移動する「のぞき窓」の背後にある物体の知覚(第23回大会 優秀発表賞抄録)

The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomic Society

Thelapanese

.fournal

ofl-]ychonomic Science

2005,Vol,24,No.1,131-132

Summary

ofAwarded

Presentation2P45

Searching

for

luminance-defined

objects

through

moving

apertures

Takahiro

KAwABE

and

Kayo

MiuRA

K],ushu

[inivet:sity*

InExperiment

1

theobservers searched fora bright,static square arnong three

dark

squares,

The

search display consisted of tenframes and was covered with

dynamic

grayscale noise which

had a random or unidircctional motion.

The

aperture was present during different_percentages of

noises. By measuring

d',

it

was revealed that

detection

was more sensitive inunidirectional, rather

than randorn, noise,

This

result suggested that temporal luminance summation _played a rniner

role

in

detection.In Experiment 2 thc observers searched fora

bright,

vertical baramong threc

dark

vertical bars inthe_presence of dynamic noisc which was perpendicular, para]leLor random,

to the bar orientation.

Detection

was most sensitive when the movement of

the

noise was

perpendicular. The result suggested thatsuccessive visual transientswhich were visible through

the apertures, and which were unrelated to rnovement of the noise, were critical for target

detection,

Key

words: aperture vision, visual transients,surface segregation

Introduetion

When

a moving object isvisible through only a

thinaperture, we often _perceiveitsentire appearancc

even though a complete image of the object

is

not

presented simultaneously. This

kind

of

_perception,

called `aperture

viewingL has been an interesting

topic

for

vision research. Although _the underlying

mechanism of typical aperture viewing

has

been

studied extensively,

how

the visual system detects

the appearance of static objects through moving apertures isunclear.

The aim of thepresent study was toexamine what

visual

information

is

necessary fordynamic apeTture viewing,

in

other words, forthe

detection

of

informa-tion concerning

the

luminance of static objects

which are viewed through moving apertures. In

Experiment

1,

we observed a minor involvement of

luminance summation. InExperirnent

2

we observed

that

detection

performance was

dependent

on the

segregation of spatiotemporal patterns of visual

transients from the occluder. More specifical]y,

when the spatiotemporal _patternsofthe visual

.sients

were more salient, the performance insearch

tasks

was

higher.

The role of occlusion _perception driven

by

spatio-* _{User Science Institute,}

_Kyushu

_{University,6-19-1}

Hakozaki,

Higashi-ku,

Fukueka

812-8581

temporal integrationof visual transients indynarnic

aperture viewing, isalso

discussed.

Method

Observers Four people incLudingtheauthor

(TK)

participatedinthe firstand second experiment,

re-spectively.

They

had

cerrect or corrected-to-normal

visual acuities. Apart from the firstauthor

(TKL

al]

observers were unaware ofthepttrposeof the

experi-ments,

Apparatus

Apersonalcomputer{SonyVAIO,Ja-pan) controlled the _presentation of stimuli and

cor-rection of data. The stimuli were _presented on a

gamma-corrected,

19-inch

CRT display

_(Nanao

Flex-Scan

T761,

Japan)

with a resolution of

1024

×768

pixelsand a refresh rate Qf

75

Hz,

Stimuli

A

motion sequence consisting of

10

frames

was _presented tothe observers ineach triaL

Each frame included a

20

×

20

_{grayscale pattern} of

random-dot$,

Each

sequence contained a search

dis-play which was covered with dynamic noise. The

dots

in

thedynamic noise were categorized

into

ei-ther

`noise

dots'which occluded search displays,or

`aperture

dots'

which passed over them, The search

display

in

"present"

trialsconsisted of one bright,

square target,and threedark,square

distracters.

_In L`absent"

trials

the

search

display

consisted of

four

dark,

square

distracters.

Inthe firstexperiment, two

The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomic Society

132

The

Japanese

Journal

ef Psvchonornic

Science

Vol.

24,

No.

1

-e-UNIDIRECI'IONAL

･Hm.RANDOM

4 4

..t.t.ng

3 ,s' 3

,.,g

₂ is 2 ltt ft" 1 .･ 1

tt.

･'

o o

m,F?ERPENDICULAR

-x-PARALLEL

.dits.RANDOM

ytf'dettt'ttt..t

./

2o 4o 6o so loo lo 2o 3o 4b so A B PERCENTAGES OF APERTURES

ON

EACH

FRAME

<%)

Figure

1.

Panels

A

and

B

show the results of

the

first

and second experiments,

respective-ly.Error bars denote standard errors ef the

mean

{IV="4).

types of dynamic noise

(uni-d{rectional

and random

noise) were superimposed on thc

first

frame

(Frame

1)

of thesearch

display.

The

number of aperture dots was selected randomly to bc either 20,40,

60,

80,

or

1OO% ofa totalof

400

dots.

The

luminance

values of

t.henoise

dots

were randoml}, selected from 128

equ-allystepped luminance values ranging from 25to

75

cdfm2. In the unidirectional noisc, a mExing of the

two motion directic)ns_prevented motion aftereffects.

Each

frame

was presented

for

13,3ms

(1

vertical

retrace interval}and thus one sequence was

pre-sented for133 ms. The speed of themoving

dots

was

14

degrees

per second.

In

the second expcriment

threctypes ef dynamic noise were used:

perpcndicu-lar,parallc], and random noise.

Procedure The ubservers were asked to

judge

whethcr, or not the search

display

contained a

bright,

square targetamong three dark,squarc

dis-tracters.Each experimental condition was repeated

40

timcs

(20

trials

for

the"target-present"

condition

and

20

for

"target-absent").

Results

Experiment

1

A

two-way repeated measures

ANOVA

with the type of dynam[c noise

<2)

×

aper-ture probability

{5)

as factorsrevealed that a main

effect of thetype of noise was marginally significant

(F{1,

3}=9,417, p<O.1) and the main effect of

aper-ture_percentage was definitelysignificant

(F(4,

12)=

60.132,

_p<O.001}.

The

interaction

between

the two

factorswas also significant

_(F(4,

12)=4.430,p<O.05).

A

post-hoc analysis of the interactionindicatedthat

d'

was significantly

different

between

_{thetwo types}

of dynarnic noise when the aperture percentage was

40%

and

60%

(F(1,

15)=7.451 and 22.677;

P<O.05,

and O.OOI,respectively).

Experiment

2

A

two-way repeated measures

ANOVA

with the type of dynamic noise

(3)

×

aper-ture probability

(5)

as factorsrevealed that a main

effect ef the type of

dynamic

noise was signjficant

(F(2,6>=31.231,

p<O,OOOI).

Multiple

comparison

tests

(Ryan's

method)

indicated

that the d'in

the

perpendicu]ar motion condition was significant]y

differentfrom that intheother two conditions

ip<

O.OOI),

The

main effect of the aperture percentagc

was significant

(F(4,

12)

-:30,778,

_P<O.OOOI),

and

in-teraction between thc

two

factors was also

sig-nificant

(F(8,

24)=9.879,P<O.OOO1),

Discussion

This study examined how

the

visual system

de-tectstheluminance of static objects which are

view-cd through moving apertures inthe _presence of

dy-namic luminance noise

(dynamic

aperture viewing).

Inthe firstexperiment, we observed that

luminance

surnrnation was only a marginal factorin dynamic

aperture viewing because

the

performance was

bet-terwhen the nojse was uni-directional, rather than

random, when the _percentage aperture was 40 or

60%.

In

thesecond experiment theresults indicated

thatdetcctionof theluminance of static objects was

moTe successfu[ when the nperture moved

perpen-dicular tothe search

items

rather than para]lel to

them. The signals of spatiotemporal integrationof

themovement thereforehindered detectionof a

tar-get with the same orientation as the path of the

movement. The results

indicate

that the

spatiotem-poral

integration

of salient visual transients_playsa