[ CogSci Summaries home | UP | email ]

Bilda Z., & Gero J. (2006). Reasoning with internal and external representations: A case study with expert architects, in R. Sun (ed), Proceedings of the Annual Meeting of Cognitive Science Society , (pp. 1020-1026). Mahaw, NJ: Lawrence Erlbaum Associates.

author = 	{Bilda, Zafer and Gero, John},
title =		{Reasoning with internal and external representations: A case study with expert architects},
booktitle =	{Proceedings of the Annual Meeting of Cognitive Science Society},
pages = 	{1020-1026},
year = 		{2006},
editor =    	{R. Sun},
address=    	{Mahaw, NJ},
publisher = 	{Lawrence Erlbaum Associates} 

Author of the summary: Nicolas Di Noia Blank, 2007, ndnoia@scs.carleton.ca

Cite this paper for:



Within a design context sketching is used as an aid to visually store and retrieve concepts. Additionally, sketching may facilitate more ideas and help in revising and refining them. This study examined whether designing is possible without using sketching. Six architects engaged in a protocol study of design within two conditions: sketching & no sketching. The resulting sketches were assessed by judges and were found to have no significant differences in quality. The analysis of the design protocols used did not demonstrate differences in the number of cognitive actions used by the architects in the two conditions. The results imply that expert designers can design without the use of external representations.


Sketches are a type of external representation. "There is evidence in a design context that sketching facilitates ideas and design concepts (Goel, 1995; Goldschmidt, 1991; Do et al. 2001; Suwa & Tversky, 1997; Purcell & Gero, 1998)"(pg. 1). The drawings used by designers are used as a tool for thinking. "Tversky (1999) states, 'Drawings provide insights into conceptualizations not just imaginings' (p.94)"(pg.1). Re-interpretation of visual information can occur when the designer inspects his/her sketches. "This draw-inspect-revise cycle has been emphasized in various design contexts (Goldschmidt, 1991; Goel, 1995; Lawson, 1990) often referred to as reflective conversation (Schön & Wiggins, 1992)"(pg.1).
A previous experiment by Athavankar (1997) showed that an industrial designer was able to evolve the shape of an object, manipulate it, evaluate modifications and add details such as color without access to sketching. The results of that study led Bilda and Gero to question whether "expert designers may be able to use only imagery in the conceptual design phase, before externalizing their design thoughts" (pg.1).
Cognitive psychology has already made links between images and human abilities in mentally manipulating them. "Design research has also questioned whether re-interpretation of images is possible with or without externalization (Verstijnen et al. 1998; Pearson et al., 2001; Kokotovich & Purcell 2001)"(pg.1). However, these experiments have not dealt with ill-structured problem solving processes such as design problems that require unique solutions. "As Pylyshyn (2003) puts it; there is a difference between 'imagining X' and 'imagining that X is the case'."(pg. 1)


Six architects participated in the study, each with more than 15 years of experience and all award winners with senior member status in their respective offices.

- Design of the case study

Three of the architects comprise the first group and they are engaged in the no sketch blindfold condition (BF) to begin with. They are given design brief 01. It requires designing a house for two artists: a painter and a dancer.
One month after the experiment condition the first group engages in another design process where they are allowed to sketch, the control condition (SK). They are given design brief 02. It requires designing a house for a couple with 5 children aged 3 - 17.
Design brief 02 uses the same construction site as design brief 01.
The second group of three architects initially engages in the SK condition and are given design brief 02. A month later they engage in the BF condition and are required to work on brief 01.

BF condition procedure:

  1. Experimenter reads the instructions explaining design activity
  2. Participant engages in think aloud exercise training
  3. Participant is given design brief, which includes site layout, site photos and neighbourhood photos.
  4. The participant is given time to learn the brief until they could recite it back with no mistakes
  5. The participant is told she/he will be expected to come up with an initial sketch for the design
  6. Participant is instructed to put on blindfold and begin thinking aloud while designing. Participant is free to ask questions about the design brief.
  7. The participant is warned when there are 5 minutes left in the session
  8. When the time is up the blindfold is removed and the participant is told to quickly put their design on paper without making any more changes.
  9. The participant is allowed to elaborate this sketch with what he/she had thought up during the session
  10. The participant is interviewed.

SK condition procedure:
First five steps are the same as BF condition.

  1. Participant is asked to think aloud and instructed to begin sketching directly
  2. Five minute warning given.
  3. At the end of the session participant is asked to summarize his/her design.
  4. The participant is interviewed.

- Study Setting

A blank room with no windows, with a digital video recorder and microphones. Designers are provided with pen, tracing paper, scaled site layout, a ruler, and a table where they sit alongside an experimenter that supervises.

- Interviews

Open ended questions, encouraging participants to talk about their experience. Duration varied from 15 minutes to 1 hour.

- Protocol Analysis

Protocols examined in both conditions were segmented by inspecting designer's intentions.

- Coding Schemes

"Based on a cognitive framework which models design thinking as: physical, perceptual, functional, and conceptual actions."(pg. 3) Each architect's sessions were coded by the experimenters, using two coding phases and an arbitration coding phase.

- Sketch Assessment

Three expert judges were selected, each an architect in practice and in teaching with at least 20 years of experience. They were provided with all the design information the participants were and they blind-judged the sketches produced at the end of both BF and SK conditions.

The judges scored the sketches within three criteria:

  1. Satisfies brief.
  2. Innovation.
  3. Practicality.


- Differences in cognitive activity

- Familiarity with design context

The total number of actions in BF sessions was higher than in SK sessions for four of the designers, the other two showed more actions in the SK condition. Since the second group did the SK condition first they were already familiar with the site layout which may have influenced their cognitive tendencies in the BF condition.

- Comparison of design outcomes

Average grades were higher in BF condition for design satisfaction and practicality while innovation showed similar average grades in both conditions.

- Interview results

The first group described frustration at being blindfolded. However, the second group was more satisfied with their design solutions and stated that the blindfolded exercise could become another way of designing for them. This is probably influenced by the order in which the groups engaged in their respective exercise conditions. (BF before or after SK).

- Coding consistency and segment durations

Based on the architects' change of intentions and attention shifts, it can be interpreted that in the BF condition the architects focused on one concept for shorter periods than in the SK condition.


"This case study showed that there are more similarities than differences between SK and BF conditions of the six architects, in terms of the percentage distribution of cognitive action categories and overall design quality. Due to the small scale of the experiment, the results cannot be generalized. Also these results are unlikely to be applicable to later stages of the design activity, since these require intensive drawing and externalization" (pg.5). Even though the quick sketching period was controlled by the experimenter, it might be argued that the reflective conversation begins as soon as the designers start drawing.

- Benefits of external representations

As implied by this study, sketching may not be the only way to design visually. The architects in the BF condition might be creating perceptual units as internal representations and be using these to have their reflective conversation.

- Skilled Imagery

"The results imply that without any visual feedback during the BF condition, architects relied on using/retrieving the visual and spatial information from their long term memories. Similar to expert chess players, expert designers could have pre-existing dynamic chunks of visual features or spatial relations encoded in their LTM"(pg. 6).


"Design reasoning via constructing internal representations may be as efficient as reasoning via constructing external representations for expert architects, in the limited period of the conceptual design phase"(pg. 6).

Summary author's notes:

Back to the Cognitive Science Summaries homepage
Cognitive Science Summaries Webmaster:
JimDavies (jim@jimdavies.org)