Keller, Schoenfeld, Cumming, and Berryman as Instructional Stimuli

The

Behavior

Analyst

1989,

12,

221-225

No.

2

(Fall)

Keller,

Schoenfeld,

Cumming,

and

Berryman

as

Instructional

Stimuli

John

A.

Nevin

University

of

New

Hampshire

When

the

Association

for

Behavior

Analysis

arranged

a

symposium

series

honoring

Fred

Keller

on

his

90th

birth-

day,

I

was

delighted

to

be

asked

to

par-

ticipate.

Thirty

years

earlier,

Keller

had

guided

my

entry

into

the

experimental

analysis

of

behavior

at

Columbia,

and

he

still

stands

as

a

model

for

much

of

my

professional

life.

Nevertheless,

when

I

was

asked

for

a

title,

I

was

at

a

loss.

I

had

no

notion

of

what

I

would

say

at

the

sym-

posium,

but

knew

that

the

title

would

determine

the

substance,

if

not

the

de-

tails,

of

the

talk.

As

I

considered

possible

titles,

I

found

myself

revisiting

Scher-

merhorn

Extension

in

reverie.

I

had

first

entered

its

dark

and

dingy

halls

in

search

of

Nat

Schoenfeld's

office

on

the

third

floor.

I

had

no

background

in

psychology,

but

was

interested

in

psychophysics

and

color

vision,

so

I

had

made

an

appoint-

ment

to

discuss

the

possibility

of

grad-

uate

study

at

Columbia.

Bill

Cumming

was

with

Schoenfeld

when

I

arrived

and

they

interviewed

me

together,

taking

turns

in

probing

my

background,

challenging

my

half-formed

ideas,

and

explaining

Columbia's

program.

Evidently

I

passed

muster,

and

Keller,

then

serving

as

de-

partment

chairman,

admitted

me

on

condition

that

sooner

or

later

I

take

the

Graduate

Record

Exams.

(He

never

spec-

ified

a

deadline,

and

I

never

took

the

ex-

ams.)

I

began

taking

courses,

and

grad-

This

paper

is

based

on

a

talk

presented

at

a

sym-

posium

on

"Columbia

University:

Discriminative

Stimuli

and

Establishing

Operations"

at

the

meet-

ings

of

the

Association

for

Behavior

Analysis,

May

1989.

All

of

us

who

participated

in

this

symposium

are

indebted

to

Celia

Wolk

Gershenson

for

bringing

us

together

to

celebrate

the

special

sense

of

excite-

ment

and

common

enterprise

that

Fred

Keller

gave

us

during

our

time

at

Columbia.

Correspondence

should

be

addressed

to

the

author

at

the

Depart-

ment

of

Psychology,

University

of

New

Hampshire,

Durham,

NH

03824.

ually

became

familiar

with

the

cast

of

characters

in

Schermerhorn

and

the

ideas

that

excited

them.

Within

my

first

semester,

I

started

working

for

Bob

Berryman,

building

ap-

paratus

for

his

and

Cumming's

studies

of

complex

discriminated

operants.

As

a

re-

sult

of

their

analyses,

they

argued

for

a

separate

"instructional"

function

of

stimuli

as

selectors

of

discriminations.

Could

this

be

a

metaphor

for

the

role

of

my

mentors

at

Columbia?

Why

not?

Hence

the

title.

Beginning

in

1959,

the

Cumming-Ber-

ryman

laboratory

on

the

second

floor

of

Schermerhorn

Extension

was

largely

de-

voted

to

exploring

variations

of

the

now-

familiar

matching-to-sample

procedure;

the

notion

that

the

sample

stimuli

in

that

paradigm

might

exercise

a

special

func-

tion

was

in

the

air.

As

Cumming

and

Berryman

(1965)

analyzed

it,

the

match-

ing-to-sample

paradigm

involved

two

si-

multaneous

discriminations

and

a

su-

perordinate

successive

discrimination.

Consider

a

pigeon

confronting

red

and

green

lights

defining

a

simultaneous

dis-

crimination

on

the

side

keys

of

a

three-

key

panel.

In

the

first

such

discrimina-

tion,

red

was

SD

and

green

was

SA;

in

the

second,

green

was

SD

and

red

was

SA.

The

superordinate

successive

discrimination

involved

the

color

of

the

sample

on

the

center

key.

In

a

matching

contingency,

a

red

sample

presented

on

some

trials

sig-

naled

that

the

first

discrimination

was

operative,

and

a

green

sample

on

other

trials

signaled

that

the

second

was

op-

erative.

In

an

oddity

contingency,

the

color

relations

were

reversed,

and

in

"symbolic"

matching,

the

color

relations

were

arbitrary

(for

example,

if

the

sample

was

blue,

the

first

discrimination

was

op-

erative,

and

if

yellow,

the

second

was

op-

erative).

Thus,

at

a

procedural

level,

the

221

222

JOHN

A.

NEVIN

sample

could

be

said

to

select

the

appro-

priate

discrimination

in

conjunction

with

the

experimenter's

specification

of

the

contingency.

A

series

of

transfer

experiments,

most

of

them

reviewed

by

Cumming

and

Ber-

ryman

(1965),

suggested

that

there

was

indeed

evidence

for

separate,

hierarchi-

cal

instructional

control

by

the

sample.

In

particular,

it

appeared

that

pigeons

generalized

broadly

to

novel

samples,

and

then

responded

to

the

comparison

stim-

uli

on

the

side

keys

on

the

basis

of

their

generalized

response

rather

than

the

sam-

ple

color

itself.

Inspired

by

Cumming

and

Berryman's

ideas,

Eckerman

(1970)

de-

vised

a

way

to

make

the

response

to

the

sample

explicit.

In

his

symbolic

match-

ing

procedure,

pigeons

were

required

to

peck

at

different

locations

along

a

strip

key,

depending

on

the

wavelength

of

the

sample,

in

order

to

produce

the

compar-

ison

stimuli

-vertical

and

horizontal

lines.

When

novel

wavelengths

were

pre-

sented

as

samples

in

a

generalization

test,

the

location

of

the

peck

on

the

strip

key

was

a

good

predictor

of

which

compari-

son

stimulus

would

be

pecked

on

that

trial.

In

effect,

the

birds

used

peck

loca-

tion

to

name

the

sample,

and

then

pecked

the

comparison

that

corresponded

to

that

name.

These

findings

and

their

interpretation

have

yet

to

be

integrated

with

the

work

of

Murray

Sidman

and

his

associates

(e.g.,

Sidman

&

Tailby,

1982)

on

the

use

of

symbolic

matching

to

establish

stimulus

equivalence

relations

in

which

the

sam-

ple

and

comparison

stimuli

are

inter-

changeable

members

of

an

emergent

stimulus

class;

but

speculation

along

these

lines

would

take

us

too

far

afield.

My

present

purpose

is

simply

to

suggest

that

Keller,

Schoenfeld,

Cumming,

and

Ber-

ryman

functioned

as

instructional

stim-

uli

for

their

students,

signaling

discrim-

inations

that

would

lead

to

reinforcing

consequences

when

we

emitted

the

ap-

propriate

behavior.

Keller

has

characterized

Columbia

in

his

and

Schoenfeld's

time

as

a

special

en-

vironment,

but

of

course

it

was

the

peo-

ple

and

the

ideas

that

made

it

special.

It

served

as

the

antecedent

for

behavior

ini-

tiated

there

and

for

consequences

that

have

ensued

over

the

years.

Using

these

terms

of

the

familiar

ABC

model,

to-

gether

with

the

notion

ofthe

instructional

stimulus,

I

will

try

to

bring

it

back

to

life.

Starting

with

the

behavior

term,

what

did

we

actually

do,

as

graduate

students,

day

after

day?

We

wired

relay

circuits;

deprived

and

trained

pigeons

and

rats;

set

timers

and

counters

to

arrange

rein-

forcement

contingencies;

varied

condi-

tions

systematically,

or

sometimes

on

a

hunch,

to

see

what

would

happen;

plot-

ted,

transformed,

and

replotted

our

data;

showed

them

to

one

another

and

debated

their

significance;

and

eventually

pre-

sented

them

in

evening

research

semi-

nars.

In

sum,

we

did

science.

The

consequences

were

tightly

tied

to

the

behavior:

the

fun

of

making

appa-

ratus

work,

the

thrill

of

controlling

be-

havior,

the

intellectual

and

aesthetic

sat-

isfaction

of

orderly

data,

the

reactions

of

fellow

graduate

students,

and

eventually

Keller's

comment,

"That's

very

interest-

ing,"

or

Schoenfeld's

challenge,

"Now

what

does

that

mean?"

The

presumably

ultimate

consequence-

successful

de-

fense

of

the

dissertation

-was

just

a

part

of

the

package.

It

was

the

ongoing

enter-

prise

that

mattered:

the

elaboration

of

what

Fred

called

"the

message,"

the

guiding

system

of

reinforcement

theory,

the

touchstone

from

which

one

could

try

to

grapple

with

a

wide

array

of

psycho-

logical

phenomena,

including

those

of

daily

life.

As

graduate

students,

our

behavior

had

to

be

shaped

and

then

brought

under

stimulus

control.

Relatively

little

of

this

took

place

in

formal

courses

and

semi-

nars.

For

example,

in

my

assistantship,

Berryman

introduced

me

to

the

delights

of

apparatus:

finding

some

potentially

useful

junk

in

the

surplus

stores

on

Canal

Street,

making

it

work,

then

(and

only

then)

figuring

out

something

interesting

to

do

with

it.

He

had

devised

a

clever

programmer

for

interlocking

schedules

of

reinforcement,

and

suggested

we

do

something

with

it.

I

had

yet

to

work

with

even

the

simplest

ratio

or

interval

sched-

ules,

but

his

enthusiasm

was

infectious,

so

we

started

an

experiment

even

though

INSTRUCTIONAL

STIMULI

223

I

had

no

real

notion

of

its

point.

Without

ever

telling

me

what

to

do,

Berryman

provided

all

the

cues

necessary

for

a

pro-

cess

in

which

each

step

was

controlled

by

the

data

as

they

accumulated

and

rein-

forced

by

the

orderliness

they

exhibited.

The

process

eventuated

in

an

article

de-

scribing

a

procedural

and

behavioral

continuum

relating

interval

and

ratio

schedules

(Berryman

&

Nevin,

1962).

A

different

piece

of

apparatus

might

well

have

led

elsewhere.

While

I

was

setting

up

the

interlocking

schedule

study

with

Berryman,

Keller

remedied

my

ignorance

of

operant

be-

havior

by

arranging

a

special

series

of

exercises

which

I

and

a

few

equally

ig-

norant

graduate

students

ran

through

af-

ter

hours

in

the

Psych

1

lab.

We

placed

our

rats

in

the

student

chambers,

and

there

it

was:

Raw

individual

behavior,

changing

in

an

orderly

way

within

the

space

of

an

hour

as

it

interacted

with

con-

ditions

that

we

could

arrange.

We

wrote

frequent

short

reports

for

Keller

to

com-

ment

on,

and

before

the

semester

ended,

he

had

taken

us

through

Skinner's

Be-

havior

of

Organisms

(1938),

with

Keller

and

Schoenfeld's

Principles

ofPsychology

(1950)

as

our

guide.

In

effect,

he

had

pro-

gressively

introduced

discriminative

stimuli

in

the

presence

of

which

appro-

priate

scientific

behavior-replication

of

the

work

of

Skinner

and

others-would

be

reinforced.

I

was

hooked

on

the

ex-

perimental

analysis

of

behavior.

For

the

next

few

years,

I

worked

con-

currently

on

reinforcement

schedules

and

complex

stimulus

control.

At

the

same

time,

seminars

and

projects

with

Clar-

ence

Graham

and

Bill

McGill

enhanced

my

initial

psychophysical

interests

and

provided

me

with

new

experimental

and

quantitative

skills.

Much

of

my

work

still

involves

a

psychophysical

approach

to

stimulus

and

schedule

control.

For

ex-

ample,

McGill's

instruction

in

the

theory

of

signal

detection

not

only

presented

a

new

approach

to

sensory

psychology,

but

also

offered

a

new

way

to

look

at

rein-

forcement

effects.

Graphs

of

response

probabilities

or

rates

maintained

by

one

schedule

in

the

presence

of

different

stim-

uli,

or

at

different

times,

against

the

cor-

responding

response

probabilities

or

rates

maintained

by

a

different

schedule

produced

the

equivalent

of

an

isosensi-

tivity

curve

(e.g.,

Nevin,

1965,

1974a),

which

I

later

called

an

isoreinforcement

curve

(Nevin,

1981).

This

led

me

to

view

both

antecedent

and

consequent

stimuli

as

functionally

equivalent

determiners

of

behavior

-a

view

that

I

am

still

working

to

elaborate

(Nevin,

1989).

The

behav-

ioral

basis

for

this

persistence

will

be-

come

clear

below.

Although

I

took

the

standard

round

of

courses

and

seminars,

most

of

my

edu-

cation

in

general

psychology

came

infor-

mally

from

Cumming.

He

taught

an

un-

dergraduate

experimental

psychology

course

in

the

mornings,

and

after

class

he

would

stand

in

the

lab,

Schaefer

beer

and

cigarette

in

hand,

discoursing

on

such

cognitive

topics

as

reaction

time

and

"mental

chronometry,"

concept

forma-

tion,

the

role

of

experience

in

scaling

ex-

periments,

or

the

significance

of

visual

"illusions"

as

opposed

to

"veridical"

perception.

Eventually,

I

recognized

that

all

these

topics

were

related

to

the

idea

of

the

discriminated

operant.

He

never

told

me

how

to

understand

behavior

in

these

situations,

but

he

gave

me

all

the

cues

that

permitted

me

to

identify

the

relevant

relations

for

myself.

The

effects

of

his

teaching

are

still

evident

in

a

recent

signal-detection

analysis

of

the

Muller-

Lyer

illusion

and

the

Kahnemann-Tver-

sky

"representativeness"

heuristic

(Nev-

in,

in

press).

The

effort

to

identify

continua

relating

phenomena

that

seemed

to

be

qualita-

tively

different

was

very

much

in

the

air,

and

remains

for

me

one

of

the

distinctive

features

of

the

Columbia

approach.

For

example,

in

a

graduate

seminar

on

aver-

sive

control,

Murray

Sidman

led

us

to

see

the

procedural

continuum

that

relat-

ed

escape

and

avoidance

behavior,

sug-

gesting

further

that

avoidance

behavior

might

be

reinforced

directly

by

shock

fre-

quency

reduction

(cf.

Herrnstein

&

Hine-

line,

1966).

He

assigned

me

to

report

on

his

own

doctoral

dissertation

(Sidman,

1

9

52),

and

I

was

entranced

by

the

striking

regularity

of

his

data

on

the

parametric

effects

of

response-shock

and

shock-shock

224

JOHN

A.

NEVIN

intervals

on

response

rate.

Inspired

by

McGill's

course

on

stochastic

timing

pro-

cesses,

I

struggled

for

weeks

to

elaborate

a

mathematical

model

based

on

shock

frequency

reduction

that

would

give

a

good

account

of

the

data,

and

came

tan-

talizingly

close;

John

Gibbon,

a

fellow

graduate

student,

eventually

succeeded

(Gibbon,

1972).

All

of

us

who

worked

with

schedules

of

reinforcement

spent

hours

studying

and

debating

the

work

of

Schoenfeld,

Cum-

ming,

and

their

former

and

current

stu-

dents

on

the

continuum

of

effects

that

could

be

produced

by

varying

only

tem-

poral

parameters

(Schoenfeld

&

Cum-

ming,

1960;

Schoenfeld,

Cumming,

&

Hearst,

1956).

One

of

their

goals

was

to

minimize

the

organism's

control

over

the

presentation

of

reinforcement,

so

that

the

schedule

would

be

a

true

independent

variable.

More

particularly,

they

showed

that

ratio-like,

interval-like,

and

various

intermediate

schedule

performances

could

be

generated

by

appropriate

selec-

tion

of

temporal

contingencies

alone.

In

retrospect,

their

approach

seems

quite

dif-

ferent

from

more

recent

analyses

that

stress

the

feedback

relations

between

re-

sponding

and

reinforcement

(e.g.,

Baum,

1973).

The

feedback

approach

takes

ob-

tained

reinforcer

rate

as

a

dependent

variable

determined

jointly

by

response

rate

and

the

scheduled

contingencies,

with

the

result

that

the

schedule

ceases

to

be

an

independent

variable.

However,

the

notion

that

feedback

relations

them-

selves

lie

along

a

continuum

(e.g.,

Rach-

lin,

1978)

is

consistent

with

the

spirit

of

Schoenfeld

and

Cumming's

approach-

as

was

Berryman's

and

my

1962

paper

on

interlocking

schedules.

The

importance

of

identifying

contin-

uous

relations

may

even

have

overshad-

owed

the

phenomena

under

consider-

ation.

At

one

point

in

my

graduate

career,

I

recognized

that

a

procedural

variable

could

bridge

the

gap

between

two

seem-

ingly

disparate

topics

in

stimulus

control

and

mentioned

this

to

Schoenfeld.

I

no

longer

recall

the

topics,

but

I

will

always

remember

Nat's

response:

"Once

you

have

learned

to

see

continuous

relations

rather

than

discrete

phenomena,

your

life

is

changed."

There

could

be

no

better

example

of

his

role

as

a

selector

of

dis-

criminations.

If

it

is

accepted

that

Keller,

Schoenfeld,

Cumming,

and

Berryman

functioned

as

instructional

stimuli,

we

are

left

with

the

problem

of

characterizing

the

special

en-

vironment

within

which

these

instruc-

tions

were

arranged.

I

would

like

to

sug-

gest

that

the

Columbia

environment

served

as

a

superordinate

instructional

stimulus,

setting

the

occasion

for

rein-

forcement

of

discriminative

operants

un-

der

subordinate

instructional

control.

To

illustrate

this

notion,

I

would

like

to

de-

scribe

part

of

a

study

conducted

in

my

lab

with

Jim

Grosch

(Nevin

&

Grosch,

in

press).

We

trained

pigeons

on

a

conventional

delayed

matching-to-sample

task

with

red

and

green

key

lights.

Delays

interposed

between

offset

of

the

sample

and

onset

of

the

comparison

key

lights

varied

un-

predictably

between

0

and

21

seconds

from

trial

to

trial.

Some

trials

were

ac-

companied

by

a

tone,

and

others

by

white

noise.

If

the

tone

was

present,

a

response

to

the

correct

comparison

color

produced

4.5

s

of

access

to

grain,

whereas

if

the

noise

was

present,

a

correct

response

pro-

duced

only

1.5

s

of

access

to

grain.

(Tone

and

noise

were

counterbalanced

across

birds.)

Thus,

the

auditory

cue

was

a

sort

of

higher-order

instructor,

signaling

not

which

response

was

correct,

but

whether

the

reinforcer

would

be

large

or

small.

Its

effects

were

clear:

All

our

pigeons

re-

sponded

more

accurately

at

all

delays

on

trials

when

the

auditory

cue

signaled

the

larger

reinforcer.

I

would

like

to

suggest

that

the

Colum-

bia

Psychology

Department

was

like

that

auditory

cue:

an

environment

in

which

instructional

control

was

especially

effec-

tive

because

the

reinforcers

were

so

am-

ple.

Much

of

my

present

research

deals

with

"behavioral

momentum"

-the

tendency

for

a

class

of

learned

behavior

to

persist

under

altered

conditions.

The

data

on

free-operant

performances

of

pigeons,

rats,

monkeys,

and

humans

in

a

variety

of

experimental

settings

are

quite

consis-

INSTRUCTIONAL

STIMULI

225

tent:

The

persistence

of

operant

behavior

increases

with

the

rate

or

amount

of

re-

inforcement

occurring

in

the

stimulus

sit-

uation

(e.g.,

Nevin,

1974b,

1979;

Mace

et

al.,

in

press).

Many

of

those

who

studied

at

Columbia

during

the

Keller

and

Schoenfeld

years

exemplify

this

prin-

ciple,

in

that

ample

reinforcement

for

the

scientific

behavior

acquired

there

has

en-

gendered

great

persistence.

My

own

con-

tinuing

preoccupation

with

stimulus

and

schedule

control

is

directly

traceable

to

that

environment

as

it

was

mediated

by

the

instructional

stimuli

provided

by

Keller,

Schoenfeld,

Cumming,

and

Ber-

ryman.

My

history

of

reinforcement

at

Colum-

bia

also

determined

the

title

ofthis

paper,

which

in

turn

determined

its

content.

The

effort

to

describe

that

history

in

the

lan-

guage

of

complex

stimulus

control

has

given

me

a

fresh

realization

that

what

we

study

and

what

we

do

are

of

the

same

stuff.

Our

analyses

and

our

actions-

professional

science

and

personal

expe-

rience-all

are

aspects

of

behavior,

sub-

ject

to

a

common

set

of

principles

that

we

grasp

more

and

more

surely

as

the

science

of

behavior

grows.

REFERENCES

Baum,

W.

M.

(1973).

The

correlation-based

law

of

effect.

Journal

of

the

Experimental

Analysis

of

Behavior,

20,

137-153.

Berryman,

R.,

&

Nevin,

J.

A.

(1962).

Interlocking

schedules

of

reinforcement.

Journal

of

the

Ex-

perimental

Analysis

of

Behavior,

5,

213-223.

Cumming,

W.

W.,

&

Berryman,

R.

(1965).

The

complex

discriminated

operant.

In

D.

Mostovsky

(Ed.),

Stimulus

generalization

(pp.

284-330).

Stanford:

Stanford

University

Press.

Eckerman,

D.

A.

(1970).

Generalization

and

re-

sponse

mediation

of

a

conditional

discrimina-

tion.

Journal

of

the

ExperimentalAnalysis

of

Be-

havior,

13,

301-316.

Gibbon,

J.

(1972).

Timing

and

discrimination

of

shock

density

in

avoidance.

Psychological

Re-

view,

79,

68-92.

Herrnstein,

R.

J.,

&

Hineline,

P.

N.

(1966).

Neg-

ative

reinforcement

as

shock

frequency

reduc-

tion.

Journal

of

the

Experimental

Analysis

of

Be-

havior,

9,

421-430.

Keller,

F.

S.,

&

Schoenfeld,

W.

N.

(1950).

Prin-

ciples

of

psychology.

New

York:

Appleton-Cen-

tury-Crofts.

Mace,

F.

C.,

Lalli,

J.

S.,

Shea,

M.

C.,

Lalli,

E.

P.,

West,

B.

J.,

&

Nevin,

J.

A.

(in

press).

The

mo-

mentum

of

everyday

human

behavior

in

a

group

home.

Journal

of

the

Experimental

Analysis

of

Behavior.

Nevin,

J.

A.

(1965).

Decision

theory

in

studies

of

discrimination

in

animals.

Science,

150,

1057.

Nevin,

J.

A.

(1

974a).

On

the

form

of

the

relation

between

response

rates

in

a

multiple

schedule.

Journal

of

the

ExperimentalAnalysis

ofBehavior,

21,

237-248.

Nevin,

J.

A.

(1974b).

Response

strength

in

mul-

tiple

schedules.

Journal

of

the

Experimental

Analysis

of

Behavior,

21,

389-408.

Nevin,

J.

A.

(1979).

Reinforcement

schedules

and

response

strength.

In

M.

D.

Zeiler

&

P.

Harzem

(Eds.),

Reinforcement

and

the

organization

of

be-

haviour

(pp.

117-157).

Chichester,

England:

Wi-

ley.

Nevin,

J.

A.

(1981).

Psychophysics

and

reinforce-

ment

schedules:

An

integration.

In

M.

L.

Com-

mons

&

J.

A.

Nevin

(Eds.),

Quantitative

analyses

of

behavior.

Vol.

1.

Discriminative

effects

of

re-

inforcement

schedules

(pp.

3-27).

Cambridge:

Ballinger.

Nevin,

J.

A.

(1989

May).

Toward

an

integrative

account

of

discriminated

operant

behavior.

In-

vited

address

at

the

meeting

of

the

Association

for

Behavior

Analysis,

Milwaukee,

WI.

Nevin,

J.

A.

(in

press).

Signal

detection

analysis

of

illusions

and

heuristics.

In

M.

L.

Commons,

M.

Davison,

&

J.

A.

Nevin

(Eds.),

Quantitative

analyses

of

behavior.

Vol.

11.

Signal

detection:

Mechanisms,

models,

and

applications.

Hillsdale,

NJ:

Erlbaum.

Nevin,

J.

A.,

&

Grosch,

J.

(in

press).

Effects

of

reinforcer

magnitude

on

delayed

matching

per-

formance.

Journal

of

Experimental

Psychology:

Animal

Behavior

Processes.

Rachlin,

H.

(1978).

A

molar

theory

of

reinforce-

ment

schedules.

Journal

of

the

Experimental

Analysis

of

Behavior,

30,

345-360.

Schoenfeld,

W.

N.,

and

Cumming,

W.

W.

(1960).

Studies

in

a

temporal

classification

of

reinforce-

ment

schedules:

Summary

and

projection.

Pro-

ceedings

of

the

National

Academy,

46,

753-758.

Schoenfeld,

W.

N.,

Cumming,

W.

W.,

&

Hearst,

E.

(1956).

On

the

classification

of

reinforcement

schedules.

Proceedings

of

the

National

Academy,

42,

563-570.

Sidman,M.

(1952).

Avoidanceconditioningwith

brief

shock

and

no

exteroceptive

"warning

sig-

nal."

Doctoral

dissertation,

Columbia

Univer-

sity,

New

York.

Sidman,

M.,

&

Tailby,

W.

(1982).

Conditional

discrimination

vs.

matching

to

sample:

An

ex-

pansion

of

the

testing

paradigm.

Journal

of

the

Experimental

Analysis

of

Behavior,

37,

5-22.

Skinner,

B.

F.

(1938).

Thebehavioroforganisms.

New

York:

Appleton-Century-Crofts.