Self-Control

Author(s): Howard Rachlin

Source: Behaviorism, Vol. 2, No. 1 (Spring, 1974), pp. 94-107
Published by: Cambridge Center for Behavioral Studies (CCBS)
Stable URL: http://www.jstor.org/stable/27758811 .
Accessed: 21/06/2014 02:41

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].

Cambridge Center for Behavioral Studies (CCBS) is collaborating with JSTOR to digitize, preserve and extend
access to Behaviorism.

http://www.jstor.org

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control1

Howard Rachlin
State University of New York at Stony Brook

We use the term "self-control" to describe decisions between

ordinarily
alternatives arriving at different times. For instance, having a roast-beef sandwich
for lunch provides a reward now; having cottage cheese instead
can
provide
a
reward tomorrow when I step on the scale. Take the temporal issue away and the
issue of self-control goes away as well. If itwere that cottage
suddenly discovered
cheese was just as fattening (and therefore had the same ultimate consequences
for me) as a roast beef sandwich and I still ate the cottage cheese, Iwould have to
admit that I simply liked the cottage cheese better. The decision would become,
like one between blue and brown suits, one of taste.
simply
When the events chosen between are continuous in the sense that they cannot
be located in a brief interval (e.g., working at
a
job and being paid
at a
temporal
certain rate) then self-control when the events differ in extent.
applies temporal
at an
The primary rewards obtained by working unpleasant job must extend
further in time than the job itself before we ascribe self-control to the worker
when he works. If the and pleasantness are co
unpleasantness completely
temporaneous, self-control is not involved. We would see' the worker as
weighing
the two factors and to work was
choosing only if the resultant pleasant. Other
? to
wise, why work? But let the worker's reward extend further in time providing
for his family, enjoying luxuries, etc. and working versus not working becomes
a matter of self-control.

Psychologists studying self-control (JVlischel, 1966; Rotter, 1954) have long

noticed that self-control is a "now" versus "later" issue. Their subjects show self
control when they prefer larger rewards in the future to smaller rewards in the
present or, symmetrically, avoid greater pain in the future in return for lesser pain

1. Preparation of this paper was supported by a grant from the National Science Foundation.
For helpful comments on a previous draft of this paper I thank G. Ainslie, W. Baum,
E. Erwin, E. Fantino, R. J. Herrnstein, F. Levine, G.H. Whitehurst, and G.T. Wilson.

94

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Kachlin

in the present. In this context visiting the dentist is showing self-control and not
visiting him shows lack of self-control. The hedonic picture looks something like
that of Figure 1. The reason for indecision between two alternatives, one
already
assumed to be better than the other, is that the better alternative is only better in

don't visit
dentist

MORE
PAIN visit dentist

LESS

PAIN

Figure 1: Hypothetical diagram of the way pain would vary on visiting the dentist and
without visiting the dentist.

the long run. The worse alternative offers immediate benefit. The difference
between someone who is controlling himself and someone who is not controlling
himself is thus not in the spatial locus of control (from inside versus from out
side his skin) as the term "self-control" seems to in the
imply but temporal locus
?
how far away from the present must we look to find the source of control.
With regard to the establishment of self-control the question to ask is how con
trol is shifted from immediate to distant consequences. Again, this is a temporal
not a one. as Skinner (1955) has
question, spatial However, pointed out, psy
have been hesitant to ascribe to events far apart,
chologists causality temporally,
from each other. They have translated the action of distant events into present
events and events inside the organism. Thus "ego strength," "in
placed the present
ternalization," "resistance to frustration," and other cog
"subjective probability,"
nitive or motivational terms have made their way into discussions of self-control.
These terms refer to mediating mechanisms which represent a past or a future event
in the present. It is my contention that these mediating mechanisms are not
necessary to understand or self-control. have served for the
empirically study They

95

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control

as the ether as a way of

psychologist, formerly served the physicist, bridging
between causes and their consequences when those causes and consequences were
an which itwas not believed that could act.
separated by entity through causality
In the case of the this entity
was unfilled space. In the case of the psy
physicists
chologists this entity is an unfilled temporal interval. As Staddon (1973) has
concluded in a recent of in ". . . limitation of the
analysis causality psychology,
causes of behavior to stimuli is without
temporally contiguous justification.,,

WHATCAN CAUSE BEHAVIOR?

A research has recently shown that the cause of
good deal of psychological
behavior can be the between that behavior and events in the environ
relationship
ment (Baum, 1973; Bloomfield, 1972; Herrnstein and Hineline, 1966; Majer,
and Solomon, 1969). To take one Herrnstein and Hineline
Seligman example,
(1966) a set of for a rat so that the rate of
arranged contingencies irregularly
delivered brief electric shocks varied inversely with rate of bar pressing. If the
rats faster the shocks came slower; if slower the shocks
pressed they pressed
came faster. The rats learned to press the bar no
although single bar press avoided
any single shock. The rate of bar presses and the rate of shocks were the critical
variables. But the rate of a discrete event has no at an instant of time.
meaning
At any instant the event is either occurring (in which case its rate is or is
infinite)
not (in which case its rate is Consider 2. The
occurring zero). Figure pips
represent shocks or bar presses in a hypothetical replication of the Herrnstein
Hineline experiment. It is generally true about the bar presses and shocks that
the more bar presses, the fewer shocks. But if all the data were not available, if
an observer saw what was between M and N, between O and or between
just

-
shocks

! il

bar presses

I II Mil II I

0
TIME

Figure 2: Hypothetical pattern of shocks and responses in the Herrnstein-Hineline experiment.

Tne pips on the top line stand for shocks and the pips on the bottom line stand for responses.
The vertical dotted lines represent restricted of 0-P and
periods observation, M-N, Q-R.

96

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Kachlin

Q and R he might conclude that shocks were causing bar presses, that bar presses
were or that the two were unrelated. an extended
causing shocks Only view of
the temporal properties of bar presses and shocks allows us to see the true relation
a rat in the Herrnstein-Hineline that presses the bar as
ship. Imagine experiment
in Figure 2. We can now ask, "What causes the bar In the
presses?" light of the
above discussion the cause of the bar presses is the relationship between bar press
as it is
ing and shocks experienced by the rat.
Suppose in the Herrnstein-Hineline experiment that each bar press costs the
rat something. We could imagine that the effort of pressing was increased or that
a low In that case the immediate con
intensity shock followed upon each press.
sequences of pressing the bar would be painful (or effortful) but the long-term
consequences of not pressing the bar would be still more painful. The picture
would be something like that shown in Figure 1. "visit dentist"
Replace by "press
bar" and "don't visit dentist" by "don't press bar" and you have a fair picture of
the rat's situation. Will rats press bars under these conditions? In a more recent
experiment, Lambert, Bersh, Hinelineand Smith (1973) arranged contingencies for
a rat so that a press a shock but avoided several
produced single immediately
shocks of equal intensity later on. The rats in the Lambert et al experiment con
the bar. Can we say that the rats were self-control?
sistently pressed exhibiting
If the criterion for an alternative which
exhibiting self-control is choosing involves
a future over a smaller present then these rats were
larger good good exhibiting
a more in the distant future in favor
self-control; they avoided painful experience
of a less painful experience in the immediate future.2 if, on the other hand, we
insist that exertions of "ego-strength," "internalization," or other
"expectancy,"
or motivational events must also go on somewhere within the rat we shall
cognitive
have the difficult task of trying to verify their occurrence. But such
explanatory
efforts are unnecessary. The behavior itself is all the evidence we need that self
control is going on. What would we say ifwe found somehow that the rat had
the appropriate motivational or but did not press the bar?
cognitive apparatus
It would be pointless, then, to claim that the rat was controlling itself. It is the
rat's behavior in relation to the contingencies
imposed that comprises self-control.
Similar arguments apply as well to all human instances of self-control. The way
in which human behavior is more complicated than a rat's behavior is not that
human behavior is controlled from inside while the rat's behavior is controlled
from outside but that the environmental events
controlling human behavior prob
occur over a wider interval than those which control the behavior
ably temporal
of the rat. When we refuse the third martini at a party (ifwe do refuse it) it is not
because of an exercise of some force within us but part of a response to con
out in time before and after we are offered the drink. The
tingencies spread widely
wider contingencies involve events on the way home and the next morning while
the narrow contingencies involve only events at the party itself. Why we should
act in accordance with wide contingencies rather than with the narrow ones which

2. Fantino (1966) showed that pigeons could learn to show self-control in a symmetrical
situation with positive reinforcement. The pigeons in Fantino's experiment could obtain an
immediate reward (followed by a penalty) by pecking a red key or they could wait a few
seconds until the key turned green and obtain a reward with no penalty. The pigeons
initially pecked the red key but with six months of training came eventually to wait for
the green key.

97

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control

dictate acceptance of the drink is another question, which we shall try to deal
with later. It suffices to say now that the question is answered no better by
to internal events or states than it iswithout them.
referring
Let us return to the rats in the Herrnstein-Hineline experiment. These rats
were the bar and few shocks. I
pressing relatively rapidly receiving relatively
that the between the bar presses and the shocks as
argued contingency experienced
rat is the cause of the bar
by the pressing. Traditional psychology would invoke
the cognitive and motivational mediating mechanisms previously discussed. What
is the purpose of those mechanisms? In order to avoid having an event at one
time caused an event at another time the concept of a state of the
by organism
is introduced. Events at one time affect the state, and the state affects the behavior.
The state in question may be motivational or are
cognitive. Past events supposed
to govern present events via the motivational state while future events are
supposed
to govern present events via the cognitive state. Herrnstein (1969), Bolles (1972),
and Seligman and Johnston (1973) have argued against motivational and for
cognitive explanations of the bar pressing of rats, but ifwe grant that events
extended in both are events
temporal directions directly caused by other similarly
extended there is no need to refer to either kind of state. The notion that cog
nitive and motivational states mediate between past and future events and present
behavior is not necessary. It obscures the search for the most direct causes of
behavior because it tends to direct that search into the organism instead of into
the past and future.
This does not imply that a given act can be manipulated by
events which occur
after the act is over. When the environmental event and the behavior are extended
in time it makes no sense to talk of an environmental sub-event at one point in
time causing a sub-act at another in time. Each sub-act is to be seen as
point only
part of the complete act; and it is the causal relation between the complete act
and the complete environmental event which concerns us. In Figure 2 a certain
rate of bar
pressing is caused by the relationship between shocks and bar presses.
At our level of inquiry no individual bar press is "caused" at all. It makes as little
sense to cause of an individual bar press in the
place the past as it does to place
the cause in the future. if I like a
painting, it is
me who likes the
Analogously,
as a whole. One could not
painting say whether my left eye liked the lower
corner of the
right painting.3
With respect to the traditional between classical and instrumental
dichotomy
conditioning, the notion that the cause of behavior extends into both the past
and future removes the usual
? temporal distinction between classical conditioning
the reinforcer the unconditioned ?
(where stimulus precedes the act) and
instrumental conditioning (where the reinforcer follows the act) and concentrates,
as has
recently been suggested (Bloomfield, 1972; Catania, 1971; Gamzu and
Schwartz, 1973; Rachlin, 1970; Staddon, 1973), on correlations between one
environmental event and another and between an environ
(classical conditioning)
mental and a behavior event (instrumental
conditioning).

3. If we were forced to consider individual sub-events as being caused we would have to admit
that an event could be changed by a subsequent event. Modern historians recognize that the
facts of the past are not easily separable from their interpretation (their context). If individual
past events were considered in isolation of their context, a modern theory of history which
revises our interpretation of those events in effect, be them.
would, changing

98

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Kachlin

To say that the origin of self-control is not in the self is not to say that the
has no or has been subtracted out of consideration. The
organism properties
biological properties of the organism determine which environmental events
control which behavioral events. What has been subtracted out are those psy
to as memory,
chological (as opposed biological) properties of the organism such
expectancy, response strength, etc. which serve to
only bridge temporal gaps.
The remaining biological are those that serve for reaction
properties of the organism
to immediate as well as If a man is stabbed and he
long-term contingencies.
bleeds, no one will be to talk about memory of the and
tempted stabbing
response strength of the bleeding intervening between stabbing and bleeding,
although the properties of his body determined that he would bleed when
stabbed. But for temporally extended events such as his dissatisfaction with a
bad job, traditional psychological analysis will invoke memory, expectancy
to cause and effect into immediate
and response strength in order bring the
temporal proximity. It is certainly something about the man that reacts to the
long hours, the low pay, the hostile boss, etc., by complaining, on strike
going
or But it is nevertheless these external temporally extended events that
quitting.
cause his behavior and not his immediate memories, and response
expectations,
strengths.
Very often widespread and narrow contingencies
cause the same behavior.
at a rewards now, in the past and in the future.
Working pleasant job provides
The relationship between these rewards and work, more
directly than anything
else, causes the work to be done. But often temporally extended events cause
behavior in conflict with that caused by temporally constricted events. When such
conflict arises a choice has to be made between the constricted and extended
consequences, the choice of extended consequences being self-control. The sub
state to "exertion of self-control" may be no different
jective corresponding
than that corresponding to any difficult choice. whether
qualitatively Deciding
to accept that third cocktail may be more difficult than deciding whether it is to
be a martini or a manhattan but the difference between the two decisions would
be simply that one ismore difficult, not that one is different in kind from the other.
To say that the cause of action can be narrow or widespread in time is not to
that events have effects whenever occur. There is often a
say equal they may
greater weight attached to constricted than to contingencies. The
widespread
and motivational theorists invoke of memory and to
cognitive gradients certainty
the reduction in control by events far in the past or future. But, given
explain
these gradients, not be first to the state of the
they need applied organism and
then to its behavior. can to behavior. Furthermore,
only They apply directly
many actions, and in the case of humans most of our significant actions, are un
related to present causes. We move from one city to another, get married or
or not because of anything that is at the
divorced, get jobs quit them happening
very moment we perform these actions (even when the actions themselves are
It has proved difficult and fruitless, moreover, to trace chains of
brief). secondary
reinforcers back from some
presumably primary reinforcer just in order to bring
the reinforcer in temporal proximity with the acts. If a man moves from
Maine to Florida he does not move because of the weather on the day he moves al
of the weather. The cognitive or motivational
though he may be moving because
theorists will say that the weather in Maine causes a certain state in the man and
the weather in Florida causes another state in the man and that moving is rein

99

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control

forced by a transition from one state to the other. But such states have been
difficult to pin down. They are awkward, unparsimonious and invite freewheeling
which can in their terms. Their one convenience, of
theorizing explain anything
causes and effects into is not worth its price.
bringing temporal proximity, simply
The cause of a man's moving from Maine to Florida is most parsimoniously
described in terms of the weather inMaine and the weather in Florida ? in terms
? occur
of mean temperatures and average snowfall, etc. nothing that could
a brief
within temporal interval.
the on the environment I am not arguing that there are
Despite emphasis above
no individual differences in ability to exercise self-control. Like other abilities,
self-control undoubtedly through
some combination of genetic and en
develops
vironmental conditions. vs. environment is not the main issue here.
Heredity
What is at issue is whether the causes of behavior we label self-control are different
in kind from the causes of behavior we label lack of self-control. Analysis reveals
that the two causes differ in degree of temporal extent, not in their place of origin.

TECHNIQUES OF SELF-CONTROL

The kind of self-control to which we have been referring might be called "brute
force" self-control. When the temptation is offered it is simply refused. The
martini is turned down at the party, the bakery is passed without a purchase, the
dessert is etc. The direct cause of such behavior is the long-term
pushed away,
correlation between the behavior and its consequences.
An objection might be raised that the view espoused here applies well enough to
brute-force self-control but not to more sophisticated techniques of self-control
such as those developed or Alcoholics or the
by Weight Watchers, Anonymous
we are in life to our own
strategies constantly inventing everyday manipulate
behavior. Consider the following ways in which a student might get himself
to
study:
1. He studies despite the temptation to go to the
simply
movies instead.
2. He rewards himself for studying by going to the movies
afterward.

3. He has previously deposited a sum of money with

fairly large
a friend. He has instructed the friend to check every half-hour
to see that he is If the friend
during the evening studying.
does not find him studying, the friend is further instructed to
to a are
send the money political party whose views exactly
contrary to those of the student.
Sofar we have discussed only alternative Number
1 and that only in theory.
Let us
reserve for the next section on how self-control
speculations might be
and turn now to alternatives 2 and 3. are both forms of self
brought about, They
control because their object is to increase the likelihood that behavior will be in
accordance with its long-term consequences. are of
They fairly representative
types of self-control often recommended behavior and might be
by therapists
called, respectively, self-reinforcement and commitment.
Let us consider self-reinforcement first. A little
analysis reveals that the reinforce
ments given to oneself do not support the behavior upon which are
they contingent.
for a moment, that the student in our increased his
Suppose, example studying by
100

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Kachlin

to the movies afterward. Now, suppose that this self-reinforcement cont

going
inued but external reinforcers, the good grades, the knowledge, the social approval
to be gained from were withdrawn. How
studying long would studying continue
without them? What would the point of studying be? Would going to the movies
continue to "work" as a One does not have
self-imposed reinforcer for studying?
to do an experiment to answer these questions In what sense, then,
negatively.4
was the movie a reinforcer? This is a critical question because the test for a re
inforcer must be whether it can support behavior. If we take away external re
inforcers leaving only self-reinforcement which supports no behavior other than
that involved in its consumption5 then self-reinforcement loses its effectiveness.
It seems likely that self-reinforcement is a form of secondary reinforcement.
to the movies have increased not because of its
Going might studying reinforcing
properties but because of its stimulus properties. This hypothesis could be tested
For instance, a
by substituting neutral, but strong, stimuli for self-reinforcers.
student who rewards himself by eating a peanut after each 10 minutes of
studying
to as well if, instead of eating it, he transfers the peanut from
ought study simply
one dish to another as a way of
counting 10-minute periods of studying. The ques
tion is empirical, but in the research on self-reinforcement, whether with humans
or animals, it has
rarely been addressed directly.6
A behavior to institute a program
therapist might ask his patient of self
reinforcement for studying as sort of a cast or mold but the therapist might not
be satisfied until the cast was removed. The program might start with going to
the movies as a self-reinforcer, switch to a more convenient and less time consum
ing activity like eating peanuts after each 10 minutes of studying, then switch to a
program of record-keeping, then nothing. con
Presumably, by this time, the
tingency between studying and grades would be controlling studying directly.
to prior supports. The point is that the
Relapses might be treated by returning
sequence from going to the movies to peanuts to
record-keeping
to
nothing may
not be a withdrawal of reinforcement but a withdrawal of stimuli.
Self-reinforcement may be like saying to oneself "Yes. I did just do that," per
the same function for the as the feedback click which we arrange
forming subject
to tell the a events affects
pigeon it has just pecked key. if the correlation between
behavior, other things being equal, correlations between intense stimuli will affect

4. Yet, recently self-reinforce ment has emerged as an area of study with humans (Bandura
1971) and even animals (Mahoney and Bandura, 1972).

5. The behavior involved in eating, chewing and swallowing, for instance, can be thought of as
reinforced by the digestion of food. Restricted to such events, self-reinforcement is a valid and
interesting concept. But the more common use of the term is in the sense of example number 2
where the behavior is not consummatory.

6. Bandura and Perloff (1967) had children set their own criterion for a task (turning a crank)
and then reward themselves with tokens for reaching their own criterion. The children who set
their own criterion turned as fast as those childrenwhose criterionwas set for them (the latter
group was also given tokens instead of rewarding themselves). The interesting part of this
experiment is the setting of the criterion, which is a question of commitment (Why didn't the
children set the criterion as low as possible?), not the self-reward. Once the criterion was set
the children would have been disobeying the rules of the "game" (they were told that they
were evaluating a game) had they rewarded themselves without reaching criterion.

101

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control

behavior more than will correlations between weak stimuli. The feedback involved
in self-reinforcement may well be more intense than the normal proprioceptive
and kinesthetic feedback of most behavior. Where this is not the case, self
reinforcement should not work.

Occasionally the term "self-reinforcement" is used the way cognitive and

motivational internal mechanisms of self-control have been used, to bring cause
and effect into temporal contiguity. If, as the cognitive and motivational theorists
assume, reinforcement must come immediately after the act being reinforced but
no immediate external reinforcer is observed then one has to be invented. Covert
is a It of un
self-reinforcement likely candidate. provides all the advantages
observable (hence, untestable) concepts while preserving the flavor of behavioral
terminology. But covert self-reinforcement, rather than lending rigor to a basically
vague underlying concept, is in danger of lending vagueness to the
basically
rigorous behavioral terminology it borrows.
Now let us turn to the sort of involved strategies exemplified by the third
alternative, where the student has made an agreement with his friend to send his
money to an if he does not This self-control
opposing political party study.
involves a commitment. It is rather like the signing of a contract
specifying various
kinds of performance in the future and setting forth penalties for failure to
comply.
George Ainslie (1970) and Leonard Green and I (1972) have argued that commit
ment of this kind follows from the of choice advanced by
simple descriptions
Logan (1965) and Herrnstein (1970). We showed that internal
by complicated
mechanisms were not necessary to commitment; that rats and
explain pigeons
were con
capable of employing commitment strategies in situations where the
involved were Ainslie's argument and his
tingencies straightforward. experiment
differ in their detail from Green's and mine, but the main arguments are the same
and will be summarized here.
Let us suppose that the long-term consequences of are valuable, but
studying
the value of not rises above The dilemma is that,
occasionally studying studying.
because of the short-term aversive consequences of studying, the time in the
presence of the stimuli which make it possible to study (a quiet room with a desk,
a book, a not
pad and sharpened pencils) is the very time that the value of study
ing rises above As soon as those stimuli are removed, say, while in
studying. lying
bed in the morning, the value of assumes its usual This is
studying high place.
where a commitment strategy is useful. The student is offered the contract (the
commitment outlined in alternative Number 3 at the beginning of this section).
The commitment is offered and at a time when the value of
accepted studying is
The of the commitment is to ? to
high. effect reduce the student's choice compel
him to
study.
a decision process in the case of alter
Figure 3 is diagram of the commitment
native 3. There are two decisions in question. Decision X takes place in the even
when the student is to Decision Y takes
ing supposed study. place the morning
before. The decision at time X is between not at
studying and studying. But
time X the short term consequences of not studying will determine the student's
choice. Thus, not studying will be the invariable result. The other decision at
time Y is whether to agree to the commitment. the commitment is ef
Assuming
fective (that the student will not tolerate the of his money to
absolutely sending
an to
opposing political party) instituting the commitment is equivalent choosing

102

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Kachlin

I TIME I
Figure 3: Flow diagram of commitment to study. Choice at X isbetween studying immediately
and not studying. It is assumed that a student would not study at X. Choice at Y is between
having a choice later (top arm) and being forced to study later (bottom arm). A student who
would not study at X might nevertheless commit himself to study by choosing the lower arm at Y.

the lower branch of Figure 3, to to

studying, and experiencing the consequences
thereof. But, given the value structure as we have outlined it, at time Y the value
of studying is higher than that of not
studying and the student must agree
to the commitment.
Once the commitment is available and once the contingencies are made effec
tive, commitment behavior follows To that exercise of
automatically. emphasize
commitment was not on ego resistance to
dependent strength, internalization,
frustration, or other or motivational Ainslie
sophisticated cognitive apparatus,
showed with rats, and Green and I, with pigeons, that
relatively naive animals
would exhibit commitment. The experiment Green and I did
closely follows the
schema of Figure 3. The choice at Xfor our pigeons was between a small immediate

103

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control

to not a
food reward (analogous studying in Figure 3) and larger food reward
to in
delayed by several seconds (analogous studying Figure 3). The pigeons in
at X. at
variably chose the small immediate reward But, Y, several seconds before
X, when they could choose to restrict subsequent alternatives, they chose
not to
a ?
have choice they chose the bottom branch of Figure 3 and obtained the larger
(but delayed) reward. The values of the delays and amounts were determined by
Herrnstein's (1970) model for choice which that preference varies
predicts
with amount and with The model further
directly inversely delay.7 predicts that
commitment will be more the the
likely greater temporal separation between the
choices at Y and X. This was also confirmed.
It is necessary to two ?
distinguish between operations the exercise of the com
mitment strategy and the invention of the strategy. While the exercise of commit
ment has to do with
nothing higher mental processes, the invention of commitment
strategies is another matter. The invention may be a higher mental process (how
ever that is
defined) and itmay be performed by the user (the student) or another
or a commitment
person (his friend his therapist) but inventing strategy is not
exercising self-control. Neither does it take ego strength, internalization, etc.
to invent a self-control strategy any more than these were necessary to
qualities
invent the cotton gin. Self-control is done by using the strategy so that one's be
havior will be in accordance with its long-term consequences. And use of the
strategy occurs once an effective is invented.
automatically strategy
To summarize, both self-reinforcement and commitment are related to self
control because they both increase the likelihood that behavior will be controlled
by its long-term consequences. Self-reinforcement makes the relation between
behavior and its consequences more vivid stimuli correlated with
by providing
those consequences. Commitment restricts choice so that behavior will automatic
ally conform to long-term consequences.

TEACHING SELF-CONTROL
we shall have little to
Unfortunately, say in this section. Most of the research
on self-control has been on the personality correlates of people who are good at
controlling themselves. The old, it seems, do it better than the young (Mischel,
the sane better than the
1958), schizophrenic (Klein, 1967), the intelligent better
than the unintelligent (Mischel and Metzner, 1962), the rich better than the poor
(Maitland, 1967), etc.
Obviously, self-control is a good to have. But
thing
environmental events which can generate self-control (i.e., shift the cause of be
havior from short-term to
long-term events) have not been systematically examined.
The to the areas:
preceding analysis would direct investigation following

7. According to Herrnstein's model, the relative value of two activities is: =

2 wnere
A2 Dl
= =
A amount and D delay of reinforcement. In our alternative 1 was a small im
experiment,
mediate reward and alternative 2 a larger delayed reward. A1/A2 always equalled 0.5, and
D2 was always 4 seconds greater than Dl. VI =
At choice X, referring to Figure 3, 0.5
that the small immediate reward would be chosen. At choice Y, 10
(^ )= infinity,predicting
VI = =
seconds prior, 0.5 (4+10/0+10) 0.7 predicting that the large delayed reward would be
chosen.

104

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Kachlin

1. Does one
practice with long-term contingencies controlling
increase the likelihood that they will control other activi
activity
ties? The notion that there are "addictive personalities" implies
that some people cannot or do not to
long-term
con
respond
in several behavioral areas. Can be to behave
tingencies they taught
in accordance with these contingencies? a per
generally Perhaps
son who has learned not to overeat will have an easier time learning
not to smoke. This implies that the way to begin in the cure of
harmful habits may be through control of other habits. For instance,
alcoholics might first be trained to keep their weight under control
or to stop
smoking. Control of eating an ice cream sundae by the
next to control of getting drunk
day's consequences might transfer
the next
by day's consequences.
2. The long-term antecedents and consequences of certain
events be isolated from those of other events by techniques
might
to make them more vivid or salient. The subject is in pretty much
the same state with respect to observation of events controlling his
behavior as the observer is. The way that we know better than
anyone else what causes our behavior is not that we have access to
our internal sources of control but we have more
simply that
behavioral data.
and timing of events with mechanical or written
Counting
aids and the techniques of self-reinforcement (and self-punishment)
are, as we have indicated before, ways to increase the salience of
the relationship between behavior and its consequences. Simply
calories has been found to be as effective in short-term
counting
as self-reward, external monetary reward, aversive
weight reduction
imagery in connection with food, and relaxation training
Wilson and 1973). The reason for
(Romanczyk, Tracey, Thorpe,
this may be that the reward for eating less is losing weight regardless
of the subsidiary rewards inserted between the two events. These
rewards do no more than the relation between
subsidiary emphasize
and a function as well
eating losing weight, performed just by
counting calories.
3. commitment strategies may be instituted. Like self
Finally,
reinforcement be in force
strategies they may kept permanently.
Behavior to is guaranteed as
according long-term contingencies
as are in force. If I habitually keep my alarm clock
long they
across the room from my bed, I will have to get out of bed every
morning to turn it off. This technique of getting myself up in the
one that can But often
morning is conveniently be used every day.
commitment involve awkward or expensive apparatus,
strategies
and because are commitments they limit choice. Ironically,
they
in the ultimate term, itmay be better to occasionally behave
long
to short-term contingencies (i.e., to act
according impulsively).
commitment does not allow such behavior and may be
Rigid
undesirable for that reason. Commitment strategies thus might
often be instituted only to bring behavior into conformity
initially
with long term consequences. Once these consequences are ex

105

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Self-control

serve to maintain behavior by themselves.

perienced they may
For instance, consider the following studies being done by G. T.
Wilson at Rutgers University's Behavior Research Laboratory.
The aim of these studies is to bring control outside of the laboratory
and into the everyday life of the alcoholic by a seriesof graded
commitments. It has been found easy to control drinking in the
live together for several weeks
laboratory (atRutgers, four alcoholics
at a time) severe electric shock ?
by following each drink but this
alone will not transfer to the life of the alco
procedure everyday
holic. In future studies alcoholics will be asked to commit them
a "contract" which
selves in the morning (the "morning after") to
specifies that they be shocked after each drink they take that
night. Eventually the commitment could be made each day and
cease in
drinking may the laboratory setting. The control of drink
ing would have been transferred from the immediate shock to the
seems more
long-term relationship between drinking and shock. It
likely that non-drinking established in this way will carry over to
everyday life, where temporally extended rewards for non-drinking
are present in other forms, than would
non-drinking established by
immediate punishment for each drink, a form of contingency
notoriously absent from everyday life.

Itmay be naive to expect that harmful habits with

complex and varied etiologies
should be curable by such a straightforward technique but, straightforward as the
not yet been tried, a to focus
technique may be, it has perhaps because of tendency
on the inner when faced with of self-control. If this
organism problems technique
or a similar one does work, it will be the best kind of evidence that such a
focus is misplaced.

REFERENCES
Ainslie, G. The learningof an op?rant to limit the subject's own future behavior. Described
in H. Rachlin, Introduction to modern behaviorism San Francisco: W. H. Freeman, 1970.
Bandura, A. Vicarious and self-reinforcement processes. In R. Glaser (Ed.), The nature of
reinforcement. New York: Academic Press, 1971. Pp. 228-278.
Bandura, A. and Perloff, B. Relative efficacy of se If-monitored and externally imposed
reinforcement systems. Journal of Personality and Social Psychology, 1967, 7, 111-116.
Baum, W.M. The correlation based law of effect. Journal of theExperimental Analysis of
Behavior, 1973,20, 137-153.
Bloomfield, T.M. Reinforcement schedules: Contingency or contiguity. In R.M. Gilbert,
and J.R. Millenson (Eds.), Reinforcement: Behavioral Analysis. New York: Academic Press,
1972. Pp. 165-208.
Bolles, R.C. Reinforcement, expectancy, and learning. Psychological Review, 1972,
79, 394-410.
Catania, A.C. Elicitation, reinforcement and stimulus control. In R. Glazer (Ed.), The nature
of reinforcement. New York: Academic Press, 1971. Pp. 196-211.
Fantino, E. Immediate reward followed by extinction vs. later reward without extinction.
Psychonomic Science, 1966, 6t 233-234.
Gamzu, E. and Schwartz, B. The maintenance of key pecking by stimulus-contingent and
response-independent food presentation. Journal of the Experimental Analysis of Behavior,
1973,79, 65-73.
Herrnstein, R.J. Method and theory in the study of avoidance. Psychological Review,
1969, 76, 46-69.

106

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions
 Howard Rachlin

Herrnstein, R.J. and Hineline, P.N. Negative reinforcement as shock-frequency reduction.

Journal of the Experimental Analysis of Behavior, 1966, 9, 421-435.
Klein, R.H. The effects of length of delay interval and expectancy control on preferences
for delayed rewards in children and normals and schizophrenic adults. Dissertation Abstracts,
1967,2S, 1199.
Lambert, J.V., Bersh, P.J., Hineline, P.M. and Smith, G.D. Avoidance conditioning with
shock contingent upon the avoidance response. Journal of the Experimental Analysis of
Behavior, 1973, i9, 361-367.
Logan, F.A. Decision making by rats: Delay versus amount of reward. Journal of Compara
tive and Physiological Psychology, 1965,59, 246-251.
Mahoney, M.J. and Bandura, A. Self-reinforcement in the pigeon. Learning and Motivation,
1972,5, 293-303.
Maier, S.F., Seligman, M.E.P. and Solomon, R.L. Pavlovian fear conditioning and learned
helplessness. In B.A. Campbell and R.M. Church (Eds.), Punishment and aversive behavior
New York: Appleton-Century-Crofts, 1969. Pp. 299-343.
Maitland, S. Time perspective, frustration failure and delay of gratification inmiddle-class
and lower-class children from organized and disorganized families. Dissertation Abstracts,
1967,27(B), 3676-3677.
Mischel, W. Preference for delayed reinforcement: An experimental study of a cultural
observation. Journal of Abnormal and Social Psychology, 1958,56,57-61.
Mischel, W. Theory and research on the antecedents of self-imposed delay of reward. In
B.A. M?her (Ed.), Progress in experimental personality research. Vol. 3. New York: Academic
Press, 1966. Pp. 85-132.
Mischel, W. and Metzner, R. Preference for delayed reward as a function of age, intelligence
and length of delay interval. Journal of Abnormal and Social Psychology, 1962, 64, 425-431.
Rachlin, H.C. Introduction to modern behaviorism San Francisco: W.H. Freeman, 1970.
Rachlin, H. and Green, L. Commitment, choice and self-control. Journal of theExperimental
Analysis of Behavior, 1972, 17, 15-22.
Romanczyk, R.G., Tracey, D.A., Wilson, G.T. and Thorpe, G.L. Behavioral techniques in
the treatment of obesity: A comparative analysis. 1973, unpublished manuscript.
Rotter, J.B. Social learning and clinical psychology. New Jersey: Prentice Hall, 1954.
Seligman, M.E.P., and Johnston, J.C. Does teleology have a place in conditioning. In F.J.
McGuigan and D.B. Lumsden (Eds.), Contemporary approaches to conditioning and learning.
New York: Halsted Press, 1973.
Skinner, B.F. What is psychotic behavior? Theory and treatment of the psychoses: Some
newer aspects. Paper presented at thededication of theRenard Hospital, St. Louis, October, 1955.
Staddon, J.E.R. On the notion of cause, with applications to behaviorism. Behaviorism,
1973,7, 25-64.

107

This content downloaded from 91.229.248.134 on Sat, 21 Jun 2014 02:41:17 AM

All use subject to JSTOR Terms and Conditions