P-2A-16 オブジェクト追跡処理に「無視された情報」が与える影響(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

TheJapanesePsychonomicSociety

lhe

1itPanes'e

fou7w.at

of

Ao'(/honomic

Science

2002,

Vol.

Zl,

No,

1,

49-50

Prizewinner's

SummaryP-2A-16

The

effects

ofthe

information

'multiple

object

Hirokazu

_OGAwA

Ktvansei

and

Gahblin

of

untracked

objects

on

tracking

Akjhiro

YAGi

Uitivei:sity'

The

present

study used a contextual

cueing

paradigm

to

investigate

contextual

information

in

a complex

visual

scene

which

changed

dynamical]y.

An

observer

had

_te

_truck

five

identical

_targets

which moved

independently

and unpredictably

among

five

identical

distracters

(a

rnultiple object

tracking

task),

The

motion

_patterns

(trajectories)

of

the

items

were

made

invariant

by

repeating

them

throughout

_the

entire

experimental

session.

The

results showed

that

repetition of

the

target

motion

_pattern

facilitated

_the

_tracking

_performance.

_A

more

irnportant

finding

was

that

repe･ating

the

motion

pattern

of

the

distractor

resulted

in

_greater

facilitation.

An

inciclental

recognition

test

confirmed

that

an observer could not notice

the

repetition

_procedure,

We

conclude

that

not, only

the

target

motion

_pattern,

but

also

the

distractor

motion

_pattem,

w,as encoded

as

contextual

information

in

an

irnplicit

manner,

Key

words :

implicit

learning,

contextual

cueing, multiple object

tracking

Recently

Chun

and

his

colleagues

(Chun

&

_Jtang,

1998,

1999)

have

developed

a

paradigm,

called

centex-tual

cueing,

te

investigate

how

a visua] context

is

'

formed

in

a

laboratory

situation,

In

their

studies

the

observer was asked

to

_perform

a

visual

search

task

in

which

the

configuration of

the

search

items

was

defined

as a context, and

_particular

display

layouts

were

_presented

repeatedly

across

the

blocks.

The

study

shuwed

that

search

performance

was

facilitated

by

a consistent

layout

witheut an observer's

aware-ness

of

such

a

procedure.

The

results suggestecl

that

the

ob$ervers

learned

the

configurat,ion as contextual

knowledge

in

an

implicit

manner.

In

the

present

study

we

investigatecl

the

nature

of

the

contextual

information

in

a

cornplex

visual scene which

changed

dynamically.

We

used a multiple

'object

tracking

paradigm

(Pylyshyn

&

Storm,

1988),

'

'

The

present

study

_posed

two

_questions.

First,

does

repetition of

the

motion

_patterns

(trajectories)

of

tracked

items

ferm

the

visual context

ancl

facilitate

'

_Department

of

Psychology,

Kwansei

Gakuin

University,

1'155,

Uegahara,

Nishinomiya,

Hyogo,

662-8501

the

_performance

of

multiple

object

trucking?

Second,

what

information

is

encoded

into

the

inemory as a

visual

context?

Method

Sixteen

undergraduate students

_{participatecl}

in

_the

experiment,

Figure1a

shows

the

sequence of events

during

a

trial.

The

display

consisted of

10

white

circle

items

(O.83

degrees

in

diameter),

on a

black

background

(14.7

×

14.7

degrees).

At

the

start of a

trial

5

of

10

items

were

designated

as

the

target

iterns

which

were

to

be

tracked

by

flashing

them

on, and off,

five

times,

After

the

targets

had

been

designated

each

item

started

to

move with avelocity

vector

that

was changed randomly every

few

hundred

milliseconds.

The

items

arso

changed

their

velocity veetors when

the

_distance

_from

_the

edge of

the

display,

or

from

other

items,

was

less

than

O.2

degrees.

A

smooth

motion

continued

for

7,500ms.

At

the

end

of

the

movement a meuse cursor appeared

at

the

center

of

the

display.

An

observer

then

had

to

indicate

five

targets

by

meuse

clicks,

Nb

feedback

was

given.

The

investigation

used

a

3

×

5

design,

Each

of

three

The Japanese Psychonomic Society

NII-Electronic Library Service

TheJapanesePsychonomic Society

50

a

The

Japanese

Journal

750ms

of

Psvchonomic

'

7500ms

Figure

1.

(a)

The

sequence of

events

during

a

trial.

function

of

the

epoch

for

each repetition

condition.

factor:

an "all

_old"

_conditien;

_an "old

_target"

condi-tion;or

an

"all

new" condition.

In

the

"all old" condition,

all

of

the

items

(targets

and

distractors)

had

motion

_patterns

which were

made

invariant

by

repeating

them

throughout

the

entire experimental sessien.

In

the

"old

target"

conditien,

the

pattern

of

the

target

items

was rnade

invariant.

In

the

d"all

new"

'

condition

all ef

the

motion

patterns

were new

genera-tions

for

each

block.

The

trajectories

for

the

first

second and

the

last

second of

the

display

period

were recalculated

for

each

repetition.

Another

'

tal

factor

involved

an

"epoch"

or

_group

of

trials.

'

Each

epoch was comprised

of

three

blocks,

each of which

jncluded

five

"all

old"

condition

trials,

five

`Lold

target''

condition

trials

and

five

"all new" condition

trials.

Aiter

the

last

block

of

the

tracking

task

the

observer was asked

to

perform

an

inciclental

explicit

recognitiQn

test

in

which

five

"all old"

patterns

and

five

newly

_generated

patterns

were

_presented

in

a

randorn

order.

The

observer was

required

to

answer whether

or

not a

_pattern

had

been

presented

in

the

preceding

tracking

task.

Results

and

Discuepien

Figurelb

shows

the

mean number of correctly

tracked

items

as a

function

of

the

epoeh

for

each

repetition

condition,

The

tracking

performances

in

the

"all

o]d"

and

the

`"old

_target"

cunditions were

better

than

those

in

the

"all

new" condition :`iall

old"

vs.

"all

new"

conditions,

F(1,l6)==28.40,

P<.eOl;

Science

Vol.

21,

No.

1

b

4.6 4.5:trt

4,4b

4.3Sij

42thmE

4.12 4.Dlt 3.9 1 2'3 4.' 5

Epoch

'

(b)

The

mean

number of correctly

tracked

items

as a

'

The

error

bars

indicate

the

standard

error

of

the

mean.

"old

_target"

vs. "all new"

conditions,

F(i,

16)=11.36,

p<.O05,

A

more

important

finding

is

that

mances

in

the

"al! old" condition were

better

than

those

in

the

"old

_target"

condition

(F(1,

16)=8.25,

P

<.05).

Although

7

of

16

observers

reported

that

they

were

aware of

the

repetition manipulation,

the

accuracy

of

expl{cit

recognition was

at

chance

level

(overall

accuracy=49.4%,

hit

rate=:42.3%,

false

alarm=

43.5%).

,

"Je

conclude

that

learning

of

the

motien

_patterns

occurred witheut

explicit

encoding uf

the

display

and

that

the

implicit

representations

of

dynamic

context

facilitated

tracking,

Furthermore,

the

results

show

that

the

motion

_patterns

of untracked

items,

which

were

to

be

ignored,

facilitated

tracking

performances.

The

present

results suggest

that

the

human

visual

system

encodes

complex

dynamic

covariation

between

objects

in

a visual scene and

this

encoding

serves

to

reduce uncertainty.

References

'

Chun,

M,M,,

&

Jiang,

Y.

1998

Contextual

cueing:

Implicit

learning

and

memory

of visual context

guides

spatial

attention.

Cagnitive

Ils),cholQgy,

36,

28-7L

Chun,

M.

M.,

&

Jiang,

Y.

1999

Top-down

attentional

guidance

based

on

implicit

leaming

of visual

iation.

i]kycholagicat

Scr'ence,

10,

360-365.

Pylyshyn,

Z.

W.,

&

Storm,

R.

W.

1988

Tracking

ple

independent

targets:

Evidence

for

a

parallel