THINKING THROUGH DRAWING: PRACTICE INTO KNOWLEDGE

THINKING THROUGH DRAWING: PRACTICE INTO KNOWLEDGE 63

Justin Ostrofsky

participants could slide it between their drawing and
the photograph to see what the drawing looked like
without interference from the photo underneath.
Participants were instructed to use the available line
segments to create a tracing that was as accurate as
possible, given the constraints of the medium. Accu-
rate realism, rather than creativity, was explicitly
emphasized. Participants were required to use all
30 pieces of tape. They could bend a segment if they
liked but could not tear it into smaller pieces; they
could also move a piece of tape after having used
it in the drawing if they decided it would go better
somewhere else. Participants were given 15 minutes
to complete their tracing.
Based on extensive pilot testing, thirty segments
appeared to be enough to make a potentially inter-
esting drawing and to permit a wide range of depic-
tions, but importantly, not enough to trace all the
information in the photo. Thus, the limited number
of line segments is an important methodological
control intended to force participants to make careful
choices about what visual information from the pho-
tograph to include versus exclude from the tracings.
We analyzed performance in this tracing task in
two ways. First, we quantified the accuracy of these
tracings by having three trained artists rate how
accurately each tracing represented the photograph
on an 8-point scale (see Figure 2 for the top 4 and
bottom 4 rated tracings). The tracings produced
by trained artists were rated as more accurate than

those produced by non-artists. Next, we calculated

the correlation between the accuracy ratings for the

tracings and the drawings of the octopus photo-

graph that the participants produced, as described

earlier. We found a moderate and statistically reli-

able positive correlation between tracing and

drawing accuracy (r = .39, p < .05), indicating that

tracings rated as highly accurate tended to be pro-

duced by individuals who made drawings that were

also highly rated on accuracy. These results support

the idea that the decision of what visual informa-

tion should be included versus excluded from an

observational drawing is an important process that

contributes to the accuracy of a drawing.

The second analysis of the tracings compared

the frequency with which artists and non-artists

spontaneously used four different kinds of verti-

ces in their renderings (derived from Biederman,

1987). This was done to examine what kind of infor-

mation might be selected by skilled artists versus

being relatively neglected by non-artists. Specifi-

cally, for each tracing, we counted the number of L-,

fork, arrow, and T-junctions that were reproduced.

Figure 3 shows that trained artists were observed to

include reliably more T-, L- and fork junctions than

non-artists. Although a trend was present for art-

ists to trace more arrow junctions than non-artists,

this was not a statistically reliable difference. These

results suggest that skilled artists are more sensi-

tive in selecting visual information to include in a

Figure 2. Examples of the limited-line tracings made by participants based on a photographic model of

an elephant. The top row presents the top 4 rated tracings and the bottom row presents the bottom 4 rated

tracings.