Representational Ecosystems in Architectural Design Studio

HAL Id: hal-01916562https://hal.archives-ouvertes.fr/hal-01916562

Submitted on 16 Nov 2018

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Representational Ecosystems in Architectural DesignStudio Critiques: Do changes in the representational

ecosystem affect tutors and students behaviors duringdesign critiques?

Julie Milovanovic, Daniel Siret, Guillaume Moreau, Francis Miguet

To cite this version:Julie Milovanovic, Daniel Siret, Guillaume Moreau, Francis Miguet. Representational Ecosystems inArchitectural Design Studio Critiques: Do changes in the representational ecosystem affect tutorsand students behaviors during design critiques?. eCAADe 2018 - 36th Annual Conference 17th-21stSeptember 2018 Lodz, Sep 2018, Lodz, Poland. pp.351-360. �hal-01916562�

https://hal.archives-ouvertes.fr/hal-01916562

https://hal.archives-ouvertes.fr

Representational Ecosystems in Architectural DesignStudio Critiques

Do changes in the representational ecosystem affect tutors and studentsbehaviors during design critiques?

Julie Milovanovic1, Daniel Siret2, Guillaume Moreau3,Francis Miguet41,2,4AAU-CRENAU,Graduate school of Architecture, Nantes (France) 3AAU-CRENAU,Ecole Centrale Nantes (France)1,2,4{julie.milovanovic|daniel.siret|francis.miguet}@[email protected]

Design studio critiques are key moments for students' learning and designingprocesses. During critiques, the representational ecosystem provides a setting forthe critique to unfold. Tutors and students, while presenting and discussingstudents' designs, interact with each other and the representational ecosystem. Inthis article, a case study illustrates our method to measure the effect of a changeof representational ecosystem on the critiques' activity. Our three settings includetraditional desk critiques, 1/50 scale mockup critiques and immersive VirtualReality critiques (with HYVE-3D). Each type of critique is analyzed by usingvideo coding as well as protocol analysis.

Keywords: studio critiques , representational ecosystem , protocol analysis,pedagogic strategies, cognitive behavior

INTRODUCTIONThedesign studio is a cornerstone in architectural ed-ucation. Its pedagogic format fosters a learning-by-doing situation. The studio organization focuses on adesigning task, where students develop a project toanswer design requirements, synthesized in the de-sign brief set by tutors. Design critiques and juriesare milestones in the students’ design developmentand learning process. Design representations act asa designing tool while students work individually ontheir project, and as a communicational and design-ing tool when they present and discuss their design

with tutors during studio hours. In this article, weexplore the effect of the use of three different rep-resentational ecosystems on tutors and students in-teractions, manipulation of representations and cog-nitive design behaviors during studio critiques. Arepresentational ecosystem includes all the types ofexternal representations produced or used during adesign activity (Dorta et al. 2016). The first repre-sentational ecosystem we studied is the traditionaldesk critique, where students bring printed draw-ings, sketches, and 3D models; the second one isa large scale mockup critique (scale 1/50) and the

CAAD EDUCATION | Concepts & Strategies - Volume 1 - eCAADe 36 | 351

last one is a critique set in the HYVE-3D (Dorta et al.2014), which is an immersive scale 1 representationalecosystem.

Systems using Virtual Reality or interactive Aug-mented Reality tabletops are increasingly broughtinto design studios (Angulo 2015; Dorta et al. 2016;Schubert et al 2016; Sopher et al. 2017) but there isa lack of empirical research to study the impacts ofsuch an alteration of the design space. In this arti-cle, we make a first step to fill that gap by propos-ing an illustrated methodology to study the effect ofa change in the representational ecosystem (here a1/50 scale model and HYVE-3D). We specifically fo-cus on its impact on three key elements that definedesign studio critiques: its format (pedagogic strate-gies and feedback), its content (design cognitive be-havior) and interactions with its settings (representa-tional ecosystem). Pedagogic strategies are molds toconvey design knowledge during critiques. In it, thecenter of activity is akin to design itself since tutorsmight demonstrate how to reframe a problem to trya new solution (see for example Petra and Quist inSchön 1985). Design representations, embedded inthe representational ecosystem, act as materials forreflexive conversations (Schön 1992) during the cri-tique. Our three key elements are connected and of-fer a global approach to better grasp what happensduring design studio critiques. Our methodology,based on in situ observations and protocol analyses(Ericsson&Simon1984), proposes a framework to an-alyze whether the use of different representationalecosystems impacts how the critique unfolds. Ourcase study of six students, comparing three critiquessettings, will highlight their differences and similari-ties regarding each element.

PEDAGOGICSTRATEGIESANDFEEDBACKSDURING DESIGN CRITIQUESThe design critiques create situations and design ex-perienceswhere students build their designerlywaysof knowing by seeing their tutors designing, reflect-ing on their design and proposing new solutions.Collaboration is promoted in the studios and gives

students anopportunity to co-construct their design-ing skills. Tutors act as coaches (Adams et al. 2016)to provide suitable feedback to help students bridgethe gap between the design knowledge they need toreach their goal and their current design knowledge.The literatureon tutors‘ typeof feedbackor strategiesduring studio critiques is rich (Adams et al. 2016; Car-doso et al. 2016; Cennamo & Brandt 2012; Dannels &Martin 2008; Goldschmidt et la. 2010; Heylighen etal. 1999; Marbouti et al. 2017; Schön 1985; Uluoglu2000; Yilmaz & Daly 2016). Based on our literature re-view, we propose a classification of tutors’ feedbacktypes into four main categories : scaffolding impliesreasoning, questioning and reflecting on a design is-sue, explaining / instructing illustrates a descriptiveand explicit approach of discussing a design issue,demonstrating / proposing involves the formulationof new ideas, changes in the design and suggesting/ exploring calls for experimentations and openingthe design process (Table 1). For our study, we an-alyzed if a change in the representational ecosystemimpacted the use of each of those feedback strate-gies.

DESIGN COGNITIVE BEHAVIOR DURINGDESIGN CRITIQUESThe content of feedback during the critique focuseson design as a process or as an object. Tutors maypoint out a problem in students design and engagein a design activity to demonstrate how to resolve itor explain why a part of the design is problematic.The content of feedback is of interest while studyingcritiques because they convey design knowledge. Ageneral way to describe design knowledge is givenby the Function Behavior Structure ontology (Gero1990). The FBS framework represents six design is-sues: a Requirement (R) includes the design briefand is outside of the designer, a Function (F) is whatthe design object is for, a Behavior (Be) representsan expected behavior of the design object, a Struc-ture (S) is an element or a structure of elements ofthe design object, a Behavior (Bs) is a behavior de-rived from a structure and a Description (D) is an ex-

352 | eCAADe 36 - CAAD EDUCATION | Concepts & Strategies - Volume 1

Table 1Description of fourcategories offeedbacks duringdesign critiquesbased on aliterature review

ternal representation of the design object. The FBSframework accounts a total of eight cognitive designprocesses showing transitions between the six de-sign issues: Formulation, Synthesis, Analysis, Evalu-ation, Documentation, Reformulation 1, Reformula-tion 2 and Reformulation 3, as showed in Figure 1(Gero 1990; Gero & Kannengiesser 2004). The FBSframework provides a theoreticalmodel of design ac-tivities that can be mapped onto the studio critiqueto analyze tutors and students cognitive design be-haviors (Gero& Jiang2016;Milovanovic&Gero2018).

Figure 1FBS framework(source Gero 1990;Gero &Kannengiesser,2004)


In this study, we explored how a change in the rep-resentational ecosystem impacted the occurrencesanddistributionof FBSdesignprocesses aswell as tu-tor / student interactions while designing.

REPRESENTATIONAL ECOSYSTEM TO SUP-PORT DESIGN CRITIQUESFeedback are delivered within the representationalecosystem. During the critiques, it sets a designingand learning environment. Tutors and students acton it, manipulate representations while presentingand discussing potential design issues. Accordingto Dorta et al. (2016), this ecosystem should havefour qualities: support hybrid representations, whichmeans that it should include physical and digital rep-resentations; integrate multiple types of representa-tions (2D, 3D, animations); includemultiple scales, ar-chitectural scales and an immersive scale 1 represen-tation; and foster an intuitive co-design situation. An-other characteristic to be added is a synchronizationof the design representations, to offer an updatedholistic perception of the project. It implies that rep-resentations forming the representational ecosystemare connected to each other. Five different character-istics defined the ecosystem, which we can synthe-size into materiality, dimensions, scales, interactionsand synchronization. In our analysis, design critiquesoccurred in threedifferent settingswithdifferent rep-resentational ecosystem characteristics that are fur-ther developed in the data description section.

Basedonour observations of studio critiques, weidentified several actions tutors and students take tocreate or interact with design representations. Rep-resentation are pointed out to refer to a special ele-ment of the design. Tutors and students might gen-erate a physical representation by drawing on a pa-per or tearing down a part of the mockup. If a digitalmodel of the projectwas brought for the critique, thedesign space can be navigated bywalk through or flyover. Some authors like, Visser (2009), noted the im-portance of gesture during design meetings and wealso observed occurrences of spatial gesture to rep-resent or explain a spatial quality of students designproject.

RESEARCHQUESTIONSIn the light of the description of each of our threekey elements, we can refine the research questionswe introduced. Our comparative study aims to tacklethe following: Does a change in the representationalecosystem affect tutors’ feedback strategies and stu-dents reactions to feedbacks? Does a change in therepresentational ecosystem impact tutors and stu-dents’ cognitive design behavior? Does a change inthe representational ecosystem have an effect on tu-tors and students’ actions on and interactions withdesign representations?

METHODOLOGYIn our study, the protocol analysis method (Erics-son & Simon 1984; Gero & Mc Neill 1998), is ex-ploited to study how the critique unfold. Each cri-tique was video recorded to be further analyzed. Weused two levels of coding for our dataset. The firstlevel of coding focuses on feedback types and actorsinteractions with the representational environment.Videos are directly coded with four types of feed-backs (scaffolding, explaining / instructing, demon-strating / proposing and suggesting / exploring) andfour types of interactions with the representationalecosystem (pointing, navigating, generating a phys-ical representation and spatial gesture) using Atlas.tisoftware. The second level of coding focuses on de-sign cognitive behavior. This analysis is at a finergrain and is based on the verbal transcripts of thedesign critiques. Each transcribed protocol is seg-mented and encoded with one of the six FBS de-sign issues (Requirement, Function, expected Behav-ior, Structure, Behavior form structure and Descrip-tion). For both level of coding, protocols are alsocoded with the speakers, either tutor or student. Thedouble coding gives more information on actors’ in-teractions during the critique. For example, it revealsif design processes are constructed by a single actoror co-constructedbetween tutors and students. Eachvideo and transcribed protocol was coded twice sev-eral weeks apart, by the same researcher, to ensuremore reliability. The two versions of each protocol


were compared and arbitratedwhen a different codewas associated with the same verbalization.

DATA DESCRIPTIONA case study with six students was conducted withintwo master architectural design studios selectedat the Graduate School of Architecture of Nantes(France).

Figure 2Threerepresentationalecosystems testedfor critiques

We studied three different representational ecosys-tems, with two students for each setting. In the firststudio, we observed the desk critique setting as wellas the mockup critique setting (Fig.2a and 2b). Stu-dents had to design an hybrid public equipment inthe Parisian suburb, that includes a city museum, asports and spa center and co-working spaces. Fromthe beginning of the studio, it was required for stu-dents to build a 1/50 scale mockup of the site sostudents could experience design critiques in thatsetting. Our case study includes two desk critiques(week 7 out of 14 studio weeks) and twomockup cri-tiques (week 10 out of 14 studioweeks) from that stu-dio. The two other cases of our dataset are HYVE-3D critiques (Fig.2c). A two-day workshop was or-ganized with a few students of the second masterstudio in order for them and the tutor to learn howto use the HYVE-3D (in collaboration with the LIDlab in Bordeaux, France and the Hybridlab in Mon-treal, Canada). The design brief was also a hybridprogram including a museum and a hotel situatedin Palm Springs, California, to explore Jacques Tati’sfilmmaking world. At the end of the two days’ work-shop, students presented their project in the HYVE-3D. The workshop took place during week 7, out ofthe 14-week long studio.

The specificities of our three settings can bedescribed based on the representational ecosystemcharacteristicswedefined (Table 2). Thedesk critiqueis a traditional setting where students brings plansand sections, printed or hand sketched. The mockupcritiques focusmainly on themockup itself but somestudents also brought plans and sketches. TheHYVE-3D critiques are immersive since the device offers a180° screen. Actors can navigate the virtual spacewith the 3D cursor, which also serves as a 2D sketch-ing interface (Dorta et al. 2015).

RESULTSFeedbacks strategiesBasedon thevideoanalysiswithAtlas.ti, weextractedthe time spend by tutors formulating each of the fourtypes of feedback: scaffolding, explaining / instruct-


Table 2Description of thecharacteristics ofeach setting used inour case study

ing, demonstrating / proposing and suggesting / ex-ploring (Fig. 3a). Students‘ reactions to feedbackwere also coded with the same categories (Fig.3b).Moments where the current project was describedare not coded. For all the critiques, tutors spent be-tween 60 and 90% of their time speaking, formulat-ing feedback on students’ design (Fig.3a). The dis-tribution of the feedback types varies across the cri-tiques. In this dataset, no trend appears concern-ing the effect of the representation of the ecosys-tem of tutors‘ feedback formulation. Students spendmost of their time speaking, presenting their project.The most active students spend around 20% of theirtime reacting to feedback with similar strategies astutors’, and the less active between 5 and 10%of theirtime (Fig. 3b). We can notice that the two students

from the HYVE-3D critiques are part of the most ac-tive ones in terms of feedback reactions.

Interactions with design representationsConcerning actors’ interactions with design repre-sentations, we can see that the time spent speakingwhile interacting with design representations variesfrom 25% of the time (HYVE_3D critique 2) to 57% ofthe time (Desk Critique 2) (Fig.4). For desk critiquesand mockup critiques, pointing is the prevailing in-teractions. For HYVE-3D critiques, pointing is not asdominant as for the other setting. Navigation wasmainly used in the HYVE-3D critiques. Generatinga new representation (in our case by sketching) oc-curred only in two critiques, one in the HYVE-3D andone in the desk critiques.

Figure 3(a) normalizeddistribution oftutors’ feedbackstrategies for eachcritiques (b)normalizeddistribution ofstudents’ feedbackreactions for eachcritiques


Figure 4Normalizeddistribution ofactors’ interactionwith representationfor each critiques

Cognitive design behaviorTo study cognitive design behavior, we exploited theFBS framework and first order Markov models to re-veal design patterns specific to our dataset. Critiquesconversations were transcribed and codedwith bothFBS design issues - Requirements (R), Function (F), ex-pected Behavior (Be), Behavior from structure (Bs),Structure (S), Description (D) - and the actor speaking- tutor or student. Each protocol was coded twice forbetter reliability. Since coding was time-consuming,wewere able to analyze only three students’ protocolfor this study, one for each representational ecosys-tem. The distribution of the design issues for a ses-sion gives a description of the nature of the designactivity (Kan & Gero 2017).

The normalized distribution of design issues peractor for our three critiques is represented in Fig-ure 5. Behavior derived from structure (Bs) is alwaysdominant, for both tutors and students for every cri-tiques. The distribution of design issues formulatedby students is similar for each representational en-vironment except for Function (F) and expected Be-havior (Be). For those two design issues, the studentfrom the desk critique formulated twice as much asthe other two. Concerning tutors’ distribution of de-

sign issues, we notice more variation than for stu-dents. The tutor from the desk critique shows amorebalanced distribution of design issues than the othertwo tutors. For the tutor from the mockup critique,Behaviors, either expected or derived from structures(Be / Bs), are the dominant design issues. Behaviorsfrom structure (Bs) and Structure (S) are prevailing forthe tutor from the Hyve-3D critique.

Figure 5Normalizeddistribution ofdesign issues peractor for eachcritiques

The interest in using thefirst orderMarkovmodelto analyze our data set is to reveal its design pat-terns (Kan & Gero 2010; Milovanovic & Gero 2018;Yu & Gero 2016). The Markov model offers a quan-titative probabilistic description of the design transi-tions, that mapped onto FBS design processes. Foreach critique, we can capture qualitative informationon the probability a design transition will occur. In-deed, a first order Markovmodel shows the probabil-ity of transitioning fromagiven state to another state(in our case design issues). The Markov analysis pro-duces a probability matrix based on the sequence ofevent states in our data set. The sequence is givenby the actor and the FBS design issue. In our data,12 states are described, that are associated to designissues (one of the 6 design issues from the FBS on-tology) and actors (tutors or students). The transition


Figure 6Representation ofmain design issuetransitions basedon their probability:(a) Desk critique, (b)Mockup critique, (c)HYVE-3D critique

probability varies between 0 and 1. Transitions witha high probability (above the selected threshold of0.17, two times the randomprobability) are represen-tative of the most probable design transitions fromthe starting design issue.

Figure 6 represents design issue transitions withthe highest probability for each type of critique. Thediagram shows transitions that are formulated by asingle actor, either tutors or students, as well as tran-sitions that are co-constructed. Moreover, we dis-tinguished two spaces within the design space : theproblem space that includes Requirements (R), Func-tion (F), and expected Behavior (Be); and the solu-tion space that includes Behavior from structure (Bs),Structure (S) and Description (D). Designing entails anavigation of the problem and solution space, which

co-evolve across time (Dorst & Cross 2001; Maher& Poon 1996). In the graphic representation of theMarkov transitions in Figure 6, we can notice whendesign processes occur within a single space or showa transition from the problem space to the solutionspace and reversely.

Most design transitions in our data set are soloconstructed, either by the student or by the tutor. Foreach critique, only one or two co-constructed tran-sitions’ probabilities are above our threshold, froma student’s formulated design issue to a tutor’s de-sign issue (Fig.6). Transitions occurredmainly in a sin-gle space, the solution space or the problem space,or show a shift from the problem space to the so-lution space. We observe more transitions from theproblem space to the solution space in the desk cri-


tiques when the tutor leads the critique’s activity(Fig.6a). Students from the desk critique (Fig.6a) andthemockup critique (Fig.6b) have similar design tran-sition patterns, whereas the student from the HYVE-3D critique (Fig.6c) shows different ones. Tutors fromthe mockup critique (Fig.6b) and the HYVE-3D cri-tique (Fig.6c) have similar design transitions patterns.In the desk critique (Fig.6a), some similar design tran-sitions appear but we can see that the activity ismainly situated in the problem space or going to-ward the problem space.

DISCUSSION, LIMITS AND PERSPECTIVESOur results illustrate differences regarding the typeoffeedback formulated by tutors, although it does notseem tobe related to the representational ecosystemused. Students‘ reaction to feedback, on the otherhand, are more dynamics in the HYVE-3D critiquesthan during the other critiques. That could be a signthat this setting is more engaging for students toparticipate. Previous studies showed how the HYVE-3D environment fosters collaboration during designstudios that relates to our observations (Dorta et al.2012). The type of interactions with the design rep-resentations differs depending on the critique rep-resentational ecosystem, especially for the HYVE-3Dcritiques. In the HYVE-3D, actors did not point at rep-resentations as much as in the other settings, whichcould be a consequence of being immersed in therepresentation. We found similarities and differencesin the dominant design cognitive processes occur-ring during critiques. In a study, Yu and Gero (2016)showed how the use of a differentmodeling environ-ment for solo design sessions impacted on the occur-rence of design processes. In our case, we also founddifferences in the occurrence of main design pro-cesses but that can hardly be connected to the rep-resentational environment used. Students’ cognitivebehaviors during the desk critique and the mockupcritique are alike, and different from the students‘cognitive behavior in theHYVE-3D critiques. That dif-ference does not match with the difference in tutors’design cognitive behavior depending on the setting.

We were expecting to observe a richer design inter-actions through co-constructed design processes inthe HYVE-3D critiques but co-constructed processeswere not the most probable in our case study.

Our sample is small to infer any general conclu-sions on the impact of the use of different setting onfeedback strategies, actors’ interactions with designrepresentations or specific design patterns. Tutorsand students form the HYVE-3D critiques only usedthis representational ecosystem for a limited time, soits manipulation is not as seamless as the desk cri-tique setting or the mockup setting. For that rea-son, the results presented in our study are to be takencarefully. Nonetheless, this case study illustrates thecomplexity and diversity of the designing and learn-ing activity during critiques. Actors’ behavior dur-ing the critiques was different but we need a widerstudy to better grasp if those changes are correlatedwith the representational ecosystem used. Designcritique, similarly to design itself, is a situated activity,actor-dependent and evolving through time. The sit-uatedness of design critiques affects tutors and stu-dents’ behavior during critiques that can also be areason for changes in theway feedback are deliveredanddesigning unfold. Our futureworkwill consist onconveying similar analysis using the samemethodol-ogy on a bigger sample, and better training for ex-periential settings to confirm trends that appeared inthe presented case study.

REFERENCESAdams, RS, Forin, T, Chua, M and Radcliffe, D 2016, ’Char-

acterizing the work of coaching during design re-views’, Design Studies, 45, pp. 30-67

Angulo, A 2015, ’Rediscovering Virtual Reality in the Ed-ucation of Architectural Design: The immersive sim-ulation of spatial experiences’, Ambiances. Environ-nement sensible, architecture et espace urbain

Cardoso, C, Badke-Schaub, P and Eris, O 2016, ’Inflec-tion moments in design discourse: How questionsdrive problem framing during idea generation’, De-sign Studies, 46, pp. 59-78

Cennamo, K and Brandt, C 2012, ’The “right kind oftelling”: knowledgebuilding in the academicdesignstudio’, Educational Technology Research and Devel-


opment, 60(5), pp. 839-858Dannels, DP and Martin, KN 2008, ’Critiquing Critiques:

A Genre Analysis of Feedback Across Novice to Ex-pert Design Studios’, Journal of Business and Techni-cal Communication, 22(2), pp. 135-159

Dorst, K and Cross, N 2001, ’Creativity in the designprocess: co-evolution of problem–solution’, DesignStudies, 22(5), pp. 425-437

Dorta, T, Kinayoglu, G and Boudhraâ, S 2016, ’A new rep-resentational ecosystem for design teaching in thestudio’, Design Studies, 47, pp. 164-186

Dorta, T, Kinayoglu, G and Hoffmann, M 2014 ’Hyve-3D:a new embodied interface for immersive collabora-tive 3D sketching’, SIGGRAPH 2014, pp. 1-1

Dorta, T, Kinayoglu, G and Hoffmann, M 2015 ’Hyve-3Dand rethinking the “3D cursor”: unfolding a naturalinteraction model for remote and local co-design inVR’, ACM SIGGRAPH 2014 Studio, New York, pp. 1-1

Dorta, T, Lesage, A and Bartolo, CD 2012 ’Collabora-tion and Design Education Through the Intercon-nectedHIS: Immature vs. Mature CI LoopsObservedThrough Ethnography by Telepresence.’, Proceedingsof the 30th eCAADe Conference, Prague, Czech Re-public, pp. 97-105

Ericsson, KA and Simon, AH 1984, Protocol Analysis: Ver-bal reports as data, MIT Press

Gero, JS 1990, ’Design prototypes: a knowledge repre-sentation schema for design’, AIMagazine, 11(4), pp.26-36

Gero, JS and Jiang, H 2016, ’Exploring the Design Cog-nition of Concept Design Reviews Using the FBS-based Protocol Analysis’, in Adams, RS and Siddiqui,JA (eds) 2016, Analyzing Design Review Conversa-tions, Purdue University Press, p. 177

Gero, JS and Kannengiesser, U 2004, ’The situated func-tion–behaviour–structure framework’, Design Stud-ies, 25(4), pp. 373-391

Gero, JS and Mc Neill, T 1998, ’An approach to the analy-sis of design protocols’, Design Studies, 19(1), pp. 21-61

Goldschmidt, G, Hochman, H and Dafni, I 2010, ’The de-sign studio “crit”: Teacher–student communication’,Artificial Intelligence for Engineering Design, AnalysisandManufacturing, 24(03), pp. 285-302

Heylighen, A, Bouwen, JE and Neuckermans, H 1999,’Walking on a thin line—Between passive knowl-edge and active knowing of components and con-cepts in architectural design: Between passiveknowledge and active knowing of components andconcepts in architectural design’, Design Studies,20(2), pp. 211-235

Kan, JW and Gero, JS 2010 ’Exploring quantitative meth-ods to study design behavior in collaborative virtualworkspaces’, New Frontiers, Proceedings of the 15thInternational Conference on CAADRIA

Kan, JW and Gero, JS 2017, Quantitative Methods forStudying Design protocols, Springer

Maher, ML and Poon, J 1996, ’Modeling design explo-ration as co-evolution’, Computer-Aided Civil and In-frastructure Engineering, 11(3), pp. 195-209

Marbouti, F, Mendoza-Garcia, J, Diefes-Dux, HA andCardella, ME 2017, ’Written feedback providedby first-year engineering students, undergradu-ate teaching assistants, and educators on designproject work’, European Journal of Engineering Edu-cation, pp. 1-17

Milovanovic, J and Gero, J 2018 ’Exploration of cogn-tivie design behavior during design critiques’, DE-SIGN2018, Dubrovnik, Croatia, p. 12

Schubert, G, Strobel, B and Petzold, F 2016 ’TangibleMixed Reality’, Proceedings of the 21st InternationalConference on Computer-Aided Architectural DesignResearch in Asia, Melbourne, Australia, pp. 333-342

Schön, D 1985, The Design Studio, RIBA, LondonSchön, D 1992, ’Designing as reflective conversation

with the materials of a design situation’, Knowledge-based systems, 5(1), pp. 3-14

Sopher, H, Kalay, YE andFisher-Gewirtzman,D2017 ’WhyImmersive? Using an Immersive Virtual Environ-ment in Architectural Education’, Education and re-search in Computer Aided Architectural Design in Eu-rope, Rome, Itlay

Uluo�lu, B 2000, ’Design knowledge communicated instudio critiques’, Design Studies, 21(1), pp. 33-58

Visser, W 2009, ’The function of gesture in an architec-tural design meeting’, in McDonnell, JT and Lloyd, P(eds) 2009, About Designing: Analysing design meet-ings, CRC Press, pp. 269-284

Yilmaz, S and Daly, SR 2016, ’Feedback in concept de-velopment: Comparing design disciplines’, DesignStudies, 45, pp. 137-158

Yu, R and Gero, JS 2016, ’An empirical basis for the useof design patterns by architects in parametric de-sign’, International Journal of Architectural Comput-ing, 14(3), pp. 289-302


Documents

Representational Ecosystems in Architectural Design Studio