Professional Documents
Culture Documents
Ten Points Every Behaviour Analyst Needs To Remember About Reinforcement.
Ten Points Every Behaviour Analyst Needs To Remember About Reinforcement.
Ten Points Every Behaviour Analyst Needs To Remember About Reinforcement.
A. Charles Catania
Operant behavior is behavior that is sensitive But other sources of resistance can be found in
to its consequences. When operant behavior becomes misunderstandings of reinforcement and how it
more likely because of the consequences it has had, works. The relations among behavior and its
we speak of reinforcement. Some consequences consequences in reinforcement seem simple, but they
produce increases in the likelihood of operant have subtle properties, some of which become
behavior, and others do not. Ubiquitous but not evident only in special contexts. When those
universal, reinforcement is not a matter of properties are not taken into account, reinforcement
associations or stimulus-response connections. can appear to be ineffective or to be accompanied by
Rather, it is a variety of selection, in many ways undesirable side-effects. It is therefore prudent to
analogous to the phylogenic selection of biological review the properties of reinforcement and to
populations over evolutionary time. In ontogenic consider the circumstances in which they may
selection, or the selection of operant behavior, mislead us as we deploy techniques of reinforcement
populations of responses are selected by their and evaluate their effects. In the interests of
consequences. As in phylogenic selection, ontogenic preventing misconceptions and misunderstandings, it
selection is constrained by the range of variations is probably even more important to remind ourselves
available in the current population, and it can either of these properties whenever we present what we
occur naturally or be arranged artificially. know about reinforcement to those outside of the
field of behavior analysis. To those who argue that
With regard to phylogenic selection, artificial reinforcement should not be studied because it can be
selection was taken for granted, as in horticulture, misused, the appropriate rejoinder is that detailed
even while arguments for natural selection continued familiarity with its properties may be the best defense
to face serious challenges long after Darwin's insight. against its misuse.
Similarly, most critics of behavior analysis allow that
procedures for the selection of behavior are useful in In its concern for accountability, behavior
some contexts, such as animal training, even as they analysis typically assesses current behavior before
deny the relevance of the ontogenic selection of modifying it in an intervention. Let us be consistent
behavior in natural environments. Shaping is with that concern by assessing our understanding of
artificial selection and is easily demonstrable, but some properties of reinforcement in a brief quiz
natural ontogenic selection is harder to document before we proceed further. Ten true-false questions
The decades of debate over whether any follow that are correlated with ten points we need to
reinforcement contingencies operate in language remember about reinforcement. This is not meant to
acquisition provide an example. imply that these ten points are mutually exclusive or
that they are the only points we need to remember; in
As suggested by Skinner (1971), some of the fact, some items on the list are implicit in other items
resistance to the concept of reinforcement can be and the list is surely incomplete.
attributed to verbal traditions in contemporary
Western culture, including the languages of freedom The reader may wish to answer these
and control. Perhaps that is one reason why research questions in writing, whether on these pages or on a
on how extrinsic reinforcers might undermine the separate paper, so that the answers will be easily
potency of intrinsic reinforcers has led some of its available for later reference. An answer key will be
proponents to argue that these presumed hidden costs offered in the context of successive discussions of the
should bar the use of reinforcement in schools, several questions, although many readers will not
businesses and other institutions (Kohn, 1993), even need it and perhaps some will even disagree with one
though the claimed effects are only inconsistently or more of the keyed answers. In the latter case,
demonstrable and are small and transient when they readers may wish to give themselves partial credit or
do occur. modify their scores in other ways as they see fit.
There is no penalty for guessing and grades will
presumably remain confidential.
useful diagnostic tool when introducing students to
A True-False Quiz on Reinforcement the topic of reinforcement.
1. Chain pulls produce food and lever pressing The vocabulary of reinforcement includes the
increases. This illustrates reinforcement. terms _reinforcer_ as stimulus and _reinforce_ as a
2. The response classes produced by verb. For example, when a rat's lever presses
reinforcement are defined in terms of their produce food and lever pressing increases, we say
forms or topographies. that the food is a reinforcer and that the lever presses
3. Reinforcers work because they make the are reinforced with pellets. The response that
organism feel good or because the organism increases must be the one that produces the
likes them. consequences. For example, if a rat's lever press
4. A more probable response may be reinforced produces shock and only the rat's jumping increases,
by an opportunity to engage in a less it would be inappropriate to speak of either pressing
probable response. or jumping as reinforced.
5. Extinction depends on an active suppression
of the previously reinforced responding. It is because reinforcement operates on
6. If reinforcers have produced problem responses that we speak of reinforcing responses
behavior, the best solution is simply to take rather than organisms. We say that food reinforced a
the reinforcers away: to reduce a child's bad rat's lever press or that a pigeon's key peck was
behavior, extinguish it. reinforced with water, but not that food reinforced the
7. Following a long string of errors, do not miss rat or that the pigeon was reinforced for pecking or
a chance to reinforce the next correct that a child was reinforced. It is too easy to be
response. ambiguous about contingencies when we fail to
8. Cases in which responses seem to be identify the response that was reinforced. If we have
insensitive to their consequences demonstrate been told only that a child has been reinforced, we do
that some responses cannot be reinforced. not know much about actual contingencies. This
9. Extrinsic reinforcers applied to behavior grammatical restriction forces us to be explicit about
maintained by intrinsic reinforcers undermine which response has been reinforced but it does not
the potency of the intrinsic reinforcers. prevent us from mentioning the organism whose
10. The advantages of reinforcement make behavior had consequences. The moral is that in our
techniques of reinforcement more likely to descriptions of reinforcement procedures we should
spread through a culture than techniques of be explicit about what behavior has been reinforced
punishment. by what consequences.
Before proceeding, the reader may wish to review the Point 2: Topography and Function
items to be sure that no answers have been omitted. Reinforcement creates response classes that
are defined by their functions and not by their forms
Point 1: Specificity of Reinforcers or topographies, so _false_ is also the correct answer
By definition, reinforcement always increases to Question 2. Common contingencies select the
responding relative to what it would have been like members of operant classes, and they do so even if
without reinforcement. Also by definition, that the relations among members are arbitrary. A lever
increase must be specific to the response that press is a lever press whether the rat presses with
produced the consequence. If response A produces right paw, left paw, both paws, chin or rump
food and only response B increases, we do not say (Skinner, 1935; consider also the arbitrariness of the
that response B has been reinforced. Thus, _false_ is sets of slides in the discriminated operant classes in
the answer to Question 1: It is not correct to say that Vaughan, 1988).
reinforcement has been demonstrated when chain
pulls produce food and lever pressing increases. The The distinction between function and
specificity of reinforcement effects to the response topography is particularly crucial as it enters into
that produces the reinforcer distinguishes diagnostic categories. The self-injurious behavior of
reinforcement from other processes that produce two children may be similar in topography, but if the
increases in behavior. Question 1 is therefore a behavior of one child is reinforced socially by
attention and the behavior of the other is reinforced
by avoidance of compliance with simple requests, analysis of self-injurious behavior, restraints that
effective treatment programs designed for the two prevented children with severe developmental
children will have to be radically different (Iwata, disabilities from poking or biting themselves were
Pace, Kalsher, Cowdery, & Cataldo, 1990). The first effective in reinforcing arbitrary responses such as
child must be taught more effective ways of engaging putting marbles in a box (Favell, McGimsey, &
the attention of others and must be brought into Jones, 1978).
situations where attention is more readily available.
Requests must be selected for the second child that Whether particular events will be reinforcers
are appropriate to the child’s competence and the can be a difficult judgment. In the final analysis, the
child’s compliance with those requests must be primary criterion for reinforcement remains whether
reinforced (perhaps in the past such behavior has the consequences of behavior have raised the
instead been punished). likelihood of that behavior. Reinforcers are defined
by their behavioral effects and not by inconsistently
What the behavior does is more important correlated properties such as reports of feelings or
than what the behavior looks like. The moral is that preferences, so the moral is that when we try to
we must define the response classes created by identify which events will be effective as reinforcers
reinforcement in terms of their functions and not in we should assess, not guess.
terms of their forms or topographies.
Point 4: Relativity of Reinforcement
Point 3: Assessing Reinforcers Reinforcement is relative, in the sense that it
Our verbal behavior is often correlated with depends on relations between the reinforced response
our nonverbal behavior, so we should not be and the reinforcer. A less probable response may be
surprised that events that are effective as reinforcers reinforced by an opportunity to engage in a more
are sometimes described in terms of positive feelings probable response. The inverse relation, as in
or strong preferences. Such descriptions are subject Question 4, does not hold. Again, _false_ is the
to the inconsistent practices of verbal communities, correct answer. For example, food is not always a
however, and we must therefore be wary of using reinforcer. When a parent only allows a child to go
them to predict whether particular events will serve as out and play with friends after the child has eaten, the
reinforcers. It is tempting to equate reinforcers with opportunity to play may reinforce the eating.
events that are colloquially called rewards. But
contrary to Question 3, which is false, reinforcers do Difficulties in estimating response
not work because they make the organism “feel probabilities at a given moment sometimes
good” or because the organism “likes” them. Our complicate analyses in terms of the relativity of
everyday language does not capture what is important reinforcement, but the reversibility of the
about reinforcers. For example, in assessments of reinforcement relation has been amply demonstrated
the reinforcers that might be effective in managing experimentally (Premack, 1962). For example,
the behavior of people with profound handicaps, levels of food and water deprivation can be selected
predictions based on staff opinion of what would so that drinking is reinforced by an opportunity to eat
work for each individual were inconsistent with the at one time and eating is reinforced by an opportunity
reinforcers identified by systematically assessing to drink at another. In providing an _a_priori_
each individual's nonverbal preferences among those means for predicting whether an opportunity to
events (Green, Reid, White, Halford, Brittain, & engage in one response will reinforce some other
Gardner, 1988; cf. Fisher, Piazza, Bowman, response, the relativity of reinforcement also avoids
Hagopian, Owens, & Slevin, 1992). the problems of circular definition that were inherent
in some earlier definitions of reinforcement. We
We can sometimes make good guesses about soon repeat ourselves if we begin by pointing to an
what will be effective as a reinforcer because increase in responding when asked how we know that
reinforcers often involve events of obvious biological a stimulus was a reinforcer and then explain that the
significance. But reinforcers are not limited to such response was reinforced when asked why the increase
events. For example, sensory stimuli such as occurred.
flashing lights can be powerful reinforcers of the
behavior of autistic children (Ferrari & Harris, 1981). The significance of reinforcers is based on
Restraint also seems an unlikely reinforcer, but in an the opportunities for behavior that they allow. For
example, when time spent in isolation was used in an reading. Responsible teaching adds extrinsic
attempt to punish the tantrums of a 6-year-old autistic reinforcers only when there are no effective intrinsic
girl, her tantrums increased substantially instead of consequences.
decreasing. This child often engaged in
self-stimulation, such as waving her fingers over her The effects of reinforcers are not permanent.
eyes to create visual flicker, but that behavior was The moral is that if we want to maintain behavior
frequently interrupted by the staff. For her tantrums, after we terminate artificial consequences, we should
time in the isolation room was a reinforcer rather than do so only if natural consequences are in place that
a punisher because the isolation room allowed her to will take over that maintenance.
engage in self-stimulation without interruption
(Solnick, Rincover, & Peterson, 1977). Point 6: Side Effects
Reinforcement and extinction sometimes
The relativity of reinforcement reminds us have side effects, effects that are independent of their
that we should not expect the effectiveness of defining properties. For example, aggressive
reinforcers to be constant across different reinforced responding is sometimes a side effect of extinction
responses, different individuals, or even different because extinction terminates reinforcer deliveries.
time samples of the behavior of a single individual. If food is suddenly taken away from a food-deprived
The moral is that when a reinforcer is effective on rat that has been eating, the rat becomes more active
some behavior in some context, we must not assume and perhaps urinates or defecates. If the food was
that it will be effective on other behavior or even on produced by lever presses, the rat may bite the lever.
the same behavior in other contexts. If other organisms are in the chamber, the rat may
attack them (Azrin, Hutchinson, & Hake, 1966).
Point 5: Reinforcement and Extinction These effects and others, though observed in
The effects of reinforcers are not permanent. extinction, are not produced by the termination of the
Extinction demonstrates that reinforcers have reinforcement contingency, because they also occur
temporary effects. When reinforcement stops, the upon the termination of response-independent food
responding that it had maintained returns to its earlier deliveries. In either case, a rat that had been eating
lower levels. Thus, the decrease in responding stops getting food. In extinction, the termination of a
during extinction is not a special process requiring a reinforcement contingency entails the termination of
separate treatment. Rather, it is simply one of the reinforcer deliveries, and the effects of the latter are
properties of reinforcement. Earlier accounts of necessarily superimposed on the decreases in
extinction treated it as a separate process that actively previously reinforced responding.
suppressed responding, but the phenomena that once
seemed to support such accounts, including If reinforcers have produced problem
spontaneous recovery, disinhibition and rapid behavior, taking them away may produce undesired
reacquisition after extinction, were eventually shown side effects. That is why extinction is not the
to be independently determined by different features method of choice for getting rid of behavior that has
of the transition from reinforcement to extinction been created by reinforcement. For Question 6, as
(Catania, 1998, pp. 71-77). It is false to say, as in for its predecessors, _false_ is the correct answer.
Question 5, that extinction depends on an active Suppose a developmentally disabled boy engages in
suppression of previously reinforced responding. severe self-injurious behavior such as head-banging
or eye-poking and we discover that his behavior is in
If the effects of reinforcement are temporary, large part maintained by staff attention as a
then once we have created new behavior with reinforcer. Because of the harm he might do to
reinforcers we cannot count on its maintenance after himself if we ignore the self-injurious behavior,
our intervention ends. Consider children learning to extinction is not well-advised. Giving him attention
read. Only long after they have learned to name independently of the self-injurious behavior is one
letters of the alphabet and to read whole words are possibility, though that might reduce the behavior
they perhaps ready to read stories, so that reading can only slowly. Another is to use attention to reinforce
become “its own reward.” Until that happens, alternative responses, and especially ones
teachers have no choice but to arrange artificial incompatible with the self-injurious behavior. The
contingencies, using consequences such as praise or self-injurious behavior will decrease as the alternative
other extrinsic reinforcers to shape the components of responses increase.
In general, the solution is not to take the not restricted to a single operant class. For example,
reinforcers away. The better way to reduce a child's a pigeon's pecks on one key may be maintained
misbehavior is to reinforce good behavior. because they are followed by reinforced pecks on a
Reinforcers that create problem behavior are second key (Catania, 1971). In a task that involves
important and, unless they are intrinsically harmful, correct responses and errors over trials, correct
taking them away will probably produce more responses and errors can also constitute two separate
problems than making good use of them by making operant classes. If we reinforce every correct
them contingent on other behavior that is more response and repeat any trial with an error until the
manageable and more productive. Presumably that pigeon gets it right, this correction procedure
is why the experimental literature from applied guarantees that any sequence of errors will be
settings provides relatively few examples of the use followed by a reinforced correct response. Correct
of extinction alone to get rid of problem behavior. responses will probably dominate eventually, because
The more typical practice in an applied setting is the reinforcer most closely follows them. But errors
instead to use reinforcers already known to be potent, may diminish only slowly and may even continue at a
by virtue of the behavior they have created, to shape modest level though they never actually produce the
other behavior. reinforcer, because they are reliably followed after a
delay by a reinforced correct response. Thus, always
Sometimes we inadvertently teach the less reinforcing a single correct response after a sequence
effective alternative, especially when we present just of errors will probably maintain errors. An extended
a few basic facts about learning, as in the technical discussion of how to reduce the
introductory psychology course. Based on a strengthening of errors while maintaining correction
superficial account of reinforcement and extinction, a contingencies would take us too far afield, but all
common misconception about what parents should do methods would involve increasing the separation
to get rid of a child’s problem behavior is that they between errors and subsequent reinforcers. One way
should not reinforce it. Instead, they should ignore is to extend the time to the next trial after every error;
it. Left unanswered are the inevitable subsequent another is to draw from a set of error-prone problems
questions, such as how the parents should handle after each error trial rather than repeating the same
things when other problematic behavior emerges that problem, so that errors for any given problem are
is maintained by the same reinforcer. We should not further removed, on the average, from the eventual
teach parents to ignore the behavior of their children; reinforced correct response.
we should teach them how to use reinforcers more
productively. The side effects of extinction typically Teachers must be alert for situations in which
make it an inappropriate method for getting rid of they may be strengthening incorrect responses along
behavior that has been created by reinforcement. with correct ones that they reinforce. A reinforcer
The moral is that rather than taking the reinforcers that follows a sequence of correct responses will
away we should make good use of them to strengthen probably do a lot more good than a reinforcer that
alternative behavior. follows a single correct response after several errors.
Thus, teachers must judge whether correct responses
Point 7: Delay of Reinforcement are so infrequent that they should be reinforced even
The effects of a reinforcer depend on other though preceded by errors or so frequent that the
responses that preceded it besides the one, usually reinforcer can wait until the student has made several
most recent, that produced it. Thus, when one correct responses in a row.
response is followed by a different reinforced
response, the reinforcer may strengthen both. That Many practical applications of reinforcement
means that the reinforcement of a single correct may include other behavior that precedes the
response after a long string of errors will strengthen behavior that we target for reinforcement. When
the errors along with the correct response. It is such behavior shares in the effect of the reinforcer,
therefore false to say, as in Question 7, that one we may mistakenly conclude that the reinforcer is not
should not miss a chance to reinforce the next correct doing its job very well. But if the reinforced
response in such circumstances. behavior includes response classes that we did not
intend to reinforce, it may simply be doing very well
The effects of delayed reinforcers on a job other than the one we wanted it to do. When
responses preceding the ones that produced them are one response is followed by a different reinforced
response, the reinforcer may strengthen both, so the must consider a treatment program that uses attention
moral is that we should keep behavior that we do not to reinforce more effective and appropriate behavior.
want to reinforce from getting consistently close to But the example should also remind us that we cannot
reinforcers produced by other responses. define response classes by what they look like.