Sketching and creative discovery
I M Verstijnen and J M Hennessey, Delft University of Technology,
Faculty of Industrial Design Engineering, Jaffalaan 1, 2628 BX Delft,
The Netherlands
C van Leeuwen and R Hamel, University of Amsterdam, Department of
Psychology, Roetersstraat 15, 1018 WB Amsterdam, The Netherlands
G Goldschmidt, Technion-Israel Institute of Technology, Faculty of
Architecture and Town Planning, Technion Haifa, Israel 32000.
In the search for helpful computer tools for sketching in the early
phases of design, the approach was taken to experimentally study
sketching behaviour. In two series of experiments two mental processes
revealed themselves as essential in the creative process: Restructuring
and Combining. These two processes are in turn influenced by expertise
in sketching and individual creativity. In this article each of the factors:
Combining, Restructuring, Expertise and Creativity, will be separately
highlighted with respect to their impact on sketching behavior. Finally,
on the basis of these results conclusions are drawn for computerized
sketching aids.  1998 Elsevier Science Ltd. All rights reserved
Keywords: drawing, sketching, creativity, design cognition, computer
supported design
O
ne day, the story goes, Archimedes jumped out of the bath and
ran joyfully and nakedly through the streets of Syracuse. Finally,
he had discovered the solution to a problem which had been bothering him. For a long time, he had wondered how to measure the volume
of an irregular object. Obviously, the discovery came to him at a moment
he had not anticipated. This aspect of discovery is more frequently encountered in anecdotal evidence and self-reports. Friedrich von Kekulé reports
drowsing in front of his fireplace when suddenly the solution, to a problem
he had been working on, showed up. The image he saw in the flames about
snakes biting their own tails hinted him to drop an assumption in organic
chemistry, which prescribed that organic molecules must be strings of carbon atoms. This enabled him to come up with the idea of a ring structure
for the benzene molecule.
Kekulé’s discovery came to him in a state of reduced awareness, but this
appears not to be the rule. Henri Poincaré was entering a bus which was
0142-694X/98 $—see front matter Design Studies 19 (1998) 519–546
PII: S0142-694X(98)00017-9
 1998 Elsevier Science Ltd All rights reserved Printed in Great Britain
DST: design studies (page 1
)
08-10-98 10:42:42
Rev 14.02x
519
zdst$$140h
to take him on a geological trip, when suddenly he saw a fundamental
property, uniting a hitherto unrelated group of mathematical functions.
These, and several other reports of scientific discoveries are frequently
reported1–3. All the anecdotes on creative discovery have the unanticipated
character of the discovery in common.
1 Dreistadt, R ‘The use of analogies and incubation in creative
problem solving’ Journal of Psychology Vol 71 (1968) pp 159–
175
2 Miller, A I Imagery in scientific thought MIT Press, Cambridge, MA (1986)
3 Rothenberg, A The emerging
goddess University of Chicago
Press (1979)
4 Goldschmidt, G ‘Serial
sketching: Visual problem solving in designing’ Cybernetics and
Systems Vol 23 (1992) pp 191–
219
5 Hennessey, J M The IDEATE
project: ‘Exploring computer
enhancements for conceptualizing’, in Automation based creative design: Current issues in
computers and architecture, T
White and A Tzonis (eds)
Elsevier Science, Amsterdam
(1994) pp 349–362
6 Kolli, R and Hennessey, J M
Deriving the functional requirements for a concept sketching
device: A case study, Proceedings of Vienna conference on
Human-Computer
Interaction
Springer Verlag (1993) pp 184–
195
7 Stuyver, R and Hennessey,
J M A support tool for the conceptual phase of design, Proceedings of the Human-Computer Interaction Conference
Cambridge University Press
(1995) pp 235–245
8 Fish, J and Scrivener, S
‘Amplifying the mind’s eye:
Sketching and visual cognition’
Leonardo Vol 23 No 1 (1990)
pp 117–126
9 Verstijnen, I M, Stuyver, R,
Hennessey, J M, van Leeuwen,
C and Hamel, R Consideration
for electronic idea-creation tools,
Companion of CHI’96, Vancouver, Canada Addison Wesley
(1996) pp 197–198
The unanticipated character of discovery makes it unlikely that these creative individuals had paper and pencil available to support the breakthrough
in their thinking with sketching. In the light of this evidence, it may seem
slightly audacious to suggest that sketching could have an important role
in discovery. Yet, this is what is frequently reported by artists and designers, and what will be concluded from our studies.
That externalization fulfils a need, is suggested by the fact that most artists
and designers use some kind of externalization, e.g. sketching and clay
modeling. They consider this essential for their creative process and will
report frustration if hindered in doing so. Sometimes, people find their way
around this frustration and this is why sketches can be found on backs of
envelopes, edges of newspapers, or napkins.
Designers have called sketches made in such a situation idea-sketches. In
contrast to presentation-sketches, idea-sketches are made in the early
phases of design. They function as a tool to interact with imagery4 and are
predominantly for private use. Because of their early appearance in the
design process, idea-sketching will have an important role in creative processes. This is the reason why many computer tools aim at supporting and
improving idea-sketching.
The IDEATE-project5–7 is one of the projects which acknowledges the
beneficial aspects of computer tools for supporting idea-sketches. This project envisages the development of a tool, the IDEATOR. The IDEATOR
will contain all facilities that are needed for storing sketches in memory,
copying and pasting, displaying imported pictures or video’s on the sketch
surface. But in addition, the IDEATOR may have to include a set of
additional facilities, based on insight in the drive or motivation behind
idea-sketches.
In order to include such facilities, the drive or motivation behind ideasketches has to be studied more closely8,9. The unpredictability of ideasketching behavior is a challenge to such study. The private nature of these
sketches, their proverbial appearance on the back of an envelope, the
absorption of the sketcher in his or her activities, makes idea-sketching a
fascinating topic, which has been an impetus, both for self-reflexive activi-
520
DST: design studies (page 2
Design Studies Vol 19 No 4 October 1998
)
08-10-98 10:42:42
Rev 14.02x
zdst$$140h
10 Garner, SW ‘Drawing and
designing: the case for a reappraisal’ Art and Design Education Vol 9 No 1 (1990)
pp 39–55
11 Goldschmidt, G Linkography: ‘Assessing design productivity’, in Cybernetics and Systems ’90, R Trappl (ed) World
Scientific (1990) pp 291–298
12 Schenk, P ‘The role of
drawing in the graphic design
process’ Design Studies Vol 12
No 3 (1991) pp 168–181
13 Verstijnen, IM and Van
Leeuwen, C ‘A book review of
Imagery, creativity, and discovery: A cognitive perspective,
B Roskos-Ewoldsen, M IntonsPeterson and R Anderson
(eds) (North-Holland, Amsterdam 1993)’ Acta Psychologica
Vol 89 (1995) pp 293–295
14 Anderson, R E and
Helstrup, T ‘Multiple perspectives on discovery and creativity
in mind and on paper’, in Imagery, creativity, and discovery: A
cognitive
perspective,
B
Roskos-Ewoldson, M J IntonsPeterson and R E Anderson
(eds) Elsevier Science Publishers (1993) pp 223–253
15 Anderson,
RE
and
Helstrup, T ‘Visual discovery in
mind and on paper’ Memory and
Cognition
Vol
21
(1993)
pp 283–293
16 Metcalfe, J ‘Feeling of
knowing in memory and problem
solving’ Journal of Experimental
Psychology: Learning, Memory
and Cognition Vol 12 (1986)
pp 288–294
17 Metcalfe, J and Wiebe, D
Intuition in insight and noninsight
problem solving Memory and
Cognition (1987) pp 238–246
18 Wallas, G The Art of
Thought Harcourt Brace, New
York (1926)
19 Hadamard, J The psychology of invention in the mathematical field Princeton University Press, Princeton, NJ (1945)
20 Koestler, A The Act of Creation London (1975)
21 Poincaré, H La science et
méthode Flammarion, Paris
(1908)
22 Solso, R L Cognitive psychology (3rd ed) Allyn and
Bacon, Boston (1991)
23 Boden, M The creative
mind: Myths and mechanisms
Weidenfeld and Nicolson (1990)
24 Rothenberg, A and Hausman, C R The creativity question
Duke University Press, Durham,
NC (1976)
ties of expert sketchers such as designers themselves, and for experimental
studies by psychologists.
1
Sketching and mental imagery
Until now, paper and pencil sketching has been discussed predominantly
among those who are themselves actively engaged in sketching, such as
industrial designers and architects. Their descriptions range from selfreports10 on the one hand, to objective descriptions of the sketch process4,11,12 on the other. This literature concentrates on the sketching activities themselves; the mental processes, from which the sketching behavior
originates are still to be investigated. Psychologists have only recently
touched upon this issue13.
Anderson and Helstrup14,15 were the first within the information processing
framework of psychology to systematically explore the field of idea sketching. Their seminal proposal was, that the decision to sketch is made when
creative processing is met with resource limitations. In their experiments,
however, they failed to obtain support for this view. This still leaves unexplained why idea-sketching could be useful.
The question, why artists and designers need externalization for discovery
in a creative process can be approached in a variety of ways. The most
direct way appears to be the introspective way. Many artists and designers
have studied their own creative processes and have launched a great variety
of ideas about the how, what and why of sketching. This ranges from casual
observations to systematic description10. Introspective methods formed in
the 19th century are the basis for the scientific revolution that led to the
study of mental processes in psychology.
Many psychologists, however, have questioned the validity of introspection
as the most objective way of studying mental processes. Objections to the
use of this method have been raised in particular for the type of discoveries
presently discussed. Metcalfe16,17 argues that the time period preceding
such a discovery (the incubation period18–22) is not open to introspection.
The wide variety and lack of systematicity in self-reports may reflect these
problems. It is possible that these self-reports reflect whichever ideology
the artist happens to believe. In an attempt to characterize these beliefs,
Boden23 distinguished two types: inspirational beliefs, reflecting a Platonic
notion of insight as the discovery of an eternal idea24, and romantic beliefs,
stressing the uniqueness of artistic and scientific talents.
The scorned method of introspection is not to be confused with another
method which has been considered useful for the study of creative dis-
Sketching and creative discovery
DST: design studies (page 3
)
521
08-10-98 10:42:42
Rev 14.02x
zdst$$140h
covery processes. This method collects and analyzes think-aloud protocols.
Think-aloud protocols are self-reports which differ in an important way
from introspection. A think-aloud protocol is collected during a process of
creative discovery instead of afterwards. Therefore, the protocol is a direct
verbalization of the ongoing process, in which self-reflexive expressions
are discouraged.
25 Shepard, R N ‘Externalization of mental images and the
act of creation’, in Visual learning, thinking and communication,
B S Randhawa and W E
Coffman (eds) Academic Press,
NY (1978) pp 133–189
26 Ericsson, KA and Simon,
HA ‘Verbal reports as data’
Psychological Review Vol 87
(1980) pp 215–251
27 Elshout, J and van
Leeuwen, C ‘Protocolanalyse
als Methode van Kennisacquisitie
voor
Expertsystemen
(Protocol Analysis as Method for
Knowledge Acquisition in Expert
Systems)’, in Kennis in organisaties: Theorie en toepassing
van kennissystemen, R Jorna
and J Simons (eds) (Knowledge
in organizations: Theory and
applications of expert systems).
Muiderberg, Coutinho (in dutch)
(1992)
28 Brandimonte, M A, Hitch,
G J and Gabbino, P Release
from verbal overshadowing.
Paper presented at the European Society for Cognitive Psychology Conference, Copenhagen,
Sweden
(1993
September)
29 Schooler, J W, Fallshore,
M and Fiore, S M ‘Putting insight
into perspective’, in The nature
of insight, R J Sternberg and J
E Davidson (eds) MIT Press,
Cambridge, MA (1995)
30 Wilson, TD, Lisle, DJ,
Schooler, JW, Hodges, SD,
Klaaren, KJ and Lafleur, SJ
‘Introspection about reasons can
reduce post-choice satisfaction’
Personality and Social Psychology Bulletin Vol 19 (1991)
pp 331–339
This latter approach suffers from the fact that most creative processes
extensively make use of visual thinking, or, in other words, there is a strong
contribution of visual imagery. These processes are not accessible to direct
verbalization. As Shepard25 put it: ‘... it seems reasonable that the most
novel ideas and radical departures from traditional ways of thinking are
not likely to arise within the very system of verbal communication that is
the primary vehicle for maintaining and perpetuating established ideas and
entrenched traditions’ (p. 156). Protocol analysis and introspection, therefore, are not appropriate to reveal this particular kind of processes26,27 and
might even obstruct them28–30.
The third way of approaching the question is the experimental method.
This approach cannot easily address the problem directly, as creative processes are notoriously unpredictable19. For this reason, the experimental
method attacks the question by inverting the problem. For instance, for the
problem what sketching contributes to the creative process, instead of raising the question why sketching is useful, the question is raised, what limitations do the mental processes have that require sketching? This strategy
is employed in the present study.
The search for limitations in processing has been pioneered by Anderson
and Helstrup14,15. They proposed resource limitations to be the decisive
impetus for sketching. Resource limitations, for example, are encountered
when memory fails to keep track of a growing information load. In their
experiments, however, these authors failed to obtain support for a crucial
role of resource limitations in sketching.
An alternative approach has been taken in the present study. This approach
rests on the idea that, since imagery plays an important role in the creative
process, the limitations are to be found in imagery. In this view, the mental
operations leading to discovery are viewed as a set of operations on a
mental image. Sketching is needed if the operations cannot be done within
mental imagery alone, or if the operations are much easier to perform externally.
This approach draws the attention away from the externalization itself and
focuses on the mental processes preceding the sketching activity. Once
522
DST: design studies (page 4
Design Studies Vol 19 No 4 October 1998
)
08-10-98 10:42:42
Rev 14.02x
zdst$$140h
these processes are understood, this will lead to insight in the origin and
function of externalization.
The psychological literature on discovery in mental imagery has studied
several different mental imagery processes. Some of these are easy and
frequent; others are difficult and rare. The latter type being candidates for
externalization support. The fact that some processes in mental imagery
were found to be difficult and others were easy, has led to a controversy
in the literature before it was realized that two different types of processes
may exist. Opposing conclusions were drawn on the one hand by Chambers
and Reisberg31, and Reed32 and on the other hand by Finke and Slayton33.
31 Chambers, D and Reisberg, D ‘Can mental images be
ambiguous?’ Journal of Experimental Psychology: Human Perception and Performance Vol 11
(1985) pp 317–328
32 Reed, SK ‘Structural
descriptions and the limitation of
visual images’ Memory and Cognition Vol 2 (1974) pp 329–336
33 Finke, RA and Slayton, K
‘Explorations of creative visual
synthesis in mental imagery’
Memory and Cognition Vol 16 No
3 (1988) pp 252–257
34 Hyman, I E and Neisser, U
Reconstruing mental images:
Problems of method Emory Cognition Report, #19, Emory University, Atlanta, GA (1991)
35 Peterson, M, Kihlstrom, J,
Rose, P and Glisky, M ‘Mental
images can be ambiguous:
Parts, wholes, and strategies’
Memory and Cognition Vol 20
(1992) pp 107–223
36 Reed, SK and Johnsen, JA
‘Detection of parts in patterns
and images’ Memory and Cognition Vol 3 (1975) pp 569–575
Chambers and Reisberg showed in their experiments that subjects were
unable to reverse the interpretation of an ambiguous figure (e.g. chef/dog,
see Figure 1.) in mental imagery. That is, when the figure was presented
to the subjects as representing a dog, they were unable to discover the
alternative interpretation of a chef before their mental eye, while, on the
other hand, this discovery took place easily when subjects were allowed
to visually inspect the figure. Subsequent investigations34,35 with a similar
paradigm did not find completely zero performance, but discovery performances remained strikingly low. Reversal involves the discovery of an unanticipated structure. Chambers and Reisberg’s results, therefore, argue for
limitations on discovery in mental imagery, as compared to visual perception, and hence for a major role for sketching in discovery.
Similar restrictions on mental imagery were obtained by Reed and
Johnsen36. They showed that the extraction of an unanticipated novel
component is difficult in imagery, as compared to visual perception. These
authors used a part-whole detection task (see Figure 2). In this task, a
composite figure is presented, followed by another one. The subject of the
experiment must decide if the second is contained as a part of the first.
Percentages of correct answers and reaction times reflect the accuracy of
processing.
Figure 1 An example of an
ambiguous figure. The figure
could be interpreted either
as a dog or as a chef, and
was used as stimulus by
Chambers and Reisberg31 to
show that reversal in mental
imagery meets considerable
difficulties
Sketching and creative discovery
DST: design studies (page 5
)
523
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
Figure 2 An example of a
configuration and several
alternative interpretations of
its components. Reed and
Johnsen36 showed that when
the configration is remembered as two overlapping
triangles (A) an alternative
composition (B to D) is very
difficult to reveal in mental
imagery
With these tasks, it can be shown for perception that certain parts are more
easily perceived and remembered as components of the whole than others.
This depends on which parts are seen as building blocks of the structure
of the whole. Usually, a whole is perceived as a unique composition of
parts, this, although it is possible in principle to see the whole as composed
in several alternative ways as well (see Figure 2 on the right). Reed and
Johnsen showed that an alternative composition was not recognized in
mental imagery.
Chambers and Reisberg used a task, which involved imposing a new structure on old components of a figure. Reed and Johnsen involved the identification of new components in a structure. Both experiments, therefore,
involve conditions in which subjects start from a given, structured pattern.
The task requires them to break up this structure. From these experiments,
the hypothesis can be drawn that mental imagery faces considerable difficulty in restructuring the initial conception of a pattern, and hence to
discover new information in a mental image.
Opposite conclusions on the issue of discovery in mental imagery were
reported by Finke33,37. These authors asked subjects to synthesize simple
elements into a recognizable object before their inner eye. For example, a
letter ‘J’ and a letter ‘D’ can form an umbrella. They found that subjects
were able to create and discover new unanticipated objects.
37 Finke, R A Creative Imagery: Discoveries and inventions
in
visualization
Erlbaum,
Hillsdale, NJ (1990)
It could be observed, however, that Finke’s task does not require restructuring. The structure of the elements as initially given remains intact. In the
524
DST: design studies (page 6
Design Studies Vol 19 No 4 October 1998
)
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
umbrella example this means that the component structure of each letter
‘J’ and ‘D’ is not violated. Therefore, the processes involved in Finke’s
Figural Combination Tasks differ essentially from those involving restructuring. The processes observed by Finke, which are easily performed in
mental imagery, are processes of combining.
Therefore the opposing conclusions from the mental imagery literature can
be reconciled on the assumption that two forms of processing in imagery
have to be distinguished, viz. restructuring and combining. The two different experimental paradigms, familiar from the literature, involve restructuring and combining processes to different degrees.
If the processes of combining and restructuring impose different loads on
mental imagery, different effects of paper-and-pencil support can be
expected. If the mental imagery task is easy, as in the Figural Combination
Task (Finke), minor effects are expected. If the mental imagery task is
difficult because of restructuring, sketching is expected to enhance performance in the Component Detection Task (Reed).
In two series of experiments this hypothesis was tested. The first series
used a Component Detection Task38. The second series used a Figural
Combination Task39. Each series will be discussed separately in the following sections with their associated theoretical processing characteristics
(restructuring or combining). Expertise and creativity were shown to have
differential impact on these two processes. Therefore, the relation of
sketching to each of the factors: restructuring, combining, expertise and
creativity, will be highlighted in separate sections.
2
38 Verstijnen, I M, van
Leeuwen, C, Hamel, R and
Hennessey, J M What imagery
can’t do and why sketching
might help (submitted-a)
39 Verstijnen, I M, Goldschmidt, G, van Leeuwen, C,
Hamel, R and Hennessey, J M
Discovery in imagery; synthesis
can be done but analysis takes
a sketch (submitted-b)
Sketching and restructuring
In this section and the following one on sketching and combining, only
the results of expert sketchers are considered, results concerning novice
sketchers will be dealt with in the section on sketching and expertise. The
experts in the experiments were all industrial design engineering students
having had at least two years of drawing education. The term ‘experts’ is
used to stress the difference with ‘novices’, but is not meant to be interpreted as designating highly experienced sketchers. Vice versa, the novices,
drawn from a population of first year undergraduate psychology students,
are not expected to lack drawing skills completely. A comparison between
novices and experts can be found in the section on sketching and expertise.
2.1
Restructuring in the component detection paradigm
The first series of experiments was inspired on the work of Reed32,36 These
experiments were performed with overlapping forms of which a novel
Sketching and creative discovery
DST: design studies (page 7
)
525
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
component had to be extracted. For this reason, the task was called a
Component Detection Task. Prior to the experiment, subjects were informed about a set of components used. These were simple 2D wire-frame
elements (e.g. square and diamond). During the experiment, configurations
of overlapping components were presented, one at a time. Each configuration was briefly presented to the subjects. The short presentation time
ensured that people were not able to grasp more information from the
configuration than the identity of the components and their relative position. Hence, Figure 3 will be remembered as slanted square, and diamond
with their lower edges aligned40.
The task was to answer the question whether the configuration contains a
subsequently presented figure, as a part. Some of these figures were indeed
contained as parts in the configuration and others were not (the false parts).
Among the parts, a distinction was made between existing ones, which
were components from which the figure was constructed, and novel ones,
which were not used in the construction but resulted from the overlap
between two or more components (see Figure 3).
40 Clark, HH and Chase, WG
‘On the process of comparing
sentences against pictures’ Cognitive Psychology Vol 3 (1972)
pp 472–517
The difference between novel and existing parts is relevant to the issue of
restructuring. Existing parts can be extracted from the configuration without restructuring, however, restructuring is to be performed if an novel part
is to be detected. This is the case, because the configuration is remembered
as a combination of the original components. Restructuring was therefore
measured by the percentage correct on novel parts. The percentage correct
of existing parts serves as a baseline, to determine the accuracy of performance without restructuring.
Three variations of this experimental paradigm were administered; the first
compared a group of subjects compelled to sketch with a group which was
Figure 3 An example of a
configuration used in the
Component Detection Task.
Detection of novel parts
requires
restructuring,
because this figure will be
remembered as a configuration consisting of a square
and a diamond. No restructuring is necessary to discover the existing part
526
DST: design studies (page 8
Design Studies Vol 19 No 4 October 1998
)
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
denied this strategy; the second experiment registered spontaneous sketch
behavior; in the third the number of elements in a configuration were
varied. A spontaneous sketching experiment approaches more closely the
circumstances in which people normally turn to sketch.
The first experiment with this paradigm, allowed half the group of subjects
to sketch (the with-sketch condition), and denied this strategy to the other
half (the without-sketch condition). Figure 4 shows the results for both
groups. Subjects who were allowed to sketch performed significantly better
on restructuring, i.e. finding novel parts, than their counterparts in the condition where sketching was not allowed. The latter performed no better
than chance-level (i.e. 50%).
41 Jacoby, L and Dallas, M
‘On the relationship between
autobiographical memory and
perceptual learning’ Journal of
Experimental Psychology: General Vol 3 (1981) pp 306–340
42 Jacoby, L The relationship
between learning and recollection: Memory attributes vs. Memory attributions Paper presented
at the Annual Meeting of the
Psychonomic Society, Boston,
Mass (1985)
Figure 4 Percentages
rect
detection
of
These findings contrast with those for existing parts, see Figure 5. On
existing parts both the without-and the with-sketch group scored equally
well. Although a slight surplus on existing parts can be noted for the withsketch group with respect to the without-sketch group, this difference is
not significant. This surplus, which was noted for both experts and novices,
could be explained on the basis of false parts only. Sketching in the case
of false parts could have been helpful to inform the sketcher on whether
he/she forgot this particular part or on whether it did not form an element
of the configuration, and thus was indeed a false part. Sketching in this
case performed a similar role as reported in the Jacoby and Dallas41,42
studies, in which they showed that previously learned, but forgotten, words
look more familiar.
cornovel
components on the Component Detection Task. Subjects not allowed to sketch
(without-sketch
condition)
act on chance level (50%),
whereas subjects compelled
to
sketch
(with-sketch
condition) raised their performance to a level significantly above chance
Sketching and creative discovery
DST: design studies (page 9
)
527
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
Figure 5 With- and without
sketch conditions in the
Component Detection Task
do not differ on parts not
requiring restructuring, i.e.
existing parts
Figure 6 illustrates the scores of the with-sketch group on novel and existing parts. Although a difference exists, this difference is not significant,
and illustrates that performance on novel parts can be raised to the level
of performance on existing parts with the aid of sketching. This is in contrast with the without-sketch group.
Figure 6 Expert
sketchers
are able to raise their performance on novel parts in
the Component Detection
Task to the level of existing
parts
528
DST: design studies (page 10 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
Similar phenomena were observed when spontaneous sketching was compared to compelled sketching. Performance on novel parts was significantly
improved by spontaneous sketching (a significant correlation of r = 0.715),
whereas sketching did not improve performance on existing parts (r =
⫺0.034).
In the compelled sketching experiment, therefore, it was concluded that
people cannot distinguish an novel part from a nonpart when not allowed
to sketch. The same could be concluded from the spontaneous-sketching
experiments. Equal numbers of subjects turned to sketch spontaneously on
novel parts as on false parts. This result demonstrates that prior to sketching, people barely have an idea of the occurrence of an novel part.
To investigate individual correlates of the restructuring ability, various tests
were administered in the spontaneous sketch experiment: the Raven43 progressive matrices test, measuring intelligence, Witkin’s44 embedded figures
test, Mark’s45 vividness of visual imagery questionnaire (VVIQ), Kunzendorf’s46 aesthetic preference test, measuring creativity, and a measure of
short-term memory-span.
43 Raven, J C Advanced Progressive Matrices Set II (1988).
44 Witkin, H A, Oltman, P K,
Raskin, E and Karp, S A A manual for the embedded figures
tests Consulting Psychologists
Press (1971)
45 Marks, DF ‘Visual imagery
differences in the recall of pictures’ British Journal of Psychology Vol 64 (1973) pp 17–24
46 Kunzendorf, RG ‘Mental
images, appreciation of grammatical patterns, and creativity’
Journal of Mental Imagery Vol 6
No 1 (1982) pp 183–201
47 Chara, PJ and Hamm, DA
‘An inquiry into the construct validity of the Vividness of Visual
Imagery Questionnaire’ Perceptual and Motor Skills Vol 69
(1989) pp 127–136
48 Reisberg, D and Heuer, F
‘Vividness, vagueness and the
quantification of visualizing’
Journal of Mental Imagery Vol 12
No 3 and 4 (1988) pp 89–102
49 Morris, P and Hampson, P
Imagery and consciousness
Academic Press, New York
(1983)
Restructuring (novel part detection) correlated with intelligence, and creativity (this latter correlation will be dealt with in the section on restructuring and creativity). Accuracy of performance without restructuring
(existing part detection) correlated with intelligence, memory span, and
embedded-figure detection. Vividness of imagery, as measured with
Mark’s VVIQ, did not correlate with any of the variables. This result can
be attributed to the lack of validity in measures of vividness47,48. It may
also be the case that vividness of imagery is more important for inspection
of imagery than for manipulation and transformation of images49.
Factor Analysis on the above correlations showed that intelligence, shortterm memory span, embedded-figure detection and the experimental measure of accuracy-without-restructuring (existing parts) are associated measures. The abundant representation of ‘smart’ variables in this factor, led to
its name: the Smart Factor. Amount of sketching on novel and existing
parts, creativity, and the experimental measure of restructuring (novel part
detection) formed a second, independent set of associated measures. This
factor was called the Art Factor, because of the occurrence of ‘arty’ variables. The Art Factor was shown to be about 5 times as important for
restructuring than the Smart Factor39.
One of the constituents of the Smart Factor is memory capacity. The fact
that the Smart Factor contributes only little to restructuring, suggests that
Sketching and creative discovery
DST: design studies (page 11 )
529
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
memory capacity is not a relevant determinant of the difficulty of restructuring in mental imagery and therefore irrelevant for the motivation to
sketch. This finding contradicts the intuitive idea that limitations in shortterm memory trigger sketching behavior50. To perform a more rigid test
on this idea, a third experiment was performed, that focused on memory.
In this experiment, the number of elements of a configuration was varied.
If short-term memory-span is irrelevant for sketching behavior, then, it was
hypothesized, no change in performance is expected when the number of
components in a configuration is raised51. Moreover, if novel parts are
used, the accuracy of detection will not depend on the number of components. Whether two or five components were presented, the number of
spontaneous sketches was unaffected by the number of components. Also,
accuracy of performance was independent of the number of components.
It was, therefore, concluded that memory limitations are irrelevant for
sketching behavior.
Many artists and designers, when asked for their motivation, ascribe a
function of memory extension to their sketching behavior. The results of
the present experiments contradict these introspective reports, instead they
confirm the conclusion of the previous experiment, that memory restrictions do not form the motivation for sketching. It cannot be excluded,
however, that sketches may fulfil a role for retention on a longer term than
the scope of the present experiments52. In general, however, it may be
concluded for the Component Detection Task, that the extraction of novel
components yields the decisive impetus for sketching, because this form
of restructuring is difficult to perform in mental imagery.
2.2
50 Ullman, DG, Wood, S and
Craig, D ‘The importance of
drawing in the mechanical
design process’ Comput. and
Graphics Vol 14 No 2 (1990)
pp 263–274
51 Cavanagh, JP ‘Relation
between the immediate memory
span and the memory search
rate’ Psychological Review Vol
79 No 6 (1972) pp 525–530
52 Reisberg, D ‘External representations and the advantages
of externalizing one’s thoughts’,
in The ninth annual conference
of the cognitive society, E Hunt
(ed) Erlbaum, Hillsdale, NJ
(1985) pp 281–293
Restructuring in the figural combination paradigm
The second series of experiments used a modified version of Finke and
Slayton’s figural combination paradigm. Before the experiment started, six
possible components (cone, cube, cylinder, sphere, diamond, flat-square)
were presented. Subjects were instructed to memorize them in combination
with their names. During the experiment only the names of three components were presented at a time. Each time this occurred, subjects had to
combine in mental imagery three simple components into an object. Half
of the subjects were allowed to sketch during the mental imagery phase,
the other half was denied this strategy. The resulting objects were drawn
on a separate sheet of paper and rated on creativity, originality and practicality by independent judges. The rated creativity of the end-products of
with-sketch and without-sketch conditions was compared.
It turned out that the creativity of the end-products was independent of
strategy. This result was in accordance with the earlier conclusion by And-
530
DST: design studies (page 12 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
erson and Helstrup14,15, who observed that discovery in mental imagery
was easy and frequent. According to the hypothesis, the reason is that the
Figural Combination Task is predominantly a combining task. Nevertheless, an investigation of whether restructuring occurred as well as performed. For this purpose, in addition to the ratings, the objects were also
scored on other properties.
For example, when a cube was one of the components presented, it was
regularly found to be transformed into a rectangular box. In Figure 7 (on
the right side) the diamond which originally was presented as having a
squared cross-section is transformed into one with a triangle as cross section. These kinds of transformations were considered to reflect restructuring. Various kinds of these component transformations were scored. From
those transformations, a restructuring measure was computed. The restructuring measure reflects the amount of structural transformations performed
on the components. Individual subjects were scored on this measure.
An other example of restructuring in the figural combination paradigm
shows in the occurrence of junctions between components. It may require
restructuring to form complex junctions between combined objects. Complex junctions are defined as those junctions, which require transformations
of the components, in order to perform a fit; if one component has to
fit to another, it is sometimes needed to perform a transformation of the
components. For that reason, complex junctions are indicative of restructuring. Interestingly, in the without-sketch conditions, the number of complex
junctions of 3D differed not from that in the with-sketch conditions, when
averaged across all situations. This measure, however, included complex-
Figure 7 Two examples of
objects created in the figural
combination
experiments.
The figure on the left is a
shower for a swimmingpool. The ball has to be
pushed up to get a shower.
This being hard, the shower
will save water. On the right
is a tool created for windowcleaners. It can be suctioned
to the wall, and be used as
a grip or a foot-rest
Sketching and creative discovery
DST: design studies (page 13 )
531
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
junctions hidden in the back of the object. When only visible complex
junctions were considered, their number in the with-sketch condition was
significantly above that of the without-sketch condition. It may be concluded that the subjects in the without-sketch condition reach equally creative end-products, because they have a way around restructuring, such
as turning the complex junction to the back, when denied the possibility
to sketch.
The mean restructuring score for the with-sketch group was significantly
higher than that of the without-sketch group (see Figure 8a: a restructuring
score of for example 2 denotes that for each subject the created objects
had an average of two restructurings of whatever type per object). Thus,
even though the products were equally creative, they were achieved by the
without-sketch and with-sketch groups in different ways: sketching subjects make use of restructuring strategies more frequently than non-sketching subjects. When Finke’s original set of 2D elements (e.g. square, triangle, circle) was used similar results for restructuring were found (see
Figure 8b.)
The method to score restructuring in the present task differs from that of
the Component Detection Task. In order to confirm that both tasks measure
the same ability, it could be considered useful to compare individual scores
across the experiments. Eleven subjects took part in the with-sketch groups
of both the Component Detection Task and the Figural Combination Task.
This made a check for the validity of the restructuring measure in the
Figural Combination Task possible. A correlation was calculated between
subjects’ score on novel parts in the Component Detection Task and their
restructuring score in the Figural Combination Task. A highly significant
correlation was found. The correlation shows that restructuring is a personal characteristic, which is consistent over tasks and stable over time.
This is an argument in favor of the validity of the measures of restructuring
in the two series of experiments and, in general, for the validity of the
construct of restructuring.
53 Hinton, GE and Parsons,
LM ‘Scene-based and viewercentered representations for
comparing shapes’ Cognition Vol
30 No 1 (1988) pp 1–35
54 Pylyshyn, Z ‘What the
mind’s eye tells the mind’s brain:
A critique of mental imagery’
Psychological Bulletin Vol 80
(1973) pp 1–24
55 Pylyshyn, Z ‘The imagery
debate: Analogue media versus
tacit knowledge’ Psychological
Review Vol 88 No 1 (1981)
pp 16–45
Based on the combined results of both series of experiments, it can be
concluded that mental images are not inspectable in the same ways as
pictures53–55. The inability to perform restructuring in mental imagery constitutes a major factor in the usefulness of sketching. Not only severe
restructuring, such as novel decomposition, is supported by sketching, also
lighter forms of restructuring, such as occurred in the Figural Combination
Task, are improved by externalization. When mental images are projected
in sketches, new structures can be seen in the sketches which could not
be obtained from the mental images before the projection.
532
DST: design studies (page 14 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
3
Sketching and combining
In contrast to Finke’s original version of the Figural Combination Task,
the one used in the current experiment was especially designed to test the
influence of sketching on combining. Combining is reflected in the figural
combination paradigm in the alignment of components of an object. Various forms of alignment are possible. For example, an object could have
been created by a simple stacking of the components (Figure 7 on the
Figure 8 Restructuring
scores of the Figural Combination Task as a function of
whether the subjects were
not allowed to sketch or
were compelled to do so in
conditions using 2D or 3D
components. In both the 2D
and 3D task, expertized sketchers are able to raise their
performance when allowed
to sketch
Sketching and creative discovery
DST: design studies (page 15 )
533
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
left) or one component could have been aligned horizontally and another
vertically (see Figure 7 on the right). Since each subject created multiple
objects, different forms of alignment used by a subject could be scored.
This score was considered to reflect the capacity to generate variety by
combining. This score was therefore considered a subject’s combining
score. According to this measure, subjects in the without-sketch condition
of the 3D task performed equal to those in the with-sketch condition
(Figure 9; a combining score for example 4 means that a subject applied
four different forms of alignment).
56 Finke, RA and Kosslyn,
SM ‘Mental imagery acuity in the
peripheral visual field’ Journal of
Experimental
Psychology,
Human Perception and Performance Vol 6 (1980) pp 126–139
57 Finke, RA and Kurtzman,
H ‘Mapping the visual field in
mental imagery’ Journal of
Experimental Psychology: General Vol 110 (1981) pp 501–517
58 Pinker, S and Finke, R
‘Emergent two dimensional patterns in images in depth’ Quarterly Journal of Experimental
Psychology Vol 29 (1980)
pp 117–133
59 Richardson, J T E Mental
imagery and human memory St.
Martin’s Press, New York (1980)
60 Spearman, C Creative mind
Appleton, New York (1931)
A slightly different combining measure was used for the 2D version of the
task. This measure, which was adopted from Anderson and Helstrup14,15
scored whether components were aligned in different orientations than
initially presented. This measure was considered also to reflect combining,
because this transformation leaves the component structure intact
(therefore, no restructuring is involved) and the orientation of the components is prescribed by the configuration. On this measure, the sketching
significantly deteriorated performance (Figure 10).
On two different measures, equal or even deteriorated performances was
reached as a result of sketching. This signifies that mental imagery is thoroughly capable of combining. Combining thus appears to be easy in mental
imagery, and new information can be glanced effortlessly from a combined
product, as was to be expected on the basis of experiments by Finke56–60.
Or, as Spearman60 put it in his Principle of Relations: ‘When two or more
Figure 9 Results of the 3D
version of the Figural Combination Task: performance
on a combining measure for
the without-sketch and withsketch condition
534
DST: design studies (page 16 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
Figure 10 Results of the 2D
version of the Figural Combination Task: performance
on a combining measure for
the without-sketch and withsketch condition
items (percepts or ideas) are given, a person may perceive them to be
related in various ways’ (in Brown61).
It can be concluded that no additional value is obtained from sketching if
combining is to be performed. The fact that sketching can even deteriorate
performance, as was the case in the 2D task, suggests that being compelled
to sketch in conditions where it is not necessary will only distract the subject.
4
Sketching and expertise
In the previous sections the results of expert sketchers (design students
with at least two years of sketching experience) were described. However,
in the majority of experiments they were compared to novice sketchers
(psychology students with little sketching experience).
4.1
61 Brown, R T ‘Creativity: what
are we to measure?’, in Handbook of creativity: Perspectives
on individual differences, J A
Glover, R R Ronning and C R
Reynolds (eds) Plenum Press,
New York (1989) pp 3–32
Expertise in the component detection paradigm
Only one major difference was obtained between experts and novices in
the experiment with compelled sketching. On novel parts in the with-sketch
condition, novices performed significantly worse than experts (see Figure
11). In all other conditions, performance was equal. Sketching has helped
experts in performing restructuring for the purpose of detecting a novel
component, but it failed to do so for novices, although also for novices a
slight increase was noted.
In an attempt to answer the question why sketching is not as helpful for
novices as for experts, it is of importance that the difference could not
Sketching and creative discovery
DST: design studies (page 17 )
535
08-10-98 10:42:43
Rev 14.02x
zdst$$140h
Figure 11 Novices
expert
sketchers
and
perform
equally on novel part detection when they are not
allowed to sketch, in the
with-sketch
condition,
experts outperform the novices
be attributed to a worse performance on, for instance, remembering the
components of the configuration. Such a difference would have led to different scores on the existing parts, which was not obtained, as Figure 12
clearly shows.
On each and every single test, the novices test scores matched those of
the experts. Therefore the difference between novices and expert sketchers
cannot be explained by individual differences in ability, as measured by
one of the various tests administered (intelligence, creativity, vividness of
mental imagery, etc.). So neither the experimental scores, nor test scores
provided a cue for the origin of the effect. For this reason, individual abilities, other than those relating directly to sketching, cannot explain the differential effect of sketching for experts and novices.
An explanation is suggested, however, from a factor analysis, the same
one as performed with the experts’ scores (see the section on Sketching
and Restructuring: Restructuring in the Component Detection Paradigm).
This analysis, when applied to novices, provided the same grouping of
variables into a Smart and an Art factor. However, one notable exception
was found. In contrast with the experts, for the novices, virtually no influence was obtained for neither the Art nor the Smart Factor on novel part
discovery. Discovery of novel parts was identified in previous sections as
a restructuring measure. This suggests, that the lower scores of novices in
536
DST: design studies (page 18 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
Figure 12 On existing parts
both the with-and without
sketch group for both the
novices and experts have
matching performances
the sketching condition for novel parts, were due only to their inability to
perform restructuring with the aid of sketches.
Differences in drawing skills may explain the novices’ inability to use
sketches for restructuring. Expert and novice sketches differ sufficiently in
appearance to warrant this conclusion: several industrial designers were
able to reliably sort the sketches into expert and novice ones. A mean
percent correct of 70% was obtained, suggesting that expertise can be seen
in the sketches.
4.2
Expertise in the figural combination paradigm
Novices took part in the 3D version of the figural combination paradigm
only. As in the Component Detection Task, the novices scored lower than
the experts on the restructuring measure in the with-sketch condition of
the Figural Combination Task, see Figure 13. In fact, novices even showed
an opposite trend. The ability to restructure therefore seems to differentiate
between experts and novices, and this argues for restructuring to be a fundamental constituent of expertise.
The combining measure did not differentiate between experts and novices.
Although in both the without-sketch and the with-sketch conditions the
novices scored lower than the experts, these differences did not reach significance, see Figure 14 (note that the scale has changed; it now reflects
Sketching and creative discovery
DST: design studies (page 19 )
537
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
Figure 13 In the
Figural
Combination Task in the
without-sketch group matching
performances
on
restructuring were found for
the novices and experts, but
in the with-sketch group the
novices are outperformed by
the experts
Figure 14 On the combining score in the Figural
Combination Task, all mean
performances
match.
It
made no difference whether
subjects were allowed to
sketch, nor when they were
expert sketchers
538
DST: design studies (page 20 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
the combining score divided by the number of objects per subject. This
manipulation was necessary because of comparison between novices and
experts). The equal performances of both groups on the combining measure
indicate that the expertise in sketching does not reside in the ability to combine.
The fact that restructuring did differentiate between experts and novices
in the with-sketch condition but not in the without-sketch condition, supports the claim that the distinctive ability of experts with respect to novices
resides in performing restructural transformations with the aid of sketches.
The ability to combine, however, was equal for experts and novices.
5
Sketching and creativity
Since only expert sketchers benefit from externalization of images, this
section returns to the achievements of expert sketchers, in order to study
the relation between sketching and creativity.
5.1
Creativity in the component detection paradigm
In the component detection experiments, the Kunzendorf test of Aesthetic
Preference46, measures the mastery of ‘visual grammar’. Creative individuals, according to Kunzendorf, have a better mastery of ‘visual grammar’,
which improves their ability to transform and restructure knowledge. It
turned out that people who scored high on this test were far more likely
to find a novel part if they had spontaneously sketched, than the low scorers
(a significant correlation of r = 0.654 was established, see Figure 15). This
finding affirms the relationship between restructuring and creativity.
Figure 15 The
found
correlation
between
creativity
scores and rate of correct
novel part detection when a
spontaneous
sketch
had
taken place
Sketching and creative discovery
DST: design studies (page 21 )
539
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
62 Wallace, D B ‘Giftedness
and the instruction of a meaningful life’, in The gifted and the talented: Developmental perspectives, F D Horowitz and M
O’Brien (eds) American Psychological Association, Washington,
DC (1985)
63 Cropley, A J ‘Improving
intelligence by fostering creativity in everyday settings’, in Intelligence: Recapitalization and
measurement, H A H Riwe
Lawrence Erlbaum, Hillsdale,
NJ (1991)
64 Guilford, J P The nature of
human intelligence McGraw-Hill
series in psychology, NY (1967)
65 Sternberg, J R Intelligence,
information processing and analogical reasoning: The componential analysis of human abilities
Lawrence
Erlbaum,
Hillsdale, NJ (1977)
66 Hennessey, B A and
Amabile, T M ‘The role of the
environment in creativity’, in The
nature of creativity, R J
Sternberg (ed) Cambridge University Press (1988)
67 Amabile, TM ‘Children’s
artistic creativity: Detrimental
effects of competition in a field
setting’ Personality and Social
Psychology Bulletin Vol 8 (1982)
pp 573–578
68 Amabile, TM ‘Social psychology of creativity: A consensual assessment technique’
Journal of Personality and Social
Psychology Vol 43 (1982)
pp 997–1013
Restructuring, creativity, and sketching were found to constitute the Art
factor. The finding of Art and Smart as two separate dimensions is in
accordance with Wallace62 notion of ‘mental power’ or ‘extraordinariness’,
which includes creativity and intelligence as separate factors63. Taken at
face value, the independence of Art and Smart factors contradicts Guilford’s64, and Sternberg’s65 idea that creativity is an aspect of intelligence.
The Art factor, however, comprises restructuring, but not combining. Guilford’s divergent thinking aspect stresses the combining aspect of creativity.
5.2
Creativity in the figural combination paradigm
The relation between combining and creativity was studied in the figural
combination paradigm. The objects combined by the subjects were rated
on creativity by independent expert judges (design teachers). Their ratings
correlated highly among each other, in contrast to those of novice judges.
This finding is in accordance with Amabile’s66–68 operational definition of
creativity, ‘A product or idea is creative to the extent that expert observers
agree it is creative’. For this reason, the expert ratings were considered a
valid indication of object creativity.
The creativity ratings correlated significantly with both the combining
score (a correlation of r = 0.482) and restructuring score (r = 0.534) in the
figural combination paradigm (only objects in the 3D variant were rated),
see respectively Figures 16 and 17.
Figure 16 The relationship
between subject’s combining
score in the Figural Combination
Task
with
their
objects’ mean creativity rating. For the with-sketch
group only
540
DST: design studies (page 22 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
Figure 17 The relationship
between subject’s restructuring score in the Figural
Combination Task with their
objects’ mean creativity rating. For the with-sketch
group only
On the basis of combining and restructuring scores, almost 70% of the
variance in the creativity ratings in the sketch condition can be explained.
The amount of explained variance provides an argument for the important
role of combining and restructuring in creative discovery in the figural
combination paradigm. These findings contrast with the view expressed by
Hennessey and Amabile66 that ‘There is one basic form of creativity, one
basic quality that observers are responding to when they label something
‘creative’.’ (p. 15). Creativity, instead, seems to have at least two constituents; restructuring and combining.
Despite the fact that these two processes can explain as much as 70% of
the variance in a single task and appear to have construct validity across
tasks, the creative process will consist of more processes than the ones
presently identified. For example, in the experiments discussed, the components for combining or restructuring were selected by the experimenter.
This was necessary for experimental control on these processes. Outside
the laboratory, the selection will be left to the creative individual. Individual preferences and differences in ability with respect to the selection of
components for mental transformation will certainly constitute another
important aspect of creativity which, however, is beyond the scope of the
present sketching study.
6
General conclusions
The distinction of combining and restructuring throws new light on the
issue of sketching. The combining process is easy to perform in mental
Sketching and creative discovery
DST: design studies (page 23 )
541
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
imagery and is not supported by sketching. The restructuring process is
difficult to perform in mental imagery and is enhanced by sketching. That
restructuring is the difficult part, is confirmed by the fact that only experts
are aided in performing it by sketching. The two processes of combining
and restructuring together constitute important elements of the creative process (see Figure 18).
From the result of these studies, it is possible to provide recommendations
for the construction of computer tools for sketching. In general, such tools
must, like paper and pencil, be intuitive; i.e. the requirements for their use
may not exceed the current level of expertise of their users. Paper and
pencil are used spontaneously by both experts and novices, which implies
that these tools must have a low threshold of accessibility and their use may
not require specialized knowledge if they are to replace paper and pencil.
Since the combining process can be easily and rapidly performed before
the mental eye, only the end-product of this process is likely to be found
externalized. Combining itself is, according to the model, never an objective for externalization, and therefore will occur only in sketches intended
to support the restructuring process. Combining objects on a computer tool,
therefore, must preferably pass effortlessly and super fast.
In addition, however, such tools must be helpful in excess of paper and
pencil tools. The present studies suggest that appropriate support given to
users, will have to allow users to raise their performance on the aspect of
the task which is most difficult to perform, viz. restructuring. In the restructuring process, the sketcher draws one particular structure but does so in
order to pick up novel structural components from perception of the sketch.
Paper-and-pencil sketches are often unspecified and vague, allowing for
perceptual creativity to flesh out a new structure. So a computer tool has
Figure 18 The
resulting
model for the role of combining, restructuring
and
sketching in creative discovery process
542
DST: design studies (page 24 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
either to support unspecified forms as input, or a flexible switching between
various structural descriptions of the input after its creation.
Kolli and Stuyver (reported in Hennessey5) studied the 3D CAD programs
Pro-Engineer, Intergraph-EMS, Sculpt 3D and GIG3DGO. The first phase
of this study revealed that with these state-of-the-art programs even a simple combining task with simple components is met with ample difficulty.
Most programs required considerable time for the alignment of components. Combining, therefore, did not proceed as effortlessly and fast as was
required for these programs to be called intuitive.
In the second phase of this study, the subjects had to perform restructuring
on the components of the previously combined objects. How the components had to be restructured during this phase was not known beforehand
to the subjects in the first phase. It turned out that some subjects had chosen
a form of combining that did not allow any alteration of the components
in the second phase. Therefore, they had to combine the object again before
they were able to restructure it. This latter result indicates that restructuring
is a cumbersome job with current 3D CAD programs. These programs,
therefore, failed to meet the requirement of being helpful as tools for ideageneration sketching in the early creative phases of the design process.
The problems of CAD programs with respect to restructuring are illustrated
in Figure 19. Suppose the left most figure was originally drawn on the
computer as two overlapping triangles. A vertical translation of these triangles with respect to each other results in Figure 19a. This new configuration preserves the original components. This transformation, which is a
form of combining, however, will not have been the motivation to sketch.
According to the present investigations, this transformation is also easy to
perform in imagery and so, there is no need for its externalization.
Figure 19 The configuration
on the left can be conceived
as
two
overlapping
tri-
angles, in that case a results
after
vertical
translation.
But since sketching is used
to restructure, a conception
of the figure as four small
triangles can be the result of
sketching two overlapping
triangles.
Translation
of
these small triangles leads
to b
Sketching and creative discovery
DST: design studies (page 25 )
543
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
Sketching is performed for the purpose of restructuring. The sketch will
be inspected for emergent structures, which will be used for further processing. For instance, the sketch could also be perceived as consisting of
four small triangles. In that case, Figure 19b illustrates a transformation
which could have been the purpose of the sketcher, viz. the vertical translation of four small triangles.
With a traditional computer program, the sketcher will meet a difficulty in
performing this transformation. The computer stored the figure on the left
as two overlapping triangles, according to the way they were originally
drawn. The program is not able to recognize the newly emerged form69.
In order to perform the vertical translation, the sketcher has to redraw the
whole configuration. This extra sketch hampers the creative process,
because it brings stagnation at a moment where progress is wanted. This
suggests that a ‘What about looking at it this way?’ provision on computers, as suggested by Tatham70, would be a fruitful extension for contemporary computer tools.
In current 3D CAD-programs, neither of the earlier-mentioned components
of the creative process, combining and restructuring, appears to be supported very well. Combining is time-consuming where it has to pass effortlessly and restructuring is hardly supported at all. Current 3D CAD-programs, therefore, do not seem appropriate for supporting the creative
process in the conceptual phase of design, where idea-sketches are usually
made. Electronic sketch tablets, which, like paper and pencil, support
unspecified input and leave the combining and restructuring to the sketchers appear to be, for the time being, more appropriate electronic ideacreation tools. Currently these tablets lack support facilities for restructuring. The efficiency of these tablets for the purpose of idea-sketching could
be considerably improved, however, if functions like the ‘what about looking at it this way’-facilities could be provided.
69 Mitchell, W J ‘A computational view of design creativity’,
in Modelling creativity and knowledge-based creative design, J S
Gero and M L Maher (eds)
Erlbaum, Hillsdale (1993)
70 Tatham, E W I’ll know what
I want when I see it.: Towards a
creative assistant Proceedings
of the Human-Computer Interaction Conference Cambridge University Press (1995) pp 267–278
The research was performed, in order to obtain recommendations from
psychological research, on how to improve the efficiency of idea-creation
tools. The present conclusions show, that such recommendations could give
a direction to the further development of such tools, such as pursued in
the IDEATE project. The wider impact for psychology of these studies,
however, will be found in the field of mental imagery research. The studies
showed that creative processes that are likely to take place in mental imagery will be combining processes. Creative processes less likely to take
place in mental imagery are instances of restructuring.
What is the significance of these studies in the context of inventive cognitive processes? It has been acknowledged that such processes rely exten-
544
DST: design studies (page 26 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
sively on visual imagery. In the light of the previous conclusions on
restructuring versus combining mental imagery, questions can be raised
about the claims by some famous creative individuals that their discoveries
were made in mental imagery. Kekulé claimed to have made his discovery
of the solution for the benzene problem while seated half asleep before his
fire place. For Kekulé, the incentive to restructure his image was the detection of a molecule with an, at that time, impossible ratio of components.
Kekulé was dozing at the fireplace, Poincaré jumped on a bus, and Archimedes was bathing. These unusual locations illustrate what these stories have
in common. All these innovators were struck by the solution for their problem at a particular moment where none of them had paper and pencil
at hand.
One possibility is that the claims simply are wrong71,72. Yet, they are taken
very seriously in the creativity literature and therefore deserve our consideration. Secondly, there is the possibility that only combining and no
restructuring took place. This would also be unlikely, certainly in Kekulé’s
case. Kekulé discovered that he had to drop the organic-molecules-canonly-exist-in-strings rule. In his days, this rule formed an essential component in the description of organic molecules. This component of the
description had to be dropped, and therefore, the whole structure had to
be reorganized. Such a process is likely to involve restructuring. Also
Archimedes, who had to decide on the volume of the King’s crown to find
out whether it was made of pure gold or if the manufacturer had played
a scurvy trick and had mislead the King, knew that he had to drop as
inappropriate all hitherto known rules for establishing volume, given the
irregular shape of the crown. Only after dropping these rules as irrelevant,
did Archimedes have the option of finding an alternative. These operations
involve the perception of novel attributes of the situation, an operation akin
to restructuring.
71 Weisberg, R W ‘Problem
solving and creativity’, in The
nature of creativity, R J
Sternberg (ed) Cambridge University Press (1988) pp 148–176
72 Weisberg, RW and Alba,
JW ‘An examination of the
alledged role of ‘fixation’ in the
solution of several ‘insight’ problems’ Journal of Experimental
Psychology: General Vol 110
(1981) pp 169–192
In most cases of cognitive innovation, the inventions involved getting free
from an original conception, which is probably identical to restructuring.
So far, the situation is parallel to the experiments; External conditions
coerce the subject into restructuring. What then facilitated restructuring in
these inventors in situations where no possibilities for externalization were
available? Maybe, a wider interpretation of what occurred in the component detection experiments will be helpful. In these experiments, when
coerced into restructuring, subjects spontaneously turned to sketching,
when allowed to. Their image would then be restructured in analogy to
their sketch. In the experiments the detection was frequently reported to
be very surprising. The surprising character of the detection matches per-
Sketching and creative discovery
DST: design studies (page 27 )
545
08-10-98 10:42:44
Rev 14.02x
zdst$$140h
fectly the usual definition of insight, e.g. ‘... previously unseen and unexpected connections reveal themselves to the mind’ (Langley and Jones73,
p. 177).
The role of analogy in creative thinking has been emphasized by many
theorists65,74,75. Possibly, extraordinarily creative individuals were able to
construct analogies within imagery, for which others, in more mundane
cases, require a sketch. The occurrence of visual analogies is a familiar
observation from the self-reports of these individuals. Kekulé discovered,
after months of confusion, that he had to drop the molecules-can-onlyexist-in-strings rule, by seeing a snake biting his own tale in the flames of
his fire-place, and spontaneously envisioned this snake to be a organic
molecule string. Archimedes saw the water-level of his bath raising while
he stepped in, and although he must have taken many baths before, he
suddenly realized the one to one relation of rising water level and the
volume placed in the water. Both individuals may have found their solution
eventually through this vehicle of visual analogy, after a long incubation
period.
73 Langley, P and Jones, R
‘Computational model of scientific insight’, in The nature of creativity, R J Sternberg (ed) Cambridge University Press (1988)
74 Perkins, D The mind’s best
work Harvard University Press,
Cambridge MA (1981)
75 Polya, G How to solve it
Princeton University Press, Princeton, NJ (1945)
Not all inventors, however, reported to have found the solution through
analogies. Poincaré is the notable exception among these self-reports of
extremely creative individuals. This suggests that even without an overt
visual analogy, a restructuring of knowledge can take place. Anyhow, it
took time before the restructuring took place spontaneously. Kekulé and
Archimedes needed time for the ‘sudden flash of insight’ to occur as well;
they happened to recognize an analogy in an event that took place in their
environment. Hence the tentative but straightforward suggestion is that
sketching provides the analogies necessary to shorten the incubation period.
To speak with Archimedes, the motto will be:
Have a bath, but make a sketch before the water runs cold!
546
DST: design studies (page 28 )
Design Studies Vol 19 No 4 October 1998
08-10-98 10:42:44
Rev 14.02x
zdst$$140h