An Analysis of the Solution of PAN Ambiguous-Cue Problem by Response Latency in Pigeons

(1)

The Japanese Psychonomic Society

NII-Electronic Library Service

The JapanesePsychonomicSociety

1'he

1dePanese

Jbzarnal

of

llsychonomic Science

19S2,

Vol.

I,

No.

1,59-62

An

Analysis

of

the

Solution

of

PAN

Ambiguous-Cue

Problem

by

Response

Latency

in

Pigeonsi)

Ben

YAGi

and

Naoto

KIMuRA

Aoyama

Gafeuin

U)tiversit.v

Pigeons

were trained on

PAN

ambiguous-cue

_problems

which

involved

three stimuli: a

positive

stimulus, a negative stimulus, and an ambigueus stimulus,

to

which responding was

sometimes reinforcecl and sornetimes nonreinforced

depending

on the

_paired

stimulus,

All

subjects attained

higher

level

of

_performance

in

terms

of correct respense on

the

trials

where the ambiguous stimulus was

paired

with the negative stimulus

than

on

the

trials

whera the ambiguous stimulus was paired with

the

positive'stimulus.

This

result could

be

interpreted

by

the

interfering

cffect ascribed

to

the

positive

property

of the ambiguous

stirnulus.

It

was,

however,

found

that

level

of performance

in

terrns

oi

latencies

of correct

responses

differed

markedly

between

the types oi

_preblems:

that

is,

responses

to

A

stimulus

on

NA

trials

had

longer

latencies

than

those

to

P

on

PA

trials,

This

finding

suggests some

interfering

effect ef alternative

_property

of the ambiguous stimulus,

the

negative

property,

which can

be

detected

by

measuring response

latencies,

Key

Words:

pigeon,

PAN

problem,

response

latency.

The

PAN

ambiguous-cue

_problem,

deviced

by

Thompson

_(1954),

is

a complex

discrimina-tion

_problem,

This

two-choice

discrimination

problem

involves

three

stimuli: a

_positive

or

consistently reinforced stimulus

(P),

a

negative

or

consistently nonreinforced stimulus

(N),

and

an ambiguous stimulus

(A),

which

is

_positive

or negative

depending

on which

of

the

other

stimulus

is

simultaneously

_presented.

Subjects

are

_presented

with

two

type6

of

trials,

PA

tria]s

and

NA

trials,

in

some mixed order.

When

each of

the

stimuli

_possesses

distinctive

properties,

much researches

indicated

that

human

and

nonhuman

primates

performed

significantly

better

on

NA

trials

than

PA

trials

(Fletcher,

Grogg,

&

Garske,

1968;

Boyer,

&

Polidora,

1972;

Fletcher,

&

Garske,

1972).

Recently

the

comparable results

have

been

obtained

by

using

_pigeons

as subjects

(Richards,

1973;

Hall,

1980).

This

superior

_performance

on

NA

_problem

was

explained

by

interfering-cue

hypothesis

_proposed

by

Zeaman

and

House

(1962).

According

to

this

hypothesis,

P

stim-ulus acquires

an

approach

tendency

as a result

l)

This

study wa$ supported

by

a

for

Scientific

Researches,

Ministry

Science

and

Culture.

Grant

in

Aid

of

Education,

of reinforced responses

andN

stimulus acquires

an avoidance

tendency

because

of nonreinforced

responses.

The

A

stimulus,

because

it

is

reinforced on some

trials

but

not on others,

acquires a net

positive

property,

if

one assumes

that

reinforcement

developes

approach

tend-ency which

is

relatively

stronger

than

corre-sponding

avoidance

tendency

resulting

from

nonreinforced

responses.

Thus,

an

inferior

perforrnance

is

predicted

on

PA

problem

be-cause

both

P

and

A

stimuli

_possess

approach

tendencies.

Although

the

interfering-cue

hYpothesis

ex-plicitly

emphasizes

a

_positive

property

of

A

stimulus,

it

may also acquire a negative

pro-perty

when

PA

as well as

NA

_problem

were

mastered on

later

sessions of

trainings.

No

study

has

indicated

any effect ascribed

to

the

negative

_property

of

A

stimulus.

The

_purposes

of

the

_present

experiment were

to

show

some

interfering

effect of

the

negative

_property

of

A

stimulus and

to

illustrate

how

animals

learn

ambiguous-cue

_problems.

For

those

pour-poses,

not

only

the

number of correct responses,

but

also

the

response

Iatencies

were

measured

in

this

experiment.

By

analysing

these

meas-ures

it

is

expected

that

the

approach-avoidance

(2)

The JapanesePsychonomic Society

60 The

Japanese

Journal

of

mechanism could

be

revealed

in

ambiguous-cue

_problerns.

Method

Subjects

Four

adult male

_pigeons

were used and

main-tained

at

80%

of

their

free-feeding

weights.

They

were all experimentally naive,

Water

and

_grit

were

freely

_available

at

their

home

cages.Apparatus

Birds

were

trained

in

an

_operant

condition-ing

chamber

on

one wall of which were

_two

response

keys,

each

2cm

in

diameter.

The

keys

were

_positioned

22

cm

from

_the

floor,

7

cm apart center-to-center, and could

be

illu-minated

from

behind

by

blue,

_green,

red, and

white

lights.

A

house

light

was

illuminated

throughout

each session.

This

chamber was

placed

in

a sound attenuating enclosure,

The

standard

_programming

and

_recording

equip-ments

were

located

in

an adjacent roorn.

Procedure

After

the

key.

_peck

response was autoshaped

to

a white

lighted

key,

Ss

received

additional

120

continuous reinforcements.

Only

one

re-sponse

key

was available

in

_this

_period.

Then

Ss

were exposed

to

the

discrimination

training

with

three

stimuli

that

differed

in

color:

P

or

N

stimulus

light

was either

_green

or red,

and

A

stimulus

light

was always

blue.

Each

trial

began

with

the

_presentation

of

a

_pair

of

stimuli,

PA

or

NA

key

lights,

and

terminated

after

one response.

A

correct response,

speci-fied

by

the

response

to

P

on

PA

_trials

_or

_to

A

on

NA

trials,

was

followed

by

the

reinforcement

of

3

sec access

to

hemp

seeds; an

incorrect

response,

_specified

by

_the

response

to

A

on

PA

or

to

N

on

NA,

_produced

no reinforcement.

The

interval

between

trials

was

always

10

sec.

Each

_problem

was

presented

30 times

in

daily

sessions.

The

order of

the

_problern

and

_the

side of correct stimuli were

determined

by

modified

Gellermann

_sequences

(Fellows,

1967)

with

following

restrictions.

First,

none of

_the

problem

could

be

_presented

more

than

three

times

in

succession.

Second,

no correct stimuli

could appear on

the

same

_side

more

than

three

times

in

succession.

Third,

no correct stimuli

could

appear

on

the

same

side more

than

two

Psychonomic

Science

Vol.

1,

No,1

times

in

succession

in

the

case

where

either

problem

was

_presented

three

tirnes

succes-sively.

The

discrimination

training

with

non-correction method was continued until response

latencies

stabilized

:

the

stabilization was

judged

when response

latencies

showed no

increasing

nor

decreasing

trends

over

3

consecutive

days.

At

this

_point

Ss

were

_presented

5

sessions

of

discrimination

training

between

P

and

N. Finally

there

followed

2 test

sessions

_to

_assess

the

latencies

to

each

of

three

stimuli

in

ex-tinction

condition.

One

of

the

three

stimuli

was

_presented

six

_times

in

_succession

on

both

keys.

A

session consisted of a

total

of

72 trials,

that

is,

24 trials

for

each stimulus.

Results

and

Discussion

To

compare

the

results

on

PA

problem

with

those

on

NA

_problem,

the

follQvving

analysis

was

carried

out:

sessions,

after

birds

reached

the

_point

of

80%

of

correct responses

on

NA

trials,

were

divided

into

6 blocks,

because

the

nurnber of

total

training

sessions

varied

among

subjects,

Three

_(S-919

and

S-921)

or

4

ses-sions

(S-922)

constituted a

block

described

above.

(The

results

from

S-916

were neglected

because

of

his

_perfect

_position

_preference

on

PA

trials.)

Figure

1 _presents

the

mean

dis-crimination

_score$

and

_the

mean correct

re-sponse

latencies

for

three

birds,

It

should

be

noted

that

Figure

1

shows

the

blocks

after

NA

_performance

was above

_the

level

of

80%

of correct re$pon$es.

For

PA

_problem

the

s ml'eO 00Q Oso

o-gZ

so og

:

0:

.70

n Z -7.6 X o m z m w v

?.s

2

" th m

123456 in BLOCKS

Figure

1. Mean

Iatency

(solid

lines)

and mean

percentage

(broken

lines>

of correct responses

en

PA

(open

circles) and

NA

(closed

circles)

(3)

B. Yagi

and

N

Kitnura:

results

in

two

blocks

_preceding

80%

_point

were also shown

in

Figure

1,

First,

it

is

clearly evident

that

acquisition of

PA

_problem

was markedly retarded

in

comparison with

that

of

NA

_problem.

To

reach

the

criterion

of acquisition

(11

correct responsesl12

trials),

Ss

required a mean of

80 trials

on

NA

pro-blem,

while a mean of

276 trials

on

PA

pro-blem.

In

addition,

the

asymptotic

level

of

performance

on

PA

problem

was slightly

below

that

on

NA

_problem.

These

results were

consistent with

the

findings

of

other

studies

using

_pigeons,

monkeys, and retardates.

Sec-ond,

latencies

of correct response

differed

markedly

between

the

types

of

_problems:

responses

to

A

stimulus

on

NA

trials

had

longer

latencies

than

those

to

P

on

PA

trials

throughout

all sessions,

in

spite

of

the

fact

that

Ss

easily

_and

_quickly

learned

_to

choose

A

stimulus

en

NA

trials.

Mean

latencies

of

12 trials,

during

which

the

criterion mentioned

above was reached, were

,67

sec on

PA

pro-blem

and

.75

sec on

NA

_problem.

An

analysis

of variance, carried out on

the

4 blocks

after

the

mastery of

_problems,

revealed a

signi-ficant

effect of

the

types

of

_problems

(p<.05).

Furthermore,

on

PA

_problem

latencies

in-creased

_gradually

as

trainings

_proceeded

and

then

decreased

constantly after

birds

learned

the

_problem.

While

on

NA

problem,

it

was

feund

that

latencies

continued

to

increase

for

two

blocks

after

birds

had

reached

80%

of

correct responses and

_then

decreased.

It

would

be

interesting

that

continuing

increment

in

latency

was observed

in

the

corresponding

blocks

where

discrimination

_scores

_on

PA

problern

were

improving.

A

rnean correct

response

latency

to

P

was

.52

sec

in

the

sub-sequent

discrimination

training

between

P

and

N. Response

latencies

to

each

of

three

stimuli

in

extinction were shown

in

Figure

2. Re-sponse

to

P

showed

_no

_sign

_of

increment

in

latency.

Latency

of response

to

A

was similar

to

that

to

P

in

the

first

session

but

it

was

lengthened

markedly

in

the

second session,

The

discrimination

scores revealed no

in-terfering

effect on

NA

_performance

ascribed

to

A

stimulus,

despite

the

fact

that

it

had

acquired

the

negative

_property

after

the

mas-tery

of

PA

_problem;

NA

_performance

was

maintained above

80%

of correct responses

PAN

Ambiguous-Cuc

Problem

61

4.oO N va 3.00 a Z 2.50 o O 2.00 u co 1.50

!

' o1.oo A z

:

f

P .5D T" 1 2 SESSIONS

Figure

2. Mean

response

latency

to

P,

A,

or

N

stimulus respectively

in

the

secutive extinction sessions.

throughout

trainings.

However,

correct

sponse

latencies

showed

the

facts

that

sponses on

NA

trials

had

longer

latencies

than

those

on

PA

trials,

and

that

continuing

crement

in

latency

on

NA

_problem

occured

concurrently with

the

mastery of

PA

_problem.

It

was

suggested

that

these

interfering

effects

of

A

_.stimulus

on

response

latency

could

be

attributed

to

the

fact

that

A

stimulus

had

acquired

negative

property

in

PA

problem.

These

findings

suggest

following

_processes

through

which

birds

learn

the

ambiguous-cue

problem:

responses

to

A

on

NA

trials

have

relatively

long

latencies

even

_at

_the

mastery

of

the

_problem,

It

may

be

due

to

the

choice

of

A

stimulus on

PA

trials

resulting

in

the

incorrect

response.

The

negative

_property

of

A

stimulus

interfers

with

NA

_performance

more and more as

PA

_performance

improves,

and

_produces

further

increment

in

correct

sPonse

latency.

But

this

effect of

A

stimulus

is

too

small

to

destroy

the

discrimination

of

NA

problem

already acquired.

Once

PA

as

well as

NA

_problem

were acquired,

it

follQws

that

correct

response

latencies

on

both

blems

gradually

decrease

with a

little

longer

one on

NA

_problem,

because

A

stimulus

is

negative

on

PA

problem.

A

number

of

studies

have

suggested

that

A

stimulus

interferes

with

PA

_performance

because

of

its

_positive

_property.

In

the

_present

experiment, response

latencies

_proved

that

A

(4)

62 The

Japanese

Journal

of

stimulus also acquirecl

the

negative

_property.

And

it

was suggested

that

this

A

stimulus

had

the

interfering

effect on

the

correct

re-sponse

latencies

on

NA

_problem.

References

Boyer,

W.N.

and

Polidora,

V.J.

1972

An

analysis

of

the

solution of

PAN

ambiguous-cue

_problerns

by

rhesus rnonkeys.

Learning

and

fi4btivation,

3,

325-333.

Fellows,

B.J.

1967

Chance

stimulus sequences

for

discrimination

tasks,

llsychogagical

Butletin,

67,

87-92.

Fletcher,

H.j.

and

Garske,

J.P.

1972

Response

competition

in

monkey's so]ution of

PAN

ambiguous-cue problems,

Learning

and

tion,

3,

334-3,IO,

Fletcher,

H.J.,

Grogg,

T.M.

and

Garske,

J.P,

1968

Ambiguous-cue

preblem

performance

of

chil-Psychonernie

Science

Vol.

1,

No.

1 dren,

retardates, and monkeys.

fozarnal

of

parative

and

Rhysiofagical

flsyckology,

66,

482. Ila!1,

G,

1980

An

investigation

of ambigueus-cue

learning

in

pigeons.

Animal

Learning

an(t

havior,

8,

282-286.

Leary,

R.W.

1958

The

Iearnjng

of ambiguous-cue

problem

by

monkeys,

American

JOurnal

of

kychelqev,

71,

718-724.

Richards,

R.W.

1973

Performance

of the

_pigeon

on

the

ambiguous-cue

_problem,

Bultetin

of

Rsp,chonomic

Sociely.

1,

445-447,

Thompson,

R,

1954

Approach

versus avoidance

in

an ambiguous-cue

cliscrimination

_problem

'in

chimpanzees.

]ournal

of

Comparative

and

siolqgical

Ilsychology,

47,

133-135.

Zearnan,

D.

and

House,

B.T.

1962

Approach

and

avoidance

jn

the

discrimination

learning

of

retardatcs,