P-2A-32 面特徴を媒介とした,探索,識別,検出,位置検出(2001年度 日本基礎心理学会第20回大会優秀発表賞)

The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomicSociety

the

Jaj,a'nese

Jo"rnat

of

1+l}'ehonotnic

Sc'ience

2002,

VoL

21,

No.

1,53-54

Prizewinner's

SummaryP-2A-32

Theeffect

of

the

surface

detection,

Takahiro

medium

on

search,

i

and

localization

KAwABE

and

Kayo

K),ushu

Uhiversity'

MIuRA

dentification

)

In

the

visual search

paradigm

using

shading

stimuli

it

is

suggested

･that

concavity

is

more

dominant

than

convexity.

However,

in

a

figure

and

ground

segregation

task,

the

opposjte

is

true.

We

proposed

that

a

discrepancy

in

dominance

could

be

based

on whether

or

not

subjects

identified

the

properties

of

objects

such

as shape and orientation,

and

we confirrnecl

this

in

four

experiments.

The

results of

the

first

and second experiments

showed

that

searching

for

and

iclentifying

items

composed of cunvex

disks

was

more

eficient

than

items

composed of concave

disks.

The

results of

the

third

and

fourth

experiments showed

that

in

detection

and

]oca]izatien

tasks,

there

were no

significant

differences

in

the

dominance

between

convexity

and

concavity.

These

results suggest

that

convexity

is

dominant

when

subjects

conduct

a

task

requiring

identification

of

the

_properties

of objects,

and

that

the

visuar system extracts uppropriate

information

according

to

task

demand.

Key

words : shape

from

shading,

search

asymmetry,

aominance

Introduction

The

visual

system

must

extract elementary

fea-tures

contaiping

meaningful

information

about

the

structure

of

the

environment,

This

function

of

the

visual

system

has

been

investigated

in

many studies,

especially

in

visua] search

_paradigms.

In

a

visual

$earch

task

using shading

stimuli,

a

target

of a

bottom-rit

disk,

_perceived

as

`concave',

pops

out among

distracters

of

top-lit

disks

_perceived

as

`con-vex'

but

not

vice

versa

(Kleffner

&

Ramachandran,

1992).

It

has

been

suggested

that

this

asymmetry

is

evidence

_that

_the

visual

system registers concavity as a more

important

and salient

feature

than

convexitv.

However,

the

dominance

of convexity

has

been

also

reported.

Our

_previous

study showecl

_that

in

figure

and

_ground

segregation,

the

region

composed of

con-vex

disks

tends

to

be

_perceived

as

figure

in

compari-son with

the

region composed oi concave

disks.

In

this

study,

we

explored

the

dpminance

in

shading

perception

when subjects were required

to

iclentify

'

_Behavioral

schools

of

University,

kuoka,

812･

and

Health

Sciences,

Graduate

IIuman-Environment

stuclies,

Kyushu

6'19-1,

Hakozaki,

Higashi-ku,

Fu-･8581

properties

of

the

objects.

Experiments

Method

Subjects

Each

of

Experiments

l

and

2

were

per-formed

with

6

observes and

5

observes were

involved

in

each of

Experiments

3

and

4.

Apparatus

All

stimuli were

_preserted

on

a

19-inch

coler rnonitor with a resolution

of

75

Hz

(Iiyama).

An

Apple

computer

(Power

Macintosh

G3

MT266)

controlled

the

_presentation

of

the

stimuli

and

the

response

reglstratlen,

Stimuli

Figure

1

shows

the

stimuli used

in

Experi-mentl.

The

entire

display

subtended

120

×

120

in

visual angle.

Each

stimulus was composed of

15

×

15

shaded

disks

with a

_gray

background

whose

luminan-ce was

35cdlm2.

Each

disk

subtended

O.8e.

The

luminance

of

the

brightest

region

in

a

shaded

disk

was

73cd/m2,

and

the

luminance

of

the

darkest

region

was

10

cdfmZ.

The

search

items

(Experiment

1)

and

the

target

surface

(Experiment2,

3,

and

4)

were composed of

top-lit

_disks

when

the

background

was _{composed of}

bottom-lit

disks,

and

vice

versa.

Procedure

The

following

Experiments

were con･

ducted

in

a

darkened

room,

To

ensurb

the

vertical

The Japanese Psychonomic Society

NII-Electronic Library Service

TheJapanesePsychonomic Society

54

The

Japanese

Journalof

PsychonomicScience

Vol.

21,

No.

1

Table

1

The

results of

the

experiments.

The

data

represent

mean reaction

times

(ms>

and

Standard

Error

Task

(experiment

nurnber)

Disk

types

convex

cencave

(a)

(b)

Fjgure

1.

The

stimuli which were used

in

ment1.

(a)

In

this

_panel

the

search

items

are

composed

of

convex

disks

whereas

the

ground

is

composed of

the

cencave

disks.

(b)

In

this

panel

the

relation

between

the

items

and

the

background

is

the

reverse of

that

shown

in

(a).

Search

(experimentl,

Identification

(experiment

2,

Detection

(experiment

3,

Localization

'

<experiment

4,

N-6)

N-6)

N-5)

N-4)

631(

±

64.7)

682(r.64.2)

498(

±

38.5)

443(

±

30.5)

748(

±

75.9)

735(

±

_44.0>

531(

±

48.6)

447(

±

32.8)

alignment of

the

head,

a

chin

rest

was

utilized.

In

Experimentl,

subjects searched

for

a

target

com-posed

of

three

shaded

disks

arranged vertically or

obliquely

among oblique or vertical

distracters

respectively.

In

Experiment2,

subjects

identified

the

orientation

(vertical

ur

horizontal)

of a surface com･

posed

of

4

×

3

shaded

disks.

In

Experiment3

the

subjects

tried

to

detect

a surface composed of

3

×

3

shaded

disks.

In

Experiment

4

the

subjects reported whether asurface

composed

of

3

×

3

_shaded

disks

was

presented

to

the

right

or

left

of

the

visual

field.

Results

Table1

shows

the

results of

Experiments

1,

Z,

3,

and

'

'

4,

The

data

represent mean reaction

times

from

correct responses.

The

statistically

significant

levels

described

below

are

from

the

analysis

of

variance

(ANOVA).

In

Experiment

1,

we confirmed

that

items

composed

of

the

top-lit

disks

were searched more readily

than

items

composed of

bottom-lit

disks,

P

<.Ol.

In

Experiment

2,

convexity

was

dominant

in

identification

of

the

surface,

P<.Ol.

However

in

Experiments3

and

4,

we

could

not obtain a result

indicating

convexity,

P>.05.

Discussion

'

In

this

study we examined

dominance

in

shading

perception

in

search,

identificatien,

detection,

and

'localization

tasks.

In

the

search

and

identification

tasks

(Experiments1

and

2),

convexity

was

domi-nant, on

the

other

hand,

this

dominance

was not observed

in

the

detection

and

localization

tasks.

The

results

in

this

study may

imply

that

the

way

to

extract

a

feature

from

a

scene

varies

according

to

the

problems

we

have

to

solve.

In

anorrnal search, where

display

is

not

mediated

by

the

surface,

we

can

locate

the

target

without consciously

identifying

its

shape

or

orientation.

Therefore,

the

saliency

is

likely

to

be

determined

by

a

deviation

from

the

standard.

How-ever,

in

identifying

such

object

properties

as shape or orientation,

the

saliency

based

on a

deviatiun

from

the

standard

is

useless.

It

is

easy

to

identify

the

front

shape of an object

in

comparison

to

the

rear shape.

Therefore,

in

identifying

the

_properties

of

the

object, cenvexity

seems

to

be

important.

References

Kleffner,

D.,

&

Ramachandran,

V.S.

1992

On

the

perception

of shape

from

shading,

PeK'option

&

'