The Japanese Psychonomic Society
NII-Electronic Library Service
TheJapanesePsychonomic Society
7Vze
fiipanese
Joitivaai
of
Ftl,i/h.onouaicSciettce
2002,
Vol.21,
No.
1,59-60Prizewinner's
SummaryP-2B-19
Is
depthimplicitly
encoded
as
visual
Jun-ichiro
KAwAHARA
Hiivshima
Udaiversily'
context?i)
The
spatialrayout
of objectscan
be
acqulred
implicitly
andfacilitate
visualsearch
under
incidental
learning
conditions
(the
contextual cueing effect;
Chun
&
Jiang,
1998).
The
present
studyinvestigated
whether
the
learned
layouts
arelimited
to
two-dimensional
configurationsor
can
encornpass
three
dimensions.
""e
found
that
the
reaction
tirnes
for
the
repetitioncondition
increased
significantly whepthe
disparity
of
the
repeatecldistracters
was reversedin
the
end ofthe
session.
These
resultsindicate
that
three-dimensiona]
layouts
canbe
preserved
asimplicit
visual spatial contexts.Key
words :contextual cueing, visual attentien,implicit
learning,
stereodisplay
Contextual
cueingis
aprocess
that
guides
visual
attentionby
enhancingthe
processing
of complex scenes.It
has
been
suggested
that
global
context,such as
the
]ayout
ofobjects,
is
extractedthrough
experience
in
a
stable environment.For
example,
when
participants
search
for
atarget,
the
visualsearch
performance
is
enhancedif
the
spatial
layout
ofitems
is
fixed
acrosstrials
in
comparison
withlayouts
which are randomlyvaried
in
everytriar,
This
enhan6ement,callecl
the
contextual cueing effect,is
obtainedimplicitly.
The
observers cannot recognisethat
a
specific
set oflayouts
have
been
presented
repetitively,
Researchers
have
shownthat
contextualcueing
can
be
obtained
withtwo
dimensional
(2-D)
spatial'
outs
(Chun
&
Jiang,
1998).
However
whether
this
effect can
be
obtained with3-D
displays
has.not
been
examined,Since
ourliving
spaceis
not only2-D
but
contains3-D
information
abeut
space, we assurnethat
a spatial
context
of
layouts
holds
in
a3-D
structure.We
therefore
examinedif
a3-D
structure
defined
by
'
Department
of
Psy6hology,
IIiroshima
sity,
Kagamiyarna
1-1-1,
Higashi-hiroshima,
s:r)241)
This
work was sponsoredby
a researchgrant
from
the
Japan
Society
for
the
Promotion
ofScience.
The
author
would
like
to
thank
Hiromi
Hosoda
for
collecting
data.
binecular
disparity
was eneocled as visual context.The
observerswere
presented
with visual searchtrials,
In
the
OLD
condition,
different
layouts
werepresented
repeatedly.In
the
NEW'
condition,
a
new
layout
wasgenerated
for
eachtrial.
In
the
last
set oftrials,
the
disparity
ofthe
distractors
in
the
OLD
condition was reversed.
If
the
observersformed
spatial contextsincluding
clisparity,
then
the
learned
layouts
would notbe
consistent withthe
layouts
after
the
disparity
reversal.The
searchperformance
on
trials
aiter
the
reversal wou]dtherefore
be
poor
in
comparison with
those
before
the
reversal.'
Method
Apparatus
a"dStimuli
The
stimuli weredis-played
on a computer rnonitoT whichthe
participants
observed
binocularly
through
ferroelectric
shutter
goggles.
The
target
a "T" which was rotated90"
to
the
left
erto
the
right.The
distractors
were
eleven
"L"s
rotated
eitherO[,
900,
1800,
or2700
from
the
upright.The
stereodisparity
was
O.5e.
Experimental
Design
and
Procedure
The
design
used
two
within-subject
factors:
layout
(NEW,
and
OLD)
and
block
(1-6).
The
visual search sessionhacl
a
total
of
720
trials,
and oneblock
consistecl of120
trials.
In
the
sixthblock,
the
disparity
of
the
dis-tractors
was reversed.The
recogriition sessioncontained
24
trials.
The
NEW
condition,in
whichthe
The Japanese Psychonomic Society
NII-Electronic Library Service
TheJapanesePsychonomic Society
60
The
Japanese
Jeurnalof
Psychonomic
location
of
all
the
items
wasdetermined
randomlyfrom
trial
to
trial,
was usedfor
half
the
trials.
The
remainder ofthe
trials
usedthe
OLD
condition.
Half
the
items
werepresented
in
the
frontal
plane
and
the
others
were atthe
back,
The
target
wasequally
Iikely
to
be
presented
eitherin
frontal
or atthe
back,
and
this
was unpredictable acrosstrials.
After
the
visual searchsession,
the
observers
engagedin
the
recognition sessionthat
consisted
ofthe
OLD
and
the
NEW
conditions<12
trials
ofeach).
The
ebservers
indicated
whetherthey
had
seenthe
display
in
the
visual
search
session,Sixteen
naive undergraduate studentsparticipatecl.
Refiults
Visual
searchThe
mean reactiontimes
for
all
ef
the
correcttrials
in
the
OLD
andthe
NEW
conditions,
as afunction
of
a
block,
are shownin
Figure
1,
A
two-v"'ay
analysis
of varjance was cenducted withtwo
within-subject
factor
(layout
:NEW,
OLD
;
block
:1-6).
The
analysis revealedsignificant
maineffects
oflayout,
F(1,
15)=6S.3,
P<
.OOI,
ancl
block
(F(5,75)==25.0,
P<.OOI),
anda
significant
interac-tion
between
layout
andblock
(F(5,75)=3,O,
P
<
,05).
To
exarninethe
effect ofthe
reversa] ofthe
diparity
'
'
1300
:t!Ig2E
1200
1100
1000
.goe
800
7001
Figure1.
The
rneanand
OLD
conditions
b]ock,
234S6
BLOCK'
reaction
times
in
the
NEW
,plotted as afunction
ofthe
Science
Vol.
21,
No.
1
of
the
distractors,
we conducted atwo-way
analysis of varianceon
the
reactiontirne
data
ofthe
last
two
blocks
withtwo
within-subjectfactor
(layout
:NEW,
OLD,
block:
5th
and6th).
The
analyses
showed
asignificant
maineffect
oflayout
(F(1,15)=63,1,
P
<
.OOI)
,and
the
interaction
between
layout
andblock
(F(L
15)
=-6.7,P<
.05)
.
Recognition
Mean
accuracy
in
the
recognitiontest
was50.2%.
Observers
correctly
indicated
the
OLD
layouts
as old on43.2%
of
trials
(hit
rate).
This
rateclid
notdiffer
from
their
false
alarmrate
of
42.7%
(F(1,15)-1.1,
n.s.).
Discussion
First,
our resultsindicated
that
a robust centextual cueing effect was obtained,Second,
the
resultsin-dicated
that
this
knowledge
of spatialIayout
was acquiredimplicitly.
Most
importantly,
the
learned
layouts
encompassed
three
dimensions.
We
expected adeteriorated
visual
search
performance
afterthe
reversal efthe
disparity
ofthe
distractors
if
the
visual contextthat
guided
attentionte
the
target
location
was sensitiveto
the
3-D
structure.The
results
in-dicated
this
was
true,
The
reactiontimes
increased
afterthe
reversal
of
the
disparity
ofthe
distractors,
whichsuggests
that
the
visual
context containsdepth
defined
by
stereodisparity.
The
present
studyrevealed
an
anisotropy
ef
spatial
tolerance
in
contextual eueingin
conjunction
withprevious
studies,
In
terms
of2-D
coordinates,Chun
&
Jiang
(1998)
showed robust contextual cueing againstjitters
which wereadded
to
the
distractors.
The
reversal of
the
disparity
of
the
di'stractors
introduced
in
the
present
study canbe
regarded
as
jitter
in
the
depth
clirection,
but
contextual cueing was reducedwith
thisjitter.
The
present
results suggestthat
the
visual
system
is
capable
of encodingdifferences
in
3-D
space ascontextual
cue
and
that
a
generalisation
in
location
may notbe
equivalentto
spatial
coerdinates.