THINKING THROUGH DRAWING: PRACTICE INTO KNOWLEDGE

(Jeff_L) #1

THINKING THROUGH DRAWING: PRACTICE INTO KNOWLEDGE 75


Ruben Coen-Cagli

Experimental Methods
Eye scan records were obtained from 35 subjects
(29 for the copy-drawing experiment, 6 for the free-
viewing control). All subjects had normal or cor-
rected to normal vision; none had specific previous
training in drawing or painting. The experimental
setup is shown in Figure 1. Subjects were presented
with a horizontal tablet 40 cm x 30 cm, viewed from
a distance such that they could comfortably draw.
In the left half of the tablet hand–drawn images
were displayed, while a white sheet (canvas) covered
the right half. The original images represented sim-
ple contours. One image per trial was shown, and
the subjects were instructed to copy its contours
faithfully on the canvas, but the instructions did
not make explicit mention of eye movements and
did not constraint the execution time (the average
execution time was 17±9 sec). For the free viewing
experiment, original images were digitized with
a scanner, and displayed on a 19-inch computer
screen for 10 seconds each, interleaved with a 5 sec-
onds blank screen. Screen resolution and viewing
distance were chosen in such a way that the images
subtended a similar visual angle as in the drawing
trials.


The subject’s left eye movements were recorded
with a remote eye tracker (ASL 5000 series) with the
aid of a magnetic head tracker. The eye position was
sampled at the rate of 60 Hz. The instrument can
integrate eye and head data in real time and deliver
a record with an accuracy of less than 1 deg in
optimal light conditions. Fixations were defined by
periods during which the gaze position was stable:
they were detected from raw data with the standard
dispersion algorithm, with threshold set to 2.0 deg
of visual angle and minimum fixation duration of
100 ms. We refer to rapid eye movements between
fixations as saccades, and to the temporal sequence
of fixation points as the scanpath. See Figure 1.

Results
Local processing bias and “edge-following” scan-
paths, in copy-drawing but not in free-viewing.
Qualitative analysis of the data in the drawing
experiments revealed that all of our subjects used
graphically continuous hand strokes (this was not
required by experimental instructions); this is a
natural motor constraint that subjects had to con-
tend with by means of some eye-hand coordina-
tion strategy. We explored the individual strategies
adopted, and found that there were remarkable
similarities across subjects. First of all, there was a
striking effect of the drawing task (as opposed to
free viewing) on the length of saccades as well as on
the distance between fixations separated by more
than one saccade, which indicated a strong local
processing bias.
This effect is quantified in Figure 2: the mean
distance between fixation points separated by few
saccades was significantly smaller in the drawing
task; in particular, the average saccade length (cor-
responding to the first point on the horizontal axis)
was almost halved in drawing. In addition, most
drawing scanpaths approximately followed the
image contours (edge-following behavior). While
this could be thought of as a strategy to obtain a
higher-resolution sampling of the image, which
may be needed to accurately reproduce it, we argue
that instead the effect is largely a consequence of the
constraint posed by motor continuity.
See Figure 2.

Edge-following was coordinated with graphically
continuous hand movements. Figure 3 depicts the
cumulative plot of fixations, and the corresponding
hand positions, of one subject at four subsequent

Figure 1. Experimental setup for eye tracking re-
cordings during the drawing task. The Subject sits
in front of a horizontal Tablet. In the left half of the
Tablet, hand–drawn images are displayed while the
Subject is instructed to copy the images on the right
half. The eye tracker integrates data from the Eye
Camera and the Magnetic Sensor and Transmitter.
Eye position is then superimposed on the Scene
Camera video stream, which takes the approximate
subjective point of view.

Free download pdf