ArticlePDF Available

What Imagery Can't Do and Why Sketching Might Help

Authors:

Abstract and Figures

The utility of enactment to overcome the limitations of imagery was investigated in three experiments using a paper- and pencil-sketching task. Novice sketchers were compared to experts. Subjects were briefly presented a configuration of several overlapping simple geometrical components and decided whether a succeeding figure formed part of it. Succeeding figures included novel parts, produced incidentally as a result of overlapping components. The ability to detect these novel parts by imagery alone was expected to be restricted. In the first experiment, forced sketching was compared to imagery-alone conditions. Forced sketching led to enhanced detection of the novel parts in expert-sketchers. By comparison, sketching had no influence on performance in novices. In the second experiment, subjects were given the option to sketch. Both novices and experts turned to spontaneous sketching in the case of novel parts, but only for experts detection of these parts was raised by it. The third experiment controlled for effects of memory load. Implications for the controversy about discovery in imagery are discussed.
Content may be subject to copyright.
EMPIRICAL STUDIES OF THE ARTS, Vol. 18(2) 167-182, 2000
WHAT IMAGERY CAN’T DO AND WHY
SKETCHING MIGHT HELP
I. M. VERSTIJNEN
Delft University of Technology, The Netherlands
C. VAN LEEUWEN
R. HAMEL
University of Amsterdam, The Netherlands
J. M. HENNESSEY
Delft University of Technology, The Netherlands
ABSTRACT
The utility of enactment to overcome the limitations of imagery was investi-
gated in three experiments using a paper- and pencil-sketching task. Novice
sketchers were compared to experts. Subjects were briefly presented a con-
figuration of several overlapping simple geometrical components and decided
whether a succeeding figure formed part of it. Succeeding figures included
novel parts, produced incidentally as a result of overlapping components. The
ability to detect these novel parts by imagery alone was expected to be
restricted. In the first experiment, forced sketching was compared to imagery-
alone conditions. Forced sketching led to enhanced detection of the novel
parts in expert-sketchers. By comparison, sketching had no influence on
performance in novices. In the second experiment, subjects were given the
option to sketch. Both novices and experts turned to spontaneous sketching
in the case of novel parts, but only for experts detection of these parts was
raised by it. The third experiment controlled for effects of memory load.
Implications for the controversy about discovery in imagery are discussed.
167
Ó2000, Baywood Publishing Co., Inc.
INTRODUCTION
Creative individuals like Einstein and Kékulé often claim that their discoveries are
a product of mental imagery alone (Ghiselin, 1952; Hadamard, 1949; Shepard,
1978). Discovery by mental imagery alone has found its laboratory-
commensurate in studies by Finke and Slayton (1988). They introduced the figural
combination task, and showed that new objects could be invented by combining
elementary forms in mental imagery (Finke, 1990).
In naturalistic studies creations are often observed to emerge from an inter-
action between mental imagery and paper & pencil sketching and people report
frustration if denied this strategy (Fish & Scrivener, 1990; Goldschmidt, 1991).
Sketching may be seen as an external aid to mental imagery in this context. Also
many creative individuals in history support this claim (e.g., Leonardo Da Vinci;
Heydenreich, Dibner, & Reti, 1982).
Anderson and Helstrup (1993ab) replicated Finke and Slayton’s (1988) figural
combination experiments, comparing mental imagery-alone conditions to sketch-
ing conditions. They found virtually no profit from paper and pencil support,
suggesting that discovery can occur in mental imagery relatively smoothly and
without a pressing need for external support.
But the figural combination task might have failed to tap specific functions that
may be impaired in imagery. As case in point, Figure 1a can be conceived in
different ways. When it is, for example, conceived as consisting of two triangles,
subjects readily detect the presence of two hourglasses in the actual presence
of the configuration, but find it particularly difficult to do so in its absence
(Reed & Johnsen, 1975). Analogously, Chambers and Reisberg (1985) showed
that subjects were almost unable to discover the reversed interpretation of a
perceptually ambiguous pattern (e.g., Figure 1b) using only mental imagery
(Hyman, 1993; Peterson, Kihlstrom, Rose, & Glisky, 1992; Reisberg &
Chambers, 1991). These functions are a likely candidate to benefit from sketch-
ing. In the Chambers and Reisberg experiments, subjects subsequently had to
sketch their image in order to check whether they had an appropriate mental
representation. On completing the sketch, several subjects reinterpreted the
image and discovered the alternative interpretation that they had previously failed
to detect.
Sketching, therefore, may facilitate reinterpretation of a configuration by
way of making it available for visual inspection. Or, in other words, sketching
enables the abandoning of one preconceived structural interpretation in favor of
a new unanticipated structural interpretation; a process we called restructuring
(Verstijnen, van Leeuwen, Goldschmidt, Hamel, & Hennessey, 1998). After
restructuring, new part-whole relationships may occur. In Figure lb for instance,
perceptual structure according to the rabbit-interpretation contains the nose of
the rabbit as a novel part, not present in the old perceptual structure, the
duck-interpretation.
168 / VERSTIJNEN ET AL.
Discovery without a need for restructuring is observed in the findings of Finke,
Pinker, and Farah (1989). These authors instructed their subjects to imagine, for
example, a capital Q, put a capital O next to it on the left, and remove the tail of the
Q and finally rotate the pattern by 90 degrees to the left. It was easy for the subjects
to detect the “8" from the resulting image. It is common knowledge that a Q
consists of two parts; a circle and a tail, so neither are novel parts. In the same
vain, restructuring is not required for Finke’s figural combination task (Finke &
Slayton, 1988) in which subjects created objects by combining simple elements.
This may explain why Anderson and Helstrup (1993a,b) found no additional value
for sketching on this task; a novel combination can be produced using mental
imagery without the need for restructuring. But an alternative explanation for their
null effect may reside in the fact that their novice sketchers may simply have
sketched too inaccurately as a result of lacking drawing skills.
The present study is aimed at testing the restructuring hypothesis: Subjects must
detect the presence of novel parts that result incidentally from the overlap of the
components of a configuration. Hypothetically, subjects will use sketching to
restructure and extract novel parts that are difficult to extract in imagery. In
addition, novice sketchers are compared to expert sketchers. In Experiment 1
compelled sketching was studied. If sketching behavior is an enactive strategy
(Reisberg & Logie, 1993), subjects will spontaneously turn to sketching, which
was tested in Experiment 2 and 3. It is predicted that if a novel part has to be
extracted in imagery, both experts and novices will choose an enactive strategy
IMAGERY AND SKETCHING / 169
Figure 1. On the left (a) a pattern used by Reed and Johnsen (1975).
If the parts (e.g., hourglasses) do not form part of the perceptually
predominant structure (e.g., overlapping triangles), discovery of these
parts is difficult in mental imagery. On the right (b) Jastrow’s
duck/rabbit (Jastrow, 1900).
(sketching). Experiment 3 was set up to study the influence of restrictions in
working memory capacity on sketching by varying the number of components in
a configuration.
EXPERIMENT 1
Method
Subjects
A total of forty-six subjects participated in this experiment, twenty-three of
them (13 female and 10 male) were undergraduate psychology students (mean
age: 22.19 years) from the University of Amsterdam who had no drawing skills.
They will be referred to as novices in sketching. These subjects received course
credit for their participation. The other subjects (also 13 female and 10 male)
were industrial design engineering students from Delft University of Technology
(mean age: 22.34 years). They received a financial reward for their participation.
Because of their intensive training in paper and pencil sketching these subjects
will be called expert sketchers.
Apparatus and Stimuli
Instructions and trials were presented on a 12" monitor at ±40 cm distance.
Responses were made by mouse clicking.
Thirty unique configurations were constructed. Each configuration contained
three out of five possible wire-frame figures: diamond, square, isosceles triangle,
right-angled triangle, and circle. An example of a configuration is shown in
Figure 2 on the left.
The configurations were standardized in three respects. First, restrictions were
imposed on the relative positions of the component figures in the configuration.
Components had to be aligned either with respect to their edges or with respect to
their centers. In Figure 2, for instance, the rightmost edge of the right-angled
triangle is aligned to the lower-right edge of the square, and the center of the
diamond is aligned to the center of the square. The second standardization regards
the size of the components. A fixed size was chosen for all wire-frame forms; the
basis of the square, the sides of the diamond and the triangles (except for the
hypotenuse) and the cross-section of the circle were all of equal length. These
standardizations were expected to facilitate the reproduction of the image by
the subject.
Sixteen of the thirty configurations were followed by one of the five compo-
nent figures. These figures existed in the set of known components, and were
called E figures. The ones that formed part of the preceding configuration are
referred to as E+, and the remaining ones as E. Eight E+and eight Efigures were
170 / VERSTIJNEN ET AL.
randomly selected from the available options. Fourteen configurations were fol-
lowed by a novel figure. Novel figures formed part of a configuration as a result
of overlapping E components (see Figure 2). They were called N figures. Only
convex N figures were chosen that generally matched the E figures in complexity
and size. Novel parts that were contained in the preceding configuration are
referred to as N+. Novel parts that were not (but formed part of another con-
figuration in the trial set) are referred to as N. Seven N+and7N
figures were
randomly selected from the available options.
Procedure
Subjects were informed that they were participating in a memory experiment.
One half of the subjects got paper-and-pencil supplies and were instructed to
sketch the configuration on every trial, even if they knew the correct answer right
away. This condition will be called the “with-sketch” condition. The remaining
subjects in the “without-sketch” condition were denied this strategy. For experts,
the with-sketch condition included eleven subjects (5 male and 6 female), and the
IMAGERY AND SKETCHING / 171
Figure 2. On the left an example of a stimulus configuration,
consisting of three wire-frame forms (square, diamond, and right-angled
triangle). On the right four alternative succeeding figures.
This particular configuration was followed by the N+figure.
without-sketch condition twelve subjects (5 male and 7 female). For novices, the
same distribution of gender and number of subjects was formed.
The instructions familiarized the subjects with the five possible E figures and
emphasized their fixed size. This was followed by an example of a configuration.
After these instructions, the subjects got two practice trials. Each trial consisted of
five-seconds presentation of a configuration, followed by a three-digit number.
For one minute subjects counted backwards from this number in steps of three,
meanwhile a fly criss-crossed the screen to erase possible after-images. There-
after, an E or N figure appeared on the screen. Unrestricted time was given for the
subjects to decide on its presence or absence in the preceding configuration. In
order for the figure to be a part of the configuration, it had to match in orientation
as well as size. Subjects in the with-sketch condition provided a response after
completing their sketch of the complete configuration as far as they were able to.
During this phase the figure remained on the screen. Subjects initiated their
response by a mouse click, whereupon a “YES” or “NO”-button appeared on the
screen; clicking the appropriate button completed the response. Feedback was
provided after each trial. Each subject received thirty trials (i.e., 8 E++8E
+7N
+
+7N
figures) in random order. The experiment was followed by a small paper
folding (punched holes) test measuring visualization ability (Ekstrom, French, &
Harman, 1976) and an aesthetic preferences test (Kunzendorf, 1982) measuring
mastery of “visual grammar.” Creative individuals are thought to have a better
mastery of “visual grammar,” which improves their ability to transform and
restructure knowledge.
Results
No significant differences between female and male subjects were obtained,
their results were therefore pooled. Percent correct scores on E figures
(E%correct) were equal for novices and experts in the without-sketch condition
(resp. 73.44 (SE = 2.79) and 70.83 (SE = 5.01)), as well as in the with-sketch
condition (resp. 79.55 (SE = 2.68) and 80.53 (SE = 2.49)). For both novices and
experts, the differences between the without- and with-sketch conditions were not
significant.
N%correct scores were equal for novices and experts in the without-sketch
condition (resp. 55.38 (SE = 4.04) and 51.79 (SE = 2.18)), but differed in
the with-sketch condition (resp. 61.04 (SE = 3.25) versus 74.42 (SE = 2.15);
t(20)=3.434; p= .003). Only for experts the N%correct scores were higher in
the with-sketch compared to the without-sketch condition (t(21) = 7.38; p< .001). In
the without-sketch condition both groups scored somewhat higher than chance
(50%), but not significantly so.
The only other significant effect was found for novices in the with-sketch
condition: a significant correlation between the punched hole test and E%correct
(r= .56; t(9) = 2.02, p< .05).
172 / VERSTIJNEN ET AL.
The results show that paper and pencil sketching can raise the detection of novel
parts, but only for expert sketchers.
EXPERIMENT 2
In this experiment spontaneous sketching was studied in order to observe
whether paper and pencil sketching is a spontaneous enactment strategy on
restructuring.
Method
Subjects
Forty-four subjects took part in this experiment: twenty-one novices (13 female,
8 male; mean age 22.08 years), and twenty-three experts (9 female and 14 male;
mean age 22.79 years). Sampling and rewarding were identical to Experiment 1.
Stimuli
The stimuli were identical to those of Experiment 1.
Procedure
In general the procedure was identical to that of Experiment 1 except that now
subjects were free to sketch if they felt the need to and their sketching behavior was
videotaped. Two sessions took place per subject. In the first session the Advanced
Progressive Matrices (Raven, 1988), the Visual Number Span Test (Ekstrom et al.,
1976), the Embedded Figures test (Witkin, Oltman, Raskin, & Karp, 1971), the
Vividness of Visual Imagery Questionnaire (Marks, 1973), and the Aesthetic
Preferences test (Kunzendorf, 1982) were administered. In the second session the
experiment took place, lasting about forty-five minutes per subject.
Scoring
A trial received a sketch score of 1 if at least one line was drawn; otherwise the
score was 0. The maximum score of a subject would therefore be 30, but occa-
sionally it was uncertain if sketching took place. These trials were left out of
the analyses of a subject. Percent sketch and correct were calculated over the
remaining trials.
Results
Female and male subjects were pooled. N%sketch was 52.26 (SE = 7.24) for
novices and 67.40 (SE = 6.82) for experts, E%sketch was 31.38 (SE = 5.56) for
novices and 37.99 (SE = 5.86) for experts. For both novices and experts more
IMAGERY AND SKETCHING / 173
spontaneous sketching occurred on N than on E figures (resp. t(20) = 3.77; p < .01
and t(22) = 7.17; p< .0l). The preference for sketching on N figures was of equal
size for novices and experts.
For both novices and experts, the correlation between E%sketch and E%correct
was not significant (resp. r= –.05 and –.03). For novices, the correlation between
N%sketch and N%correct was not significant (r= .22). For experts this correlation
was significant (r= .72, t(21) = 4.69; p< .01). The latter two correlations differ
significantly (z= 2.08, p< .05). A scatterplot of these correlations with regression
lines is shown in Figure 3. The fact that only for experts the correlation for N
figures reaches significance could be interpreted as in support of the restructuring
hypothesis. Sketching in both novices and experts is performed systematically in
order to extract a novel part, but only the latter group was able to discover the
correct answers from their drawings.
Four design teachers were reasonably accurate (70% correct) in splitting up the
sketches into ones produced by novice sketchers and ones by experts. This finding
combined with the equivalency of novices and experts on all tests, suggests that
superior performance of expert sketchers on N figures resides in the ability to
control the sketch gestures.
174 / VERSTIJNEN ET AL.
Figure 3. Scatterplot of the correlation between individual scores on
percent correct on N figures and percent sketch on N figures for
novices and experts. The plot illustrates how experts profit from their
sketches in detecting N figures.
The regression lines in Figure 3 are useful for comparing the current data with
those of Experiment 1. N%sketch = 100 in the present experiment matches the
with-sketch condition of Experiment 1, while N%sketch = 0 matches the previous
without-sketch condition/ N%sketch = 100 corresponds on the abscissa with a
value of N%correct of 78.23 for experts and of 63.55 for novices. For N%sketch =
0 we read respectively 43.66 for experts and 54.39 for novices. These values agree
well to those of Experiment 1.
The difference in correlation between N%sketch and N%correct for novices and
experts suggests an investigation of the role of creativity (as measured with
the Aesthetic Preferences Test) within the experts group. A three-way relation
between creativity, sketching and percent correct in the case of experts can be
observed by restricting the analysis to those N trials in which sketching took place;
the correlation between N%correct and creativity scores in these trials shows a
significant value of .39 (t(21) = 1.95; p< .05). In contrast, a negative correlation
existed for novice sketchers (r= –.49; t(19) = 2.43; p< .05). A possible explanation
why no effect of creativity was observed in Experiment 1 may involve the
spontaneous character of sketching in the current experiment.
Both novices and experts sketched when presented with an N figure, but it made
no difference whether or not the N figure presented formed part of the preceding
configuration (for novices: N+%sketch = 56.01 (SE = 7.37) and N%sketch = 51.79
(SE = 7.65); for experts respectively 67.91 (SE = 7.31) and 66.77 (SE = 7.12)). This
suggests that presence or absence of N figures was barely detectable before
sketching. In contrast, subjects more frequently sketched on Ethan on E+figures
(for novices: E%sketch = 38.03 (SE = 5.78) versus E+%sketch = 24.01 (SE =
5.77), t(20) = 4.050; p< .01; for experts respectively 46.20 (SE = 7.26) versus 29.74
(SE = 4.81); t(22) = 4.286; p< .01). This result illustrates that, in contrast with the N
figures, subjects had a feeling of knowing about presence or absence of an E figure
before sketching.
Table 1 shows the correlations between the experimental variables and
the administered test scores for novices and experts. The VVIQ did not show
any significant correlation with any of the dependent variables (see also
Chara & Hamm, 1989; Ernest & Paivio, 1971) and was left out of the tables.
Embedded Figures test scores were obtained for six novices only, and was also
left out.
Principal Component analyses (orthogonal transformation, varimax rotation)
on Table 1 (Embedded Figures Test by novices excluded), yielded two main
factors for both novices and experts (see resp. Figure 4a and 4b). For novices
55.3 percent of the total variance was attributable to these two factors and for
experts 65.9 percent. One factor loaded high on Raven, Visual Number Span
(VNS), Embedded Figures (experts only), E%correct, and low on Kunzendorf,
N%sketch, and E%sketch. High scores on Raven, VNS, and Embedded Figures
could indicate the ability to extract and retain the composing wire-frame figures
for the duration of a trial. This factor was therefore labeled the “Smart”-factor. The
IMAGERY AND SKETCHING / 175
other factor loaded high on Kunzendorf, N%sketch, E%sketch, and low on Raven,
VNS, Embedded Figures (experts only), and E%correct. According to loading
pattern it was labeled the “Art”-factor.
The main difference between novices and experts occurs on N%correct. The
“Art” and the “Smart”-factors loaded approximately 0 for novices on N%correct,
but for experts 79.38 percent of the variance of this variable was explained by
these two factors. For experts, both the “Art”-factor and the “Smart”-factor load on
N%correct. The “Art”-factor explained an approximately five times more of the
variance than the “Smart”-factor. This suggests that both are needed to score as
experts in this condition.
To summarize: both experts and novices turn to sketching more frequently
for N(ovel) than for E(xisting) figures, but expert performance is raised only.
“Smart”-factors load predominantly on accuracy for E figures; “Art”-factors
predominantly load on that of N figures. This result suggests that the problem in
detecting N figures lies in the quality of the sketch rather than in the ability to
memorize the components of a configuration.
EXPERIMENT 3
In the previous experiment, it was found that memory span is not related with
detecting N figures; the Visual Number Span test (Ekstrom et al., 1976) had no
correlation with correct detection. This finding argues against the intuitive idea
that sketching helps to keep up with an ever-growing memory load. Experiment 3
provides an explicit test of this hypothesis.
176 / VERSTIJNEN ET AL.
Table 1. Correlation Matrices for Novices (Lower-Left Part, N= 21)
and for Experts (Upper-Right Part, N= 23)
EXPERTS
NOVICES
Raven
VNS
Emb. Figures
Kunzendorf
N%sketch
E%sketch
N%correct
E%correct
Raven
VNS
Emb. Figures
Kunzendorf
N%sketch
E%sketch
N%correct
E%correct
1
.29
.7
.16
.12
–.14
–.10
.45
.38
1
.61
.09
.01
.22
.12
.33
.68
.49
1
.35
.40
.74
.30
.69
.26
.18
.13
1
.48
.29
.20
.05
.31
–.02
.34
.29
1
.66
.22
.15
.01
.09
.16
.34
.80
1
.22
–.05
.38
.22
.26
.39
.72
.66
1
.35
.45
.38
.61
.15
.07
–.03
.34
1
IMAGERY AND SKETCHING / 177
Figure 4. Results from a Principal Component Analysis on the correlation matrix of novice and expert sketchers.
Factor 1 has been labeled the “Art” factor, Factor 2 the “Smart” factor.
Method
Subjects
Eighteen subjects took part in this experiment, nine female and six male. All
were expert sketchers sampled and rewarded as in Experiment 1.
Stimuli
The subjects received thirty-two trials. These trials contained eight trials with
configurations composed of two E figures, eight trials of three E figures, eight
trials four E figures, and eight trials of five E figures. In sixteen trials the
configuration was followed by an E figure (8 E+and8E
figures), the rest was
followed by an N figure (8 N+and 8 Nfigures), equally balanced over different
types of configuration.
Procedure
The procedure was identical to that of Experiment 2, except that only one test
was administered: the Ekstrom et al. (1976) “punched holes” test.
Results
Male and female subjects were pooled as in the previous experiments. With all
trials combined, the present experiment replicated the earlier obtained effects: the
mean E%sketch (34.03, SE = 7.75) was lower than the mean N%sketch (49.31 (SE
= 9.04); t(17) = 3.49; p< .01), while the mean E%correct (64.94, SE = 2.95), was
similar to the mean N%correct (63.54, SE = 3.51). On configurations with 2, 3, 4,
and 5 figures the mean E%correct was respectively 65.28 (SE = 4.63), 56.94 (SE =
8.29), 66.39 (SE = 6.18), and 70.83 (SE = 4.17). The mean N%correct was
respectively 70.83 (SE = 4.63), 55.56 (SE = 5.55), 54.16 (SE = 7.89), and 73.61 (SE
= 5.53). Again a correlation was found between N%sketch and N%correct (r= .49:
t(16) = 2.24; p< .05) and an insignificant one (r= .10) between E%sketch
and E%sketch. The mean score established on the paper folding test was 17.56
(SE = .50), this test correlated with N%correct only (r= .56; t(16) = 2.73; p< .01).
The number of components in a configuration did not influence the rate of
sketching, neither for N figures nor for E figures. This result argues against an
explanation for the results of Experiments 1 and 2 based on the assumption that
sketching is a consequence of memory load.
GENERAL DISCUSSION
It is intuitively plausible that enactment strategies, such as sketching, are
frequently employed if our mental capacities fail. But previous research on
sketching (Anderson & Helstrup, 1993a,b) however, yielded disenchanting results
178 / VERSTIJNEN ET AL.
about the advantages for enactive strategies. One reason for this failure could be
that in these studies, the emphasis was on the end products, without much concern
for the process. End products were rated on creativity by judges. The criteria
for creativity employed in these studies will presumably reflect the novelty or
uniqueness of the end product. Novelty and uniqueness could, in principle, be
achieved in a variety of ways. Some of these may be possible with or without
sketching; other may depend exclusively on sketching. If experimental conditions
exclude a specific strategy that requires sketching, a subject may flexibly switch
to another strategy (Verstijnen et al., 1998).
Our study extends on Anderson and Helstrup’s (1993a,b) findings by focusing
on the process: it confirms the original intuition that sketching can be useful. We
assumed that enactment strategies like sketching are applied to compensate for a
specific flaw of imagery: its frail capacity to restructure a given configuration.
In the Anderson and Helstrup (1993a,b) task, novel part extraction was not
required. In accordance with our hypothesis, novel part extraction was performed
at chance-level without the aid of sketching. Furthermore, Anderson and Helstrup
(1993a,b) investigated only novel sketchers. Our results, however, suggest that
only expert sketchers can improve their accuracy with the aid of a sketch. Probably
only skillfully made externalizations of images can be helpful in making the
incidental information available.
The effect cannot be attributed to subjects having inadequately represented the
figure due to, for instance, memory overload. Experiment 3 showed that for both
experts and novices that information load on components did not influence the
number of spontaneous sketches. This result eliminates an explanation for the
sketching data based on insufficient representation or memory overload. The
present results are consistent with an explanation in terms of a system that, for
reasons of economy, represents in imagery only the information that is sufficient
to reconstruct the original pattern. This information will be sufficient to recog-
nize existing components (the present study) and create new objects by figural
combination. All these tasks could be performed without having to revise the
component structure of the image. The notion of restructuring may, therefore,
explain the difference between the findings of Chambers and Reisberg (1985) who
claimed that discovery in imagery is hard versus Finke (1990) who claimed that
discovery in imagery is fluent. It may be argued that two types of discovery could
be distinguished; one type, combining, is predominant in Finke’s figural com-
bination task and is easy and frequent and can be performed in imagery without
sketching, the other, restructuring, is predominant in the Chambers and Reisberg’s
reversal task, is difficult and rare and can be facilitated by sketching.
Often two processes are distinguished in creative imagery. Our distinction,
however, does not map straightforwardly on to ones made in the creativity
literature. For instance, Finke’s Geneplore model distinguishes generation and
exploration (Finke, 1993; Smith, Ward, & Finke, 1995). Exploration is thought to
exploit emergent properties. Emergent properties, however, can either result from
IMAGERY AND SKETCHING / 179
figural combination or from restructuring. Since our studies show that restruc-
turing and combination play different roles in imagery, we propose to distinguish
these processes as a basis for modeling processes of imagery, creativity, and
discovery.
The ease of creating a new object by figural combination in imagery may be
interpreted as evidence for the important role of the visual system in imagery
(Kosslyn, 1980). Perception has often been regarded as a process of creative
synthesis of sensory components. The synthesis process could also be performed
without relevant sensory stimulation, that is the case when the visual system is fed
with components from memory. However, in such an account, it cannot easily be
explained why imagery is unable to restructure, whereas this is easily done in
the presence of a visual display.
In a recent model of perceptual self-organization (van Leeuwen, Steyvers, &
Nooter, 1997) unstable object structures quickly emerge, allowing for rapid object
recognition. These unstable structures will have to be rebuilt after a certain time
interval. If rebuilding occurs under similar sensory conditions, it is possible to
detect a previously unanticipated component. This may explain why restructuring
in visual perception is relatively easy.
We may, therefore, conclude that perceptual organization and creative imagery
could be understood through the operation of the same mechanism dependent of
the presence or absence, respectively of a relevant visual display. Sketching as a
strategy to overcome the absence of a display could be useful in the context of
specific tasks where restructuring is required for creativity.
ACKNOWLEDGMENT
The author would like to thank Romke Rouw for her assistance in testing
the subjects.
REFERENCES
Anderson, R. E., & Helstrup, T. (1993a). Visual discovery in mind and on paper. Memory &
Cognition, 21, 283-293.
Anderson, R. E., & Helstrup, T. (1993b). Multiple perspectives on discovery and creativity.
In B. Roskos-Ewoldsen, M. J. Intons-Peterson, and R. E. Anderson (Eds.), Imagery,
creativity and discovery: A cognitive perspective (Advances in Psychology, 98).
Amsterdam: North Holland.
Chara, P. J., & Hamm, D. A. (1989). An inquiry into the construct validity of the Vividness
of Visual Imagery Questionnaire. Perceptual and Motor Skills, 69, 127-136.
Chambers, D., & Reisberg, D. (1985). Can mental images be ambiguous? Journal of
Experimental Psychology: Human Perception and Performance, 11, 317-328.
Ekstrom, R. B., French, J. W., & Harman, H. H. (1976). Kit of factor-referenced cognitive
tests. Princeton, NJ: Educational Testing Service.
180 / VERSTIJNEN ET AL.
Ernest, C. H., & Paivio, A. (1971). Imagery and sex differences in incidental recall. British
Journal of Psychology, 62, 67-72.
Finke, R. A. (1990). Creative imagery: Discoveries and inventions in visualization.
Hillsdale, NJ: Erlbaum.
Finke, R. A. (1993). Mental imagery and creative discovery. In B. Rosuos-Ewoldsen, M. J.
Intons-Peterson, & R. E. Anderson (Eds.), Imagery, creativity, and discovery: A
cognitive perspective (Advances in Psychology, 98). Amsterdam: North Holland.
Finke, R. A., Pinker, S., & Farah, M. J. (1989). Reinterpreting visual patterns in imagery.
Cognitive Science, 13, 51-78.
Finke, R. A., & Slayton, K. (1988). Explorations of creative visual synthesis in mental
imagery. Memory & Cognition, 16, 252-257.
Fish, J., & Scrivener, S. (1990). Amplifying the mind’s eye: Sketching and visual cognition.
Leonardo, 23, 117-126.
Ghiselin, B. (1952). The creative process. New York: American Library.
Goldschmidt, G. (1991). The dialectics of sketching. Creativity Research Journal, 4(2),
123-143.
Hadamard, J. (1949). The psychology of invention in the mathematical field. Princeton,
NJ: Princeton University Press.
Heydenreich, L. H., Dibner, B., & Reti, L. (1982). Leonardo the inventor. London:
Hutchinson.
Hyman, I. E. (1993). Imagery, reconstructive memory and discovery. In B. Roskos-
Ewoldsen, M. J. Intons-Peterson, & R. E. Anderson (Eds.), Imagery, creativity and
discovery: A cognitive perspective (Advances in Psychology, 98). Amsterdam: North
Holland.
Jastrow, J. (1900). Fact and fable in psychology. Boston, MA: Houghton Mifflin.
Kosslyn, S. M. (1980). Image and mind. Cambridge, MA: Harvard University Press.
Kunzendorf, R. G. (1982). Mental images, appreciation of grammatical patterns, and
creativity. Journal of Mental Imagery, 6, 183-201.
Marks, D. F. (1973). Visual imagery differences in the recall of pictures. British Journal
of Psychology, 64, 17-24.
Peterson, M., Kihlstrom, J., Rose, P., & Glisky, M. (1992). Mental images can be
ambiguous; Parts, wholes, and strategies. Memory & Cognition, 20, 107-223.
Raven, J. C. (1988). Advanced Progressive Matrices. Set II.
Reed, S. K., & Johnsen, J. A. (1975). Detection of parts in patterns and images. Memory
& Cognition, 3, 569-575.
Reisberg, D., & Chambers, D. (1991). Neither pictures nor propositions: What can we
learn from a mental image? Canadian Journal of Experimental Psychology, 45,
288-302.
Reisberg, D., & Logie, R. (1993). The ins and outs of working memory. In B. Roskos-
Ewoldsen, M. J. Intons-Peterson, & R. E. Anderson (Eds.), Imagery, creativity and
discovery: A cognitive perspective (Advances in Psychology, 98). Amsterdam: North
Holland.
Shepard, R. N. (1978). Externalization of mental images and the act of creation. In
B. S. Randhawa & W. E. Coffman (Eds.), Visual learning, thinking and communica-
tion (pp. 133-189). NY: Academic Press.
Smith, S. M., Ward, T. B., & Finke, R. A. (1995). The creative cognition approach.
Cambridge, MA: MIT Press.
IMAGERY AND SKETCHING / 181
van Leeuwen, C., Steyvers, M., & Nooter, M. (1997). Stability and intermittency in
large-scale coupled oscillator models for perceptual segmentation. Journal of
Mathematical Psychology, 41(4), 319-344.
Verstijnen, I. M., van Leeuwen, C., Goldschmidt, G., Hamel, R., & Hennessey, J. M.
(1998). Creative discovery in imagery and perception: Combining is relatively easy,
restructuring takes a sketch. Acta Psychologica, 99(2), 177-200..
Witkin, H. A., Oltman, P. K., Raskin, E., & Karp, S. A. (1971). A manual for the embedded
figures tests. Consulting Psychologists Press.
Direct reprint requests to:
I. M. Verstijnen
University of Leuven
Department of Psychology
Tiensestraat 102
B-3000 Leuven, Belgium
e-mail: ilse.verstijnen@psy.kuleuven.ac.be
182 / VERSTIJNEN ET AL.
... The investigation of sketching-behavior as a means to gain insight in mental imagery is taken up by only a small number of psychologists and usually they study novice sketchers (e.g. Anderson & Helstrup, 1993ab), although there are some recent exceptions (Verstijnen, van Leeuwen, Goldschmidt, Hamel, & Hennessey, 1998; Verstijnen, van Leeuwen, Hamel & Hennessey, 2000). Sketching is usually seen as a form of 'Ena cted Imagery' (Reisberg & Logie, 1993), a term which does not only cover the close bond between sketching and imagery, but also suggests that sketching fulfills a need that arises from imagery. ...
... While the process of combining enrolls quite easily before the mental eye, another process called restructuring (or reconstruals) causes considerable difficulty (e.g. Reisberg, 1987; Verstijnen et al., 2000). Restructuring involves a structural change in the information that is loaded from memory. ...
... In a number of experiments, the hypothesis stating that paper and pencil sketching aids restructuring, proved to be fruitful. In one experiment for example (Verstijnen et al., 2000), figures like those inFigure 1 on the right were shown to the subjects (experienced sketchers). It was asserted that the subjects stored just one particular interpretation in memory, say a twotriangle interpretation. ...
Conference Paper
Full-text available
Paper and pencil sketching, visual analogies, and creativity are intuitively interconnected in design. This paper reports on previous and current research activities of a psychologist (the 1 st author) and an architect/designer (the 2 nd author) on issues concerning sketching and analogies, and analogies and creativity respectively. In this paper we tried to unite these findings into a combined theory on how sketching, analogies, and creativity interrelate. An appealing theory emerges. It is hypothesized that with no paper available or no expertise to use it, analogies can be used to support the creative process instead of sketches. This theory, however, is a tentative one that needs more research to be confirmed.
... And while some people may be more 'insightful' than others (Verstijnen et al., 2000), we can safely assume that the cognitive processes underlying (insight and noninsight) problem solving are rougly the same in every normal person. That can indeed be a reason why including (a), stating that the problem must be well within the competence of the average subject, in a definition is redundant. ...
... In these kinds of experiments, the focus is often on 'insight' versus 'noninsight,' so it would be interesting to focus on recombination and restructuring instead. The results could also be compared with experiments on the recombination/restructuring difference in visual perception (Verstijnen et al., 1998;Verstijnen et al., 2000). ...
Data
Full-text available
... However, it is a relatively unexplored avenue in cognitive science how creative cognitive phenomena (such as analogy, imagery, and mental models) interact with the presence or absence of external support systems. In the literature, such interaction studies have typically been limited to lab experiments contrasting sketching with no sketching in order to examine characteristics of imagery, such as the possibility of restructuring (e.g., Anderson & Helstrup, 1993;Finke, 1990;Verstijnen, van Leeuwen, Goldschmidt, Hamel, & Hennessey, 1998;Verstijnen, van Leeuwen, Hamel, & Hennessey, 2000). This literature on imagery has revealed that, at least under some conditions, cognition with the support of sketching is superior to cognition without external support (Verstijnen et al., 1998;Verstijnen et al., 2000;Roskos-Ewoldson, Intons-Peterson, & Anderson, 1993). ...
... In the literature, such interaction studies have typically been limited to lab experiments contrasting sketching with no sketching in order to examine characteristics of imagery, such as the possibility of restructuring (e.g., Anderson & Helstrup, 1993;Finke, 1990;Verstijnen, van Leeuwen, Goldschmidt, Hamel, & Hennessey, 1998;Verstijnen, van Leeuwen, Hamel, & Hennessey, 2000). This literature on imagery has revealed that, at least under some conditions, cognition with the support of sketching is superior to cognition without external support (Verstijnen et al., 1998;Verstijnen et al., 2000;Roskos-Ewoldson, Intons-Peterson, & Anderson, 1993). ...
... Graphical and industrial designers (Verstijnen, Goldschmidt, van Leeuwen, Hamel, & Hennessey, 1998;Verstijnen, van Leeuwen, Hamel, & Hennessey, 2000), architects (Suwa & Tversky, 1997), and artists (van Leeuwen, Verstijnen, & Hekkert, 1999) often make sketches as intermediate products before producing their final design. Several studies have addressed the role of sketching in creative design processes (Anderson & Helstrup, 1993a, 1993bRoskos-Ewoldson, 1993;Suwa & Tversky, 1997;van Leeuwen et al., 1999;Verstijnen et al., 1998Verstijnen et al., , 2000. ...
... Graphical and industrial designers (Verstijnen, Goldschmidt, van Leeuwen, Hamel, & Hennessey, 1998;Verstijnen, van Leeuwen, Hamel, & Hennessey, 2000), architects (Suwa & Tversky, 1997), and artists (van Leeuwen, Verstijnen, & Hekkert, 1999) often make sketches as intermediate products before producing their final design. Several studies have addressed the role of sketching in creative design processes (Anderson & Helstrup, 1993a, 1993bRoskos-Ewoldson, 1993;Suwa & Tversky, 1997;van Leeuwen et al., 1999;Verstijnen et al., 1998Verstijnen et al., , 2000. In principle, the design process may depend on these preliminary sketches in several ways. ...
Article
Full-text available
Novice designers produced a sequence of sketches while inventing a logo for a novel brand of soft drink. The sketches were scored for the presence of specific objects, their local features and global composition. Self-assessment scores for each sketch and art critics' scores for the end products were collected. It was investigated whether the design evolves in an essentially random fashion or according to an overall heuristic. The results indicated a macrostructure in the evolution of the design, characterized by two stages. For the majority of participants, the first stage is marked by the introduction and modification of novel objects and their local and global aspects; the second stage is characterized by changes in their global composition. The minority that showed the better designs has a different strategy, in which most global changes were made in the beginning. Although participants did not consciously apply these strategies, their self-assessment scores reflect the stages of the process.
... It also characterizes some highly cultured aspects of perception. Discovering the hidden structure in a visual display plays a role in the creative process of designers (Verstijnen, van Leeuwen, Goldschmidt et al. 1998a,b;Verstijnen, van Leeuwen & Hamel 2000) and artists (van Leeuwen, Verstijnen & Hekkert 1999), who typically use externalizations of their creative process, such as sketches, to detect the surplus structure in their images and guide the further course of their design process. Towards the end of the design process, the various structures are then assembled to be integrated in the final design (Jaarsveld & van Leeuwen 2005). ...
... In the literature, such interaction studies have typically been limited to lab experiments contrasting sketching with no sketching in order to examine characteristics of imagery, such as the possibility of restructuring (e.g. Anderson & Helstrup, 1993;Finke, 1990;Verstijnen, van Leeuwen, Goldschmidt, Hamel, & Hennessey, 1998;Verstijnen, van Leeuwen, Hamel, & Hennessey, 2000). Even less studied than sketching, however, is cognition with the external support of physical objects, such as prototypes. ...
Article
Although theories of mental simulations have used different formulations of the premises of ‘thought experiments’, they can be fitted under a minimalist hypothesis stating that mental simulations are run under situations of uncertainty to turn that uncertainty into approximate answers. Three basic assumptions of mental simulations were tested by using naturalistic data from engineering design. Results from the design protocols showed (1) initial representations in mental simulation had higher than base-rate uncertainty, (2) uncertainty in mental simulations were lowered after simulation runs, (3) resulting representations had more approximations than base-rate or initial representations. Further, the reference to external representational systems (sketches and prototypes) was examined. It was found that prototypes had fewer technical/functional simulations compared to sketches or unsupported cognition. Although prototypes were associated with more approximation than unsupported cognition, the different external representation categories did not differ in information uncertainty. The results support the minimalist hypothesis of mental simulations. Copyright © 2008 John Wiley & Sons, Ltd.
Chapter
Diagrams figure prominently in human reasoning, especially in science. Cognitive science research has provided important insights into the inferences afforded by diagrams and revealed differences in the reasoning made possible by physically instantiated diagrams and merely imagined ones. In scientific practice, diagrams figure prominently both in the way scientists reason about data and in how they conceptualize explanatory mechanisms. To identify patterns in data, scientists often graph it. While some graph formats, such as line graphs, are used widely, scientists often develop specialized formats designed to reveal specific types of patterns and not infrequently employ multiple formats to present the same data, a practice illustrated with graph formats developed in circadian biology . Cognitive scientists have revealed the spatial reasoning and iterative search processes scientists deploy in understanding graphs. In developing explanations, scientists commonly diagram mechanisms they take to be responsible for a phenomenon, a practice again illustrated with diagrams of circadian mechanisms. Cognitive science research has revealed how reasoners mentally animate such diagrams to understand how a mechanism generates a phenomenon.
Article
Full-text available
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.
Article
Full-text available
A figure combination task, in which three components are combined into an object, was administered under imagery-alone and externalization conditions to subjects with different levels of sketching expertise. In externalization conditions the imaged combinations were sketched. In accordance with earlier studies, the combinations were rated equally creative across conditions. The combinations were scored with regard to the novelty of their spatial configurations of the components (combining score), and with regard to the novelty of the structure of the components (restructuring score). For expert sketchers the latter score was found to be increased by sketching. Creativity ratings correlated with both combining and restructuring scores in the sketching condition, but only with combining scores in the imagery condition. The results are interpreted in terms of a model in which creative processes use combining and restructuring strategies in a flexible way. Whereas restructuring draws heavily on both externalization and expertise in externalization, combining can be used independently of externalization and expertise.
Chapter
Full-text available
The reconstructive view of image construction suggests some constraints on discovery using mental images—namely, discovering something about the gist or consistent details should be fairly straightforward, discovering something about less important details or something that contradicts the gist should be more difficult. This reconstructive view of imagery led to thetwo types of experiments discussed in the chapter. The first line of experiments investigated whether classic ambiguous figures can be reinterpreted using mental imagery. The results of these experiments suggest that a view of image creation as a reconstructive memory task may be productive. The other line of experiments is based on ongoing research that more directly investigates the relation between reconstructive memory and mental imagery. People with high mental imagery ability may be better able to manipulate their images or better able to visualize more extraneous details than people with low imagery ability (Kaufmann and Helstrup). A primary distinction between perception and imagery is the source of the information for the visual system: in perception that source is the world; while in imagery it is memory.
Article
Full-text available
Subjects performed in an embedded-figures detection task which required them to judge whether one pattern was a part of another. In the perception condition, the part was presented before the complete pattern, but in the imagery condition, the part was presented after the complete pattern. Subjects made fast, but inaccurate, responses in the perception task when RT s were recorded, but the error rate declined substantially when they were given 10 sec to make a decision. In the latter condition, subjects failed to detect a part on 14% of the trials in the perception condition and on 72% of the trials in the imagery condition when a correction was made for prior perception of the part. A subsequent experiment showed that the high error rate in the imagery task was not the result of the subject's inability to remember the complete pattern. The complexity of mental operations and the limited accuracy of visual images are considered as possible alternative explanations of the results.
Chapter
This chapter outlines a general, cognitive approach to the study of creative thinking and discovery. Both methodological and evaluative issues are considered. The idea that imposing constraints can help to study creative thinking might seem paradoxical, in that when a person is no longer free to respond in particular ways, the person might be less creative. However, certain types of restrictions ought to enhance creativity. Thus, imposing such restrictions would not only place the creative act under experimental control, it would also tend to encourage creative thinking. In general, whether a particular restriction enhances or inhibits, creativity can be determined empirically. Another useful technique is to suspend or delay the application of expert knowledge. This is done in the spirit of trying to avoid conventional mental sets. A subject might be asked to generate a mental image of a structure without initially knowing what purpose the structure is supposed to have.
Book
ALDERMAN CALL NUMBER: BF1031 .J25 1900