The Japanese Psychonomic Society
NII-Electronic Library Service
The JapanesePsychonomic Society
V'heJaPaneseJ?)umaloflts'.vchonomicScience
20IO,Vol.29,No.],69 70
Summary
ofAwardedPresentation5-113
The
influence
of
duration
on
the
perception
of
gaze
direction
and
being
looked
at
Yutaka
NAKAJiMA*,
Yusuke
TANI*2,
Yuko
IsoGAyA*2,
andTakao
SATo*2
lhe
Uhiversits,of
Toleyo'・
*2To investigatehow quickly we can perceive being watched by other people,we evaluated the
effect of
duration
on the quantitativejudgment
of gazedirection
and the perception of "beingLooked at", The stimu!i were models ef heads generated by acomputer, and thcrotation angle and
gaze directionof each head was varied indepenclently,
The
participants were asked tojudge
whether the stimulus was looking at them
(being-looked-at
judgment),
and tojudge
where thestimuLus was looking
(quantitative
gaze directionjudgment).
The
resu]ts revealed thatneither thedirection
thatgives thebest
perception ofbeing-looked-at,
nor the perceptually straightdirection
of the participantswere affected by the duration(100
ms, 1000 ms, or an unlimited duration}.These results indicatethatthe
judgment
of whether or not alooker's
gazeis
directed
towards one completes at an earlylevel
in
thegaze perception process and within100
ms.Key
words: gazedirectien,
perception ofbeing
looked
at,duration.
The perception of "being looked
at"'
is
one of theimportant
methods of communication, Itis
known
thata person who islooked at perceivcs a looker's
gaze as
"being
looked at" when the looker'sgazeis
directedtotheirfaee
(Gibson
& Pick.1963)or within3
degrees
of theirnose.The
perceptionis
also inde-pendent of theviewing clistancc(Sato
&Matsuzaki,
2eOl). Not only the gaze itselL
but
also the looker'shead
rotation affeets thegaze
perception.
When
the tooker'shead isturned to therighL the gaze directionthat givesthe best"being-looked-at"
feeling
shifts tothe left,relative tothat of a frontal direction,and
vice versa
(the
head-turn effect,Anstis,
Mayhew,
&
Morley,
1969).As
noted abovc, prev]'ousstudieshave fecused on the spatial property of gaze
percep-tion,
but
few
have
investigated
the temporalprop-erty. Two questions may beasked. First,does the
"being-looked-aC'
perception occur after the
judg-ment of gazedirection?
Or,
does
the processing ofthisperccption and the
judgment
of the looker'sgazedirectionoccur simultaneously?
The purpose of this study was to examine the
temporal property of gaze perception by varying the
presentation time
(duration)
of a stimulus.The
par-* IntelligentMode]ing Laboratory, The
Universi-ty of Tokyo, 2ml1-16,Yayoi,Bunkye-ku, Tokyo
113-8656
Copyright201O,The
japanese
Psychonomic Societyticipantsperforrnedtwo tasks,
The
first
was tojudge
whether or not the
gaze
was directedtowards them(being
looked at, or not). The second task was toquantitatively evaluate the gaze angle.
The
resultsof these tasks were compared, so that the
relation-ship
between
theperceptionof"being
looked
at"andgaze directioncould bcclarified.
Methods
Four types
(2
males and 2females) of head modelswere prepared in Poser 7
(Srnith
MicroSoftware,
Inc.),and used as stimuli. They were presented ina
random ordcr on a
17-inch
CRT
display
inallexperi-mcnts. We varied thehead rotation angle and gaze
direction
of the models independent]y, The rotationangles were
-30,
-15.
0,
+15, and30
degrees,
andthe gaze djrectionswere
-18,
-15,
-12,
-9,
-6,
-3,
O, +3, +6, +9, +12, +15, and +18 degrecs,respectively.
We
defined
the straight-aheaddirec-tiontothc participantsas Odegree.
Positive
valuesindicated a head rotation towards the right, and
negative values to-,ards the left,from the partici-pants'viewpoinL
The
sizeof thestimuli was close toan actual human
head
size.The
durations
werefixed
within each experimental block and was 100ms,
1OOO ms, or unlimited. Fer the being-looked-at
judg-ment, the participantsused
3AFC
method(leftward,
Allrights reserved.
The Japanese Psychonomic Society
NII-Electronic Library Service
The JapanesePsychonomic Society
70
The
Japanese
Journal
of PsychonomicScience
Vol,"being looked
at", rightward} to
judge
whether the stimuLus waslooking
at them.Each
stimu]us was presented 20 times. For the quantitative gazedirec-tion
judgment,
they were asked toestimate the gazedirection
by
moving a marker placedin
front
of them(cf.
Anstiset al. 1969),Each stimulus was pre$ented4
times.Eight
(being-looked-atjudgments)
orsix(gaze
direction
judgments)
naive participantswith normal or corrected-to normal vision participated.All
ofthe
experiments were conducted ina dark room.
Results
For the being-looked-at
judgrnent,
we firstcalcu-lated
the ratio of thebeing-looked-at
rcsponsesfor
each condition and participant, We then estimated
the
perceived gazedirection
ofthe
looker andthe
maximum ratie of the perceivedbeing-looked-at
re-sponse, by
fitting
Gaussian functions to thecalcu-lated
data,
The averageddata
across allparticipantsare shown
in
Figure IA. These results show thatadifferenceof
duration
dicl
not affect the perceivedgaze
directien
of the looker.In
addition, themaxi-mum ratio at a
duration
of 1OO ms was significantlylower
than at any otherduration,
especially whenthe
head
orientation was ±30
degree,
For thequantitativeestimation, we calculated the
average of the
judged
gazedirection
for
each-?
A
e .ts! 2.ii?
ek1
ls 3bISio
Ee.E
tio.1lg r osFeee engla (deg}
Figure 1.
Results
of receivedjudgments. ments. iIi・llit/
;..-.i.;Ii':'-./xtt..-//-i'tttttt;ttt;t.t.ttl.t/・,gli:'.''={l,1..Ll'
''t-tt:rr,:i'
'' 11,{''i'fr.{i''
"a-{salssu -sg,
i,
gl
B
15}i'
v
.30
"15
O 15 Feoeangle{deg) experiments.A.
gaze-B.
gaze-directien judg-se29,
No.
1
tion and participant,and estimated the perceived
straight djrection
(which
we defined as a perceptual zeroclirection)
and regression coeMcient(which
is
regarded as accuracy index) with
liner
regressionfrom
the
averageddata.
The
averageddata
acrossparticipantsare shown inFigure IB. These results show that the
duration
did not have a significanteffect on a perceptual zere
direction,
whereas theregression coethcient at a
duration
of100ms
wassignificant]y Iower than at any other
duration.
Discussion
Our results suggest thatittakesup to 100ms to
aceomplish the
judgments,
whether or not thegazeof a looker
is
directed towards oneself.This
is
be-cause boththe
gaze direction that gives the be$tbeing-looked-at
responses and the perceptual zerodirectionwere not affected
by
adifference
ofdura-tion.
Thedecrease
of the ratio ofbeing-looked-at
responses fora duration of
100
ms andfor
rotatedheads might be related to the higher accuracy of
quantitative
judgment
at the100-ms
duration.
When thedurations were 1OOO ms andlor unlimited,
the gaze
direction
judgrnents
were less accurate:more precisely,overestimation of the
looker's
gaze directionoccurred at thesedurations. Thisoveresti-mation may serve to
distingui$h
a gaze toward andaway
from
participants,which leadstoimprovementof the
S/N
ratio.Accordingly,
we propose thatthediscrete
judgment
of whether or not alooker'sgazeis
directed
toward one, would eccur at an earlier levelthan the
level
where the precisequantitativcjudg-ment takesplace. The results of such a quantitatjve
judgment
may feed back and modify the finalper-ception of gaze direction.
References
Anstis,
S.
M.
Mayhew,J.
W.
&
Morley,
T.{1969).
The perceptionof where aface
or television"portrait"islooking.American
fournat
of
Rsychoiogy,82,489.Gibson,
J.
J.
&
Pick,
A.
D.
(1963),
Perception ofother person's
looking
behavior.
Americanlbunzat
of
Rsychotogy,
76,386-394,
Sato,
T,,
&
MatsuzakL
N,
{2001},
Effects
of viewing