Striking deficiency in top-down perceptual reorganization of two-tone images in preschool children

Abstract

Two-tone images (black and white transformations of gray-scale photographs) can be difficult for adult observers to recognize. However, following a brief presentation of the original photograph from which the two-tone image was created, adults experience rapid and long-lasting perceptual reorganization, such that after the initial presentation, the two-tone image becomes immediately and easily recognizable. Following a previously reported observation [1], we present evidence that, in contrast to the effortless recognition seen in adults, preschool-aged children are generally unable to recognize two-tone images even when the photograph is simultaneously available. When asked to draw corresponding parts of the photo and two-tone images, children often marked correct regions of the photo and nonsensical regions of the two-tone image. A control experiment showed that children are fully able to mark corresponding parts of two identical photographs. These results point to a dramatic lack of cue-driven perceptual reorganization in young children under conditions that trigger instant recognition in adults. We suggest that this robust phenomenon may provide a window into the development of top-down mechanisms of perceptual learning and consider interventions (e.g., |2-4]) that may improve young children's ability to use one image to reorganize another.

Full text

Page 1

Striking Deficiency in Top-Down Perceptual
Reorganization of Two-Tone Images in Preschool
Children

J.M.D. Yoon* J. Winawer
Stanford University Stanford University
Stanford, CA 94305 Stanford, CA 94305
jennifer.yoon@stanford.edu winawer@stanford.edu
N. Witthoft
Stanford University
Stanford, CA 94305
witthoft@stanford.edu
E.M. Markman
Stanford University
Stanford, CA 94305
markman@stanford.edu


of gray-scale photographs) can be difficult for adult observers to
recognize. However, following a brief presentation of the original
photograph from which the two-tone image was created, adults
experience rapid and long-lasting perceptual reorganization,
such that after the initial presentation, the two-tone image
becomes immediately and easily recognizable. Following a
previously reported observation [1], we present evidence that, in
contrast to the effortless recognition seen in adults, preschool-
aged children are generally unable to recognize two-tone images
even when the photograph is simultaneously available. When
asked to draw corresponding parts of the photo and two-tone
images, children often marked correct regions of the photo and
nonsensical regions of the two-tone image. A control experiment
showed that children are fully able to mark corresponding parts
of two identical photographs. These results point to a dramatic
lack of cue-driven perceptual reorganization in young children
under conditions that trigger instant recognition in adults. We
suggest that this robust phenomenon may provide a window into
the development of top-down mechanisms of perceptual learning
and consider interventions (e.g., [2-4]) that may improve young
children's ability to use one image to reorganize another.


Index Terms – Development, Perceptual Reorganization,
Preschool Children, Two-Tone Images.

INTRODUCTION

* Supported by an NSF Graduate Research Fellowship
Abstract - Two-tone images (black and white transformations

textbooks that illustrates how information not derived directly
from the image can be used to guide perceptual organization.
In this case, naïve observers often fail to see the Dalmatian
dog in the snow and instead perceive a disorganized
smattering of black and white patches. However, when
instructed to look for a Dalmatian, viewers find the image
suddenly and vividly transformed into a coherent percept [5,
6]. Moreover, following reorganization, it is difficult if not
impossible to see the image as a meaningless array of patches.
This serves as a powerful demonstration of the ability of top-
down information to drive perceptual reorganization.
Other cues, such as a single presentation of the photo from
which the two-tone image was derived, can also trigger
immediate perceptual reorganization of the corresponding
two-tone image in human and non-human adult primates.
Long-lasting recognition for the two-tone image obtains
Figure 1 shows a two-tone image often seen in perception
without further need for the cue, resulting in enhanced
responsivity in neurons in the inferior temporal cortex [7] as
well as increased functional connectivity between temporal
and frontal cortical regions [8] when viewing previously cued
two-tone images in comparison to equivalent uncued and
unrecognised images. That an extrinsic cue can trigger
perceptual reorganization following a single presentation
suggests the influence of a process high in the visual hierarchy
driving a top-down reinterpretation of an otherwise unchanged
stimulus.


Fig. 1 Dalmatian dog two-tone image (hint: head is on the left).
Kovacs and Eisenberg [1] suggested that this ability
might only develop over time. They presented two-tone
images and their corresponding photos to 4-5 year-old
children. None of the eight children observed were able to
verbally identify the two-tone images, even with simultaneous
presentation of the corresponding photo. This phenomenon is
striking in that children appear to not just be slower or less
accurate; rather, when viewing the two-tone images and
photos side-by-side, they fail to see what adults see
automatically. The experiments reported below document our
efforts to probe the robustness of Kovacs and Eisenberg’s
observations, develop a measure that allows children to show
us how well they can or cannot perceive the correspondence
between a two-tone image and its cue, and to quantify the
1-4244-1116-5/07/$25.00 ©2007 IEEE.181

Page 2

difference between children and adults in two-tone image
recognition.
EXPERIMENT 1

preschool children seven two-tone images and the
corresponding photos from which the two-tone images were
derived. We recorded the children’s spontaneous remarks
about the similarity between the two-tone and photo image
pairs. In cases where children claimed that the two-tone image
was the same as the photo, we asked the children to draw
corresponding parts of the two images. The seven test trial
image pairs were preceded by two practice image pairs using
the same procedure. For those two-tone images that children
failed to recognize at initial presentation, we were interested in
how often children would report perceptual re-organization
when shown the photo, and how well they would be able to
draw corresponding parts of the two-tone and photo images.

Participants
Twelve preschool children (3y3m – 4y11m, average
4y6m) from Bing Nursery School participated in the study.
Three children were excluded as a result of the child not
finishing the study (n=1) or experimenter error (n=2).

Following Kovacs and Eisenberg [1], we showed

Fig. 2 Seven two-tone / photo pairs and two practice items (bottom right).

Stimuli
Two-tone images were created by thresholding gray-scale
photographs of a kitten, dolphins (the example image used in
the Kovacs & Eisenberg, 2004), a cheetah, a dog, a polar bear,
a tiger, and an athlete (Fig. 2). Each trial included a two-tone
image paired with the original gray-scale photograph from
which it was derived. Two additional practice image pairs
were created using simpler image transformations. A
photograph of a deer was paired with a blurred version of the
same photograph and a house was paired with a higher
contrast version of the same image. Each image was printed
onto a 12 x 12 cm card.

Procedure
The experimenter sat across from the child at a small table
and explained the procedure to the child by saying, “In this
game, I’m going to show you some pictures. You just tell me
what you think is in this picture. Sometimes, the pictures will
be very clear and sometimes they will look all fuzzy and
blurry.” Each trial consisted of at least three stages: (1) uncued
two-tone identification, (2) photo identification, and (3) cued
two-tone identification. An optional fourth stage occurred
depending on the outcome of the third stage: (4) drawing
corresponding parts of two-tone and photo images.
In stage (1), uncued two-tone identification, the
experimenter placed a two-tone image card before the child
and asked, “What do you think this pictures is?” If children
did not provide an identifying label, the experimenter asked
them to make their best guess, but did not offer suggestions.
In stage (2), photo cue identification, the experimenter
moved the two-tone image card to one side, and then placed a
photo card before the child and asked, “What about this
picture? What do you see?” The photo card was always the
one that corresponded to the two-tone image card that the
child saw in stage (1).
In stage (3), cued two-tone
experimenter placed the two-tone card from stage (1) next to
the photo card and pointed to the two-tone image card, saying,
“Let’s look at this picture again. Does it look different to you?
Do you see anything new in this picture?”

Drawing
When children gave the same label for the two-tone and
photo cards or otherwise indicated that the two were the same,
the trial continued to stage (4) and they were then invited to
draw corresponding parts of the two-tone and photo image.
The experimenter picked two or three features of the named
image and asked, “Can you draw for me where the [feature] is
in this picture?”, pointing to the photo. Then the experimenter
asked, “Can you draw for me where the same [feature] is in
this picture?”, pointing to the two-tone image. The
experimenter gave positive feedback no matter what the child
drew, “Wow, that looks great! Thank you.”
For each trial, drawings on the paired two-tone and photo
images were then given a single score of 0, 0.5 or 1. Drawings
earned a score of 1 if all parts drawn on the two-tone image
corresponded correctly to the parts drawn on the two-tone
image. Drawings earned a score of 0.5 if at least one part (e.g.,
nose) drawn on the two-tone image corresponded exactly with
the part drawn on the two-tone image, but some remaining
drawn parts did not correspond. Drawings received a score of
0 if no parts drawn on the two-tone image corresponded to the
same parts drawn on the photo.

Results and discussion
Performance on the practice items demonstrated that
children understood the task. Every participant recognized
both practice items correctly without the photo cue (the
blurred deer and the high contrast house). Participants also
identification, the
182

Page 3

successfully drew corresponding parts between the practice
items and photos (mean score of 0.94, SD = 0.11).
In contrast, participants recognized only 19% of two-tone
images without a cue (SD = 14%). Moreover, on 40% of test
trials (SD = 43%), participants did not claim to perceive a
correspondence between the two-tone and photo image pairs.
On these trials, the participants gave different labels for the
paired two-tone and photo images, and at stage (3), cued two-
tone identification, either said that they did not see anything
new or different about the two-tone image after seeing the
photo, or offered a new incorrect label for the two-tone image.
Incorrect labels for the two-tone images were often highly
descriptive and fanciful (e.g., “the whole sky and earth”, “lines
in the water”, “a moose walking in trees”, “really deep grass”).
On the remaining 41% of test trials, the participants first
gave an incorrect label to the uncued two-tone image during
stage (1) and then identified the cued two-tone image at stage
(3) using the same label used to identify the photo. The
children often explicitly indicated that they had changed their
mind and that the two-tone image was the same as the photo
(e.g., “Instead of that’s a bat, that’s a dolphin”, “I mean that’s
a puppy dog”, “Oh, I know! This is a polar bear and this is a
polar bear, too”).


Fig. 3 An example in which the child indicated a match but could not draw
corresponding parts. In red, a 4-year-old has drawn the nose (arrow 1) and
claws (arrow 2) of the bear in the photo, and what she claims are the same
parts in the two-tone image of the bear.
Curiously, on these trials in which the child verbally indicated
that the photo cue had reorganized their perception of the two-
tone image, the average drawing scores across children was
only 0.22 (SD = 0.19), indicating that their drawings had few
if any correctly corresponding parts and largely earned scores
of only 0 or 0.5. This is much lower than the average of 0.94
on the practice items (P < .001, two-tailed paired t-test),
suggesting that neither lack of task understanding nor poor
drawing skills were responsible for the inability to draw
corresponding parts. The drawing score was about the same
when children had first recognized the two-tone image uncued
(mean = 0.23, SD = .22), although this was only a small
number of trials (a mean of 1.0 per child). The only items
recognized uncued by more than one child were the tiger and
the cheetah. The fact that they did not draw corresponding
parts on these items well perhaps suggests that they identified
them by features such as stripes or spots but did not clearly see
the figures.
EXPERIMENT 2

successful recognition; if the child gave the photo and the two-
tone image different labels, then s/he was not asked to draw
corresponding parts. Furthermore, we found that even when
the child did assert that the photo matched the two-tone image,
the drawings (as in Fig. 3) suggested that they did not in fact
see the match. To further explore the robustness of the
phenomenon, we tested another group of children with a
modified procedure. During side-by-side presentation in
Experiment 2, the experimenter explicitly told children that
the two-tone and photo images were the same. We were
interested in whether drawing scores would improve under
these conditions, particularly in “candidate perceptual
reorganization trials”, the subset of trials in which children did
not initially recognize the two-tone image (prior to the photo
cue). In addition, we were able to collect drawing scores on
every trial for every child with this design, rather than only on
about 60% of trials as in Experiment 1.

Participants
Eleven preschool children from Bing Nursery School
(3y10m – 5y0m, average 4y8m) participated in the study.
None had participated in previous versions of the experiment.
Two children were excluded as a result of the child not
finishing the study (n=1) or experimenter error (n=1).

Stimuli
The same seven two-tone and photo test trial pairs were
used as in Experiment 1. The same two practice trial image
pairs were also used.

Procedure
The procedure was highly similar to that described in
Experiment 1. Stage (1), uncued two-tone identification, and
stage (2), photo cue identification, were exactly the same.
The procedure differed in stages (3) and (4). In stage (3),
cued two-tone identification, the experimenter placed the two-
tone and photo images side-by-side, pointed back at the two-
tone image and said, “Let’s look at this picture again.” She
then explicitly told children that the two-tone and photo
images were the same. If the child had given the incorrect
label for the uncued two-tone image, the experimenter pointed
to the images in turn and said, “Actually, this is a picture of
the same thing as this. So there is a [name of object] in this
picture and in this picture” If the child had already provided
the correct label for the uncued two-tone image, the
experimenter still explicitly stated the two images were the
same, saying, “Okay! You said this is a picture of the same
thing as this. So there is a [name of object] in this picture and
in this picture.”
Because stage (3), cued two-tone identification, always
resulted in the child deciding or being told that the two-tone
and photo images corresponded, all trials concluded with stage
(4), drawing, which was described in the procedure for
Experiment 1 and conducted in the same way.

In experiment 1, the drawing measure was contingent on
183

Page 4

Results and discussion
On 16% of the test trials (SD = 0.15), participants labelled
the uncued two-tone image during stage (1) correctly. On the
remaining 84% of test trials, participants first gave an
incorrect label to the uncued two-tone image during stage (1).
However, at stage (3) the cued two-tone image was identified
using the same label used to identify the photo.
On these candidate perceptual reorganization trials, the
average drawing score was 0.33 (SD = 0.44). This score was
much lower than the average drawing score on practice trials
(0.89, SD = 0.13; t(8) = 10.3, p < .001, paired two-tailed t-
test), but did not differ from the average drawing score on
candidate perceptual reorganization trials in Experiment 1
(0.24, SD = 0.19; t(14) = 0.99; p = 0.34, unpaired two-tailed t-
test) when children were not explicitly told about the
correspondence between two-tone and photo image pairs. In
Figure 4, the drawing scores are plotted for all experiments,
collapsed across trials in which the two-tone image was
successfully recognized uncued and those trials in which it
was not successfully recognized uncued since these two trial
types did not differ from each other.
Deficient perceptual reorganization of two-tone images as
revealed by inability to draw corresponding parts was
therefore replicated in Experiment 2. In addition, the deficit
was found to be robust, persisting in the face of explicit
instruction from a knowledgeable experimenter about the
identity between the two-tone and photo images.


Fig. 4 Performance in drawing corresponding parts on image pairs. Data are
combined from trials when the two-tone image was or was not successfully
recognized without first seeing the photo-cue. Children are significantly worse
at drawing corresponding parts of two-tone and photo image pairs
(Experiments 1-3) compared to adults and compared to children drawing on
two identical photographs. Neither instruction about the identity between the
two-tone and photo images (Experiment 2) nor reducing perceptual
interference (Experiment 3) helped to improve drawing scores significantly.
EXPERIMENT 3

image, their incorrect initial interpretation may interfere with
When children fail to recognize the uncued two-tone
the formation of a new interpretation upon presentation of the
original image (e.g., [3]). The incorrect perceptual hypothesis
formed when children are asked to identify the uncued two-
tone image might be what is blocking the top-down
reorganization of the two-tone image that is triggered by the
corresponding photo cue. Although such “perceptual
interference” has been found in both adults and children [2, 3],
it is thought to be more pronounced in children and may have
contributed to their difficulty in reinterpreting two-tone
images in the previous experiments, since they were first
asked to evaluate the two-tones prior to seeing the cue. We
therefore removed the uncued two-tone presentation in order
to reduce perceptual interference, first showing children the
photograph, and then the corresponding two-tone image, thus
eliminating the chance to make a cue-free, incorrect perceptual
hypothesis about the two-tone image. If perceptual
interference is at all responsible for children’s low drawing
scores, this manipulation should result in improved drawing
scores, reflecting increased perceptual reorganization.

Participants
Twelve preschool children from Bing Nursery School
(4y2m – 5y1m, average 4y7m) participated in the study. None
had participated in previous versions of the experiment. None
were excluded.

Stimuli
The same seven test trial photos (kitten, dolphins, cheetah,
dog, polar bear, tiger, athlete) and two practice photos (house,
deer) were used as in Experiments 1-2.

Procedure
The procedure was a similar but truncated version of the
procedure described in Experiment 2. Instead of beginning
each trial with stage (1), uncued two-tone identification, all
trials instead began at stage (2) photo identification and
proceeded identically through stages (3) and (4) as in
Experiment 2. The procedure reduced possible perceptual
interference by omitting stage (1), uncued two-tone
identification.

Results and discussion
The average drawing scores on all test trials was 0.32 (SD
= 0.21) (Fig. 4). This score was much lower than the average
drawing score on practice trials (0.83, SD = 0.12; t(11) =
10.73, p < .001, paired two-tailed t-test), but not significantly
different from the average drawing score on all test trials in
Experiment 2 (0.33, SD = 0.19; t(19) = .20; p = 0.84, unpaired
two-tailed t-test), when children were explicitly told about the
correspondence between two-tone and photo image pairs but
asked to identify the two-tone image prior to seeing the
corresponding photo.
Reducing perceptual interference did not result in
significant improvement in perceptual reorganization (as
indexed by the drawing measure) in our task. It is possible that
a different task or more fine-grained scoring metric would
detect an improvement, but this improvement is likely small
184

Page 5

and subtle. While we cannot rule out the possibility that
perceptual interference may hinder cue-driven perceptual
reorganization, or results suggest that some other, more
powerful barriers remain.

EXPERIMENT 4

practice trials persisted in Experiments 2 and 3 in the face of
clear information from the experimenter about the relationship
between the two-tone and photo image pairs, we conducted a
control experiment to confirm the sensitivity of the drawing
measure. In Experiment 4, children were asked to draw
corresponding parts of identical gray-scale photographs. The
procedure was otherwise unchanged from Experiment 2.

Participants
Ten preschool children from Bing Nursery School
(3y9m – 4y11m, average 4y7m) participated in the study.
None had participated in previous versions of the experiment.
None were excluded.

Stimuli
The same seven test trial photos (kitten, dolphins, cheetah,
dog, polar bear, tiger, athlete) and two practice photos (house,
deer) were used as in Experiments 1-3. However, image pairs
for practice and test trials were identical gray-scale
photographs.

Procedure
Aside from the change in stimuli, the procedure from
Experiment 2 was duplicated.

Results and discussion
Average drawing scores were 0.94 (SD = 0.04) (Fig. 4).
Drawing scores for Experiment 4 were much higher than those
obtained through experimental trials in Experiments 1-3 and
were even marginally higher than those obtained through
practice trials in Experiments 1-3 (F(3,38)=2.76, p=0.06, one-
way ANOVA across four groups). These results support our
contention that the drawing measure employed in Experiments
1-3 is indeed sensitive and that the low scores obtained in
those experiments cannot be attributed to children’s
generalized inability to correctly draw corresponding parts on
image pairs.

Because the drawing score deficit on experimental versus
EXPERIMENT 5
Although we had informal evidence from adults that two-
tone images were trivial to recognize when placed side-by-
side with the corresponding photo cue, we wanted to quantify
the difference between children and adults in perceptual
reorganization using the same drawing measure used in
Experiments 1-4. We therefore tested adults using the same
procedure as Experiment 2 in order to obtain a drawing
measure for each trial.

Participants
Ten undergraduates from Stanford University participated
in the study. None were excluded.

Stimuli
The same practice and test image pairs were used as in
Experiments 1-3.

Procedure
The procedure from Experiment 2 was duplicated exactly.

Results and discussion
On 66% of the test trials (SD = 0.22), adult participants
labelled the uncued two-tone image during stage (1) correctly.
On the remaining 34% of test trials, participants first gave an
incorrect label to the uncued two-tone image during stage (1).
At stage (3) the experimenter identified the two-tone image
with the same label used to identify the photo. However, when
adults failed to identify the uncued image correctly, they
always spontaneously indicated that they could re-organize the
two-tone image before the experimenter provided the label.
For example, when a subject guessed that the two-tone polar
bear was “a car”, upon presentation of the photo the subject
spontaneously added “I mean a polar bear”.
Adults had significantly fewer candidate perceptual
reorganization trials than the children in Experiment 2 (2.4 per
adult vs 5.7 per child; t(8) = 6.5, p < .001), due to adults’
superior recognition for uncued two-tone images.
Average drawing scores for adults on all trials was 0.95
(SD = .06). Adults were equally good on the candidate
perceptual reorganization trials, with an average score of 0.98
(SD = 0.05).

DISCUSSION

children’s robust and dramatic deficit in perceptual
reorganization by asking them to draw corresponding parts of
two-tone and photo images. We found that even when children
spontaneously claimed the two-tone and photo images were
the same, or were explicitly instructed that the two-tone and
photo images corresponded to each other, the deficit remained
unchanged. Children were able to draw corresponding parts in
practice trials with blurred and contrast-altered practice
images, and in a control task with identical image pairs,
showing that neither deficiencies in drawing ability nor lack of
task understanding or motivation were responsible for the
effect. Finally, the use of simultaneous side-by-side
presentation of the photo and cue make it possible for children
to ‘cheat’ if they wanted to, and yet their drawing scores are
strikingly low.
The effortless success of adults in Experiment 5 is in
striking contrast to children’s failure. Even when they did not
recognize a two-tone image on its initial, uncued presentation,
adults often felt that it was impossible not to recognize a two-
tone image with simultaneous presentation of the photo cue.
In these experiments, we have documented preschool
185

Page 6

information to influence perceptual reorganization develop
and why is it so dramatically different in adults and children?
One possibility is the late maturation of top-down control of
perceptual processes, perhaps in the ability to inhibit
interference from alternative, cue-independent organizations
of the two-tone image. Reducing such perceptual interference
by relieving children from the requirement to report one of
these alternative organizations (Experiment 3), however, did
not lead to a measurable improvement in their ability to draw
corresponding parts. Difficulty inhibiting an initial cue-free
and possibly incorrect perceptual hypothesis about the two-
tone image therefore does not appear to be the primary
mechanism blocking perceptual reorganization.
Another possible explanation for children’s deficient
perceptual reorganization is that in order to use the photograph
as a cue for successful reorganization of the corresponding
two-tone image, young children must appreciate the dual
nature of the photograph as both a concrete object in and of
itself, and a transformed representation of something other
than itself—the two-tone image. That is, children may not be
able to simultaneously keep in mind the two-tone image and
photo as images in and of themselves, as well as
representations of one another that are mutually informative.
This difficulty in achieving dual representation may be what
prevents children from understanding that the photograph
corresponds to the two-tone image and from capitalizing on
this understanding to use the photo as a cue to help interpret
the two-tone image. Eliminating the need to achieve dual
representation has been shown to aid young children’s use of
symbolic objects such as scale models. For example,
DeLoache and colleagues [9] have demonstrated that children
fail to find a hidden toy when shown its location in a scale
model, but succeed if they are convinced that the model is in
fact the large room put through a shrinking machine. To
explore this possibility, we plan to test whether eliminating the
need for dual representations will rescue children’s
performance. We will show children photos that are
transformed by a special “blurring machine” so that children
believe that the two-tone images are in fact the photos, only
transformed.
A third possibility is that the photo cues in fact do trigger
perceptual reorganization of two-tone images in children when
they report that it does, but that the very act of drawing
somehow destroys the fragile reorganization that exists. For
example, the child who claimed to see the polar bear in the
two-tone image shown in Figure 3 may have in fact seen it
correctly, but upon attempting to draw the paw and nose
became distracted by salient features in the two-tone image
that encourage a photo-discrepant organization. If children do
in fact reorganize the two-tone images, but their
reorganizations are somehow more fragile or incomplete than
adults’, then more sensitive measures like eye-tracking could
reveal implicit consequences
reorganizational capacities present from much earlier in
development than supposed here.
How does the mechanism that allows top-down
of the perceptual

differences in basic perceptual processing and/or neural
maturation between children and adults and that cognitive
interventions such as those discussed above will not improve
children’s ability to see the figures hidden in two-tone images.
For example, long-range connections between retinotopic
visual areas required for low-level visuospatial integration are
thought to continue maturing through early adolescence [10].
Such long-range connections may be especially crucial in
organizing two-tone images, since figures are often comprised
of distinct contours spatially separated from one another.
Other candidate perceptual mechanisms required for
successful recognition of two-tone images include figure-
ground segmentation and Gestalt grouping. These possibilities
can be tested by looking for concurrent improvements in a
candidate perceptual process and in uncued recognition of
two-tone images, as the developmental trajectory of uncued
two-tone image organization should follow closely the
maturation of the required perceptual mechanism.
We look forward to further exploring these possibilities in
future work.

A final consideration is that there are fundamental
ACKNOWLEDGMENTS

stimulus preparation and data collection, the teachers and
children at Bing Nursery School, and the students and faculty
of the Stanford developmental, cognitive, and neuroscience
area for helpful feedback.

We would like to thank Caitlin McLean for helping with
REFERENCES
[1] I. Kovacs and M. Eisenberg, "Human Development of Binocular
Rivalry," in Binocular Rivalry, D. Alais and R. Blake, Eds. Cambridge,
MA: MIT Press, 2004, pp. 101-116.
[2] D. M. Bernstein, G. R. Loftus, and A. N. Meltzoff, "Object identification
in preschool children and adults," Dev Sci, vol. 8, pp. 151-61, Mar 2005.
[3] J. S. Bruner and M. C. Potter, "Interference in Visual Recognition,"
Science, vol. 144, pp. 424-5, Apr 24 1964.
[4] J. S. Deloache, "Becoming symbol-minded," Trends Cogn Sci, vol. 8, pp.
66-70, Feb 2004.
[5] E. B. Goldstein, Sensation and Perception, 7th ed. Boston, MA:
Wadsworth Publishing, 2006.
[6] R. L. Gregory, "The Medawar Lecture 2001 knowledge for vision: vision
for knowledge," Philos Trans R Soc Lond B Biol Sci, vol. 360, pp. 1231-
51, Jun 29 2005.
[7] M. J. Tovee, E. T. Rolls, and V. S. Ramachandran, "Rapid visual learning
in neurones of the primate temporal visual cortex," Neuroreport, vol. 7,
pp. 2757-60, Nov 4 1996.
[8] R. J. Dolan, G. R. Fink, E. Rolls, M. Booth, A. Holmes, R. S.
Frackowiak, and K. J. Friston, "How the brain learns to see objects and
faces in an impoverished context," Nature, vol. 389, pp. 596-9, Oct 9
1997.
[9] J. S. Deloache, "The credible shrinking room: Very young children's
performance with symbolic and nonsymbolic relations," Psychological
Science, vol. 8, pp. 308-313, 1997.
[10] I. Kovacs, P. Kozma, A. Feher, and G. Benedek, "Late maturation of
visual spatial integration in humans," Proc Natl Acad Sci U S A, vol. 96,
pp. 12204-9, Oct 12 1999.


186