Analytic mode

2