Professional Documents
Culture Documents
BMOD
BMOD
MODIFICATION
M D Kirk-Smith
1
TABLE OF CONTENTS
Summary
1. Introduction
3. Operant Conditioning
4. Behaviour Modification
2
SUMMARY
This report sets out the philosophy of Behaviourism and the principles and techniques of Operant
Conditioning and Behaviour Modification. This is a branch of psychology concerned with how
environmental antecedents and consequences control, and can be used to manipulate, habitual
behaviours. In contrast to conventional consumer research, the behaviourist approach does not
consider consumers' self-report necessary to explain the causes of behaviour.
The behaviours involved in interactions with consumer products, the major class of marketed
products, whether in the supermarket, watching TV or in the bathroom and kitchen, have a
routine or habitual character. Psychological research suggests that conventional consumer
questionnaire methods, based on self- report, are unlikely to determine how such habitual
behaviour is controlled by environmental or technical factors.
The behaviourist approach emphasises the direct experimental determination of how these
factors, e.g., product attributes and effects, control routine chains of responses. It is therefore
likely to be appropriate. Examples of the behaviourist approach are given to illustrate its
concepts, as currently applied in psychology and in potential applications to consumer research.
Finally, the conflict between behaviourism and the cognitive approach, another major branch of
psychology, is discussed with reference to the brain systems for habits and memory.
3
1. INTRODUCTION
Learning theory has been the most influential theory and experimental approach in psychology
this century. It is influential because it has been successfully applied to solve real-world
problems - up to now mostly within the clinical and educational areas. Behaviour Modification is
beginning to be used to investigate the control of habitual behaviour in consumers. It is based on
the principles of Operant conditioning, a branch of learning theory founded on the philosophy of
Behaviourism. This report introduces the background principles.
Commercial applications of learning theory have been limited, possibly because of the
predominance of the non-experimental questionnaire methods currently used in market and
consumer research, which rely on the assumption that peoples' self-report will uncover the
reasons for their behaviour (1).
Many consumer products may be considered low-involvement products, i.e., the consumer is not
predominantly engaged in complex or rational thinking or decision-making during use or
purchase of the product. The behaviours involved may therefore considered habitual or
stereotyped.
In such circumstances it is unlikely that the conventional questionnaires and interviews of market
research will uncover how environmental or product factors control consumers' behaviour, e.g.,
dosing levels, since such factors appear to act at the non- verbal level (2), this possibly being a
function of brain structure (3). Consumers may simply not be aware of the influences on their
behaviour. The explanations that people give may be related instead to the most culturally
plausible or verbally available explanations, the presentation of themselves to the interviewer,
and other "demand" characteristics of the situation.
In short, people may not know why they do something, but can easily provide rationalisations,
thus the assumption that questionnaire data will always provide causal answers for consumer
behaviour is not supported by formal psychological research.
In practical terms, a baseline behaviour may determined within the habit, e.g., the amount of
toothpaste used. Then various factors may be systematically changed to see how they control the
behaviour and its variability, e.g., the amount of toothpaste surfactant, the flavour and the brush
design. The influence of contingent behaviours, e.g., face washing might also be investigated to
see how they drive or control the behaviour. By these means a causal model of how different
factors actually affect habitual behaviour can be determined.
4
2. THE BEHAVIOURIST PHILOSOPHY
Up to about 1900 psychology was defined as the study of mental experience and its data were
largely self-observations as introspection, i.e., individuals would self-reflect on their thoughts
and feelings. The problem is that introspective observations are private; this distinguishes them
from observations in other areas of science, where replication is possible by anyone.
In the early 1900's, and in reaction to this prevailing view, the American psychologist William
James proposed that the sole subject matter of psychology should be the study of behaviour - this
is Behaviorism. He argued that if psychology were to be an objective science then its data must
be observable and measureable. Since only overt behaviour is measureable, introspection was a
futile approach to develop a science.
The Behaviourist view is that it is unnecessary to invoke internal or "mentalistic" processes such
as drive, motives, attitudes, etc., to explain an individual's behaviour. All these are simply
hypothetical constructs inferred from the behaviours themselves. As they cannot be physically
taken out or experimentally manipulated, or measured, speculation about their existence or
influence can serve no useful purpose in explaining behaviour.
For example, in the marketing literature consumer "needs" are frequently cited as an explanation
for why people buy products. The behaviourist view is that needs, being inferred from behaviour,
are in no sense an explanation of the behaviour and add nothing to the analysis. A need could be
redefined as a condition resulting from deprivation and characterised by a special probability of
response. At best, needs and other mental events can be viewed as epiphenomena - secondary
phenomena accompanying and caused by others.
As such, they may be shaped by environmental factors, just like the expressed behaviour, but
need not have any causal link to the behaviour. That is, behavioural and mental events are seen as
running in parallel, both being shaped by the environment, but need not necessarily be causally
related.
What CAN be measured is behaviour and the conditions under which it occurs. These conditions
can be experimentally varied to determine how each affects the behaviour of the organism.
3. OPERANT CONDITIONING
The applied side of Behaviorism is the systematic application of different contingencies, e.g.,
rewards, to manipulate behaviour. Behaviourists focus exclusively on the environmental factors
that control behaviour. No inferences are made about why an animal did something; the
behaviour is simply recorded and the circumstances noted, controlled, or varied. In other words,
5
Behaviorism is an input-output analysis. The methodology to do this was developed primarily by
BF Skinner from about 1930 on, mostly in highly controlled experimental work using individual
rats and pigeons in the "Skinner Box". This is a box in which a pressed lever delivers food.
Typically, the effects of different food reward schedules on rates of lever pressing would be
intensively studied. The emphasis was very much on control, experimentation and replication,
just as in the 'hard sciences. The technology and rules of changing such environmental
contingencies to control behaviour came to be called "Operant Conditioning".
4. BEHAVIOUR MODIFICATION
Behaviorism was the dominant force in academic psychology until the mid-1960's, by which
time the animal experiments in Operant Conditioning had developed many useful principles and
methods.
This represented a revolution in thinking about the treatment of mental illness. Psychoanalysis
viewed an abnormal behaviour as merely a symptom of an underlying (hypothetical) deep-rooted
personality or psychic disorder, which had to be uncovered by long-term guided introspection
into the past and present. In complete contrast, the Behaviorists stated that the only problem was
that the person had learnt an unsuitable response, the result of an unfortunate conditioning
history. Since they knew the laws of learning all that was required to be done was to unlearn the
old abnormal response and to relearn a new, adaptive, response. There was no need to invoke or
invent complicated internal psychological processes or to delve into the person's impressions of
the past, since behaviour is controlled by the current situation.
In Operant Conditioning psychologists have taken simple everyday principles, defined them
more rigorously, and developed them systematically into a "technology of behaviour change",
similar to the way mathematics has been developed from simple principles. This section explains
the main concepts of Operant Conditioning.
6
Most applications of behaviour modification have been in clinical psychology. Consumer
research currently invokes internally factors such as needs and attitudes to explain consumer
choice, and there has been little experimental work on consumer behaviour using an explicitly
behaviourist model. Thus, the practical examples given below are derived from clinical research
with some speculative interpretations of consumer behaviour in behaviourist terms.
There are two learning theories or approaches: Classical conditioning and Operant conditioning.
Before outlining the key concepts in Operant conditioning it will aid clarity to distinguish the
two. Both are experimentally derived and both deal with HOW things are learnt and behaviours
altered -the difference is in WHAT is being learnt.
Classical (or Pavlovian) conditioning is concerned with the rules and conditions governing how a
natural or innate response to one object can be transferred to an object which does not currently
cause this response. The most famous example is Pavlov's experiment where a dog's natural
response of salivation to meat is transferred to the sound of a bell. Other arousals or responses
that might be transferred are fear, the eyeblink reflex, and avoidance behaviour. The important
point is that something must first happen to elicit the related arousal response, i.e., the animal is
passive. The commercial applications of classical conditioning arise when the arousal is
evaluation or liking, what are the rules underlying the transfer of evaluation or liking, i.e., how
do we get someone to like something, and in the transfer of connotation, i.e., how can the
attributes of one object be transferred to another.
In contrast, Operant conditioning is concerned with the rules of manipulation of any spontaneous
actions through changing the consequences or the setting of the behaviour. For example, a
hungry pigeon may wander around a box, perhaps accidentally pecking a lever, whereupon food
is presented and the pigeon may peck again. Operant conditioning might investigate, for
example, how this pecking behaviour may be controlled or increased by how often the food is
given, or the effects of signalling the food by a light. The important point is that, unlike Classical
conditioning, the animal must first actively "operate" on the environment before experimental
manipulations can begin, hence the term "Operant". The commercial applications arise when the
behaviours are routine or stereotyped, i.e., the habit of using the product or of purchasing in the
supermarket. The questions which operant conditioning might address are -what are the technical
factors that control the behaviour, to what extent do they control the behaviour, how do the chain
of behaviours comprising the habit influence each other, and how can we alter environmental
factors or consequences to alter the habit?
The key terms used by behaviourists in their interpretation and analysis of operant behaviour
follow.
Behaviourists believe that the key to understanding, controlling and predicting an individual's
behaviour is the identifying or arranging of events that follow the behaviour.
A positive reinforcer is simply defined as a stimulus that increases the frequency of the preceding
response or behaviour. "Reward" is not used since it implies something that gives pleasure or
satisfaction - terms which are introspective and thus cannot be clearly defined or measured.
7
The procedure of introducing a reinforcer immediately following a response, with a resultant
change in response frequency, is called operant conditioning. The main rule is that a reinforcer is
most effective in increasing or conditioning a response when it immediately follows the
response.
Rewarding consumers immediately for purchasing our product would be expected to increase
purchases. For example, in the supermarket the most immediate reward that the consumer
receives on actually choosing by picking up, a packet is the feel of the packet. This implies that
the surface texture of the packet should be made pleasurable, since it has a primary position in
the sequence of benefits arising from purchasing the product.
In everyday life, if the conditioned reinforcement is money the task or behaviour is called
"work"; if the conditioned reinforcement is winning or social interaction the same task is called
"play".
Generalised secondary reinforcers like money have acquired their effectiveness through
association with a large variety of other reinforcers. For that reason a generalised secondary
reinforcer will be effective most of the time.
A primary reinforcer will always be stronger than a secondary reinforcer. Thus, the organism
must be deprived of a primary reinforcer for a secondary reinforcer to work. In addition, if a
conditioned reinforcer is not occasionally associated with another reinforcer, its effectiveness
will eventually disappear.
The smell of a product may become a secondary reinforcer through association with its
effectiveness, which in itself can be seen as a secondary reinforcer, because it has been
associated with positive events arising from getting the work done quicker, i.e., being able to do
what you want to do. An important aspect is that the secondary reinforcers are presented sooner
after the required behaviour than the primary reinforcer.
8
5.3 Extinction
One of the first examples of behaviour modification was based on this principle. In mental
hospitals there is little normal social reinforcement. The best way to get attention is to act crazy.
This unfortunate fact may contribute greatly to the development and maintenance of the crazy
behaviour. For three years a patient had a high rate of psychotic talk about an imagined
illigitimate child and men who were pursuing her. We are used to thinking about psychotic talk
as reflecting some inner turmoil or distorted perception. Indeed, nurses would listen attentively
and try to get to "the roots of her problem". It is difficult to imagine her talk as simply a
conditioned response like any other conditioned response - the behaviour is the problem. The
treatment was simply for nurses to ignore her "crazy talk" and only pay attention when she talked
normally, i.e., the "extinction" of crazy talk and the "reinforcement" of normal talk. In the week
before treatment 91% of her talk was psychotic; after nine weeks of this treatment it had fallen to
less than 25%.
In commercial terms, the continued lack of effect of a product will stop people using it. The issue
may be which consequence of using the product actually maintains the behaviour, it may not
necessarily be an obvious consequence such as chemical effectiveness, despite what the
consumer may say. If there is an occasional effect, this can reinforce or increase use more than if
there is always an effect (see Section 5.8).
5.4 Punishers
Punishing events can also be used to establish and maintain a response as well as to eliminate it.
Just as a response which produces attention or food will increase in rate (positive reinforcement),
a response which eliminates electric shock will also increase in rate. In other words, the same
kind of events that acts as punishers for one response can act as reinforcers for another response,
provided that the latter response results in stopping the events. When stimuli are used in this
manner they are no longer called punishers but are called "negative reinforcers". The general
9
procedure is called "negative reinforcement" and the response is called "escape behaviour". In
layman's terms you do what you have to do to escape pain.
This procedure has successfully been used to treat shell-shock paralysis. The patient is told that
he will not be allowed to leave (escape) the treatment room until he is cured (i.e., until normal leg
movement is regained). An electric shock is then given to a "paralysed" limb. If any limb
movement is observed the shock electrode is removed immediately, i.e., the limb response
prevents application of the shock. This can be built up until full limb use is achieved, which may
take a matter of minutes.
In escape behaviour the organism always experiences the stimulation before making a response
that allows escape. Another response is "avoidance behaviour", which is when the organism can
completely evade stimulation by prior action.
In a supermarket, if the selection of one product enabled the consumer to avoid queueing or other
disliked stimuli, the "avoidance behaviour" of buying the product would increase.
As a general rule, any substance or event that has been shown to be a positive reinforcer for one
response may condition any response it regularly and immediately follows. There are constraints
to this imposed by the brain neurology, e.g., it is easier to condition pecking for a food reward
than wing flapping; for escape, pigeons learn faster if the response required is wing flapping,
both of which make sense from an evolutionary perspective. There may therefore be behaviours
which are more conditionable by different contingencies in human beings.
Reinforcement can therefore accidentally condition whatever response happened to precede it. In
pigeon studies, the application of a regular 15 minute food reward causes the stereotyped or
"superstitious" repeating of whatever behaviour happened to initially precede it. It may be
possible that a product is selected in a supermarket through such accidental reinforcement, e.g., if
a product is bought and then the consumer receives the reinforcing smell of fresh bread, the
bakery counter being next on the route. Although the product might have been bought for a
particular reason on the first occasion, future purchases might be maintained by the accidental
reinforcing consequences.
A "response class" is a grouping of all responses which have at least one thing in common. This
might be by similar movements such as eye blinks, or by similar effects of the responses on the
environment, e.g, a lever might be pushed or a product purchased in many ways, but the effect is
the same.
"Response differentiation" is when one member of a response class is reinforced while others are
not, so that it occurs with greater frequency than unreinforced responses. This is called
"differential reinforcement".
A classic example is when students differentially reinforce a lecturer's choice of location in the
classroom; at one position they pay attention, at another they look down. The lecturer will then
10
spend more time at the reinforced location than the other. Differential reinforcement may also be
used to alter verbal behaviour, indeed, it appears that psychotherapists may be reinforcing
patients' positive statements with warmth, understanding and affirmation and extinguish negative
statements by neglecting them, so that the patients eventually learn to say the right things.
5.7 Shaping
Entirely new responses can be generated through reinforcing gradual changes in existing
responses. This is called "shaping" or the "method of successive approximations".
One early example of the use of shaping is in the treatment of a catatonic schizophrenic. He had
been silent for 19 years in mental hospital from the age of 21, there being no known treatment. A
visiting behavioural psychologist dropped chewing gum by accident and noticed that the patient
showed interest. He asked to treat the patient, with intention of creating a "terminal behaviour" of
fluent and appropriate speech. He held a stick of chewing gum in front of the patients' eyes and
when the eyes moved to the gum he was given it. Soon the patient looked towards the
psychologist each time the gum was produced. This is not even an approximation to the terminal
behaviour, but at least one response was now under the control of reinforcement. The response
can now be generalised to something which could be "shaped up". When the experimenter saw a
spontaneous slight movement of the lips he reinforced it with gum immediately. By the end of
the week the patient was making regular eye and lip movements. Now an additional requirement
for reinforcement was added -he had to make a vocal sound as well. This, a croak, was rewarded
immediately. The patient was then encouraged to say "gum" and rewarded as this word began to
be said. By the sixth week, the patient suddenly said "Gum, please", and by the end of the session
the patient was answering any question asked. Talking was then gradually generalised to other
people. Taking the lecturer example above, suppose the students reinforced the lecturer when he
moved slightly to the right of the lectern. Each time he moved right again he would have to go a
bit further right to get the same reinforcement. Eventually, the lecturer would find himself on the
right hand side of the classroom through having his behaviour "shaped", and may not be aware of
what had happened (in fact, this is common trick used by psychology students on their lecturers).
The use of loss leaders or a door prize to entice people into a shop might be considered a form of
shaping. Once people are in the shop, the likelihood of purchasing other items is increased. An
in-use aspect might be the gradual change in the form of a product so that the consumers are
gradually manipulated into using a different behaviour with an effectively new product, this may
be more acceptable than suddenly providing the novel product ab initio.
Shaping of verbal behaviour was widely and successfully used in the "brainwashing" of
American soldiers in the Korean war, e.g., in encouraging informing and anti-American
statements.
Shaping is different from response differentiation in that although both involved differential
reinforcement in the latter the response to be reinforced already occurs. In shaping the terminal
response does not yet occur, and may never have occurred.
11
5.8 Schedules of reinforcement
Much research has gone into how reinforcers might be used to increase (or decrease) the
frequency of a behaviour. The scheduling of reinforcers has been much studied, the possibilities
are to give reinforcement after:
Another variant is the "Differential reinforcement of other behaviour" or the DRO schedule.
Here reinforcement is delivered only when the person is not exhibiting an unwanted behaviour.
This schedule only specifies what responses will not be reinforced.
Under a "fixed ratio" there is a pause in responding after reinforcement and then a sudden burst
of rapid responding, which is reinforced once the required number of responses has been reached
-"break and run" behaviour. It is as if once the organism makes the first response, he responds as
fast as he can to complete the requirement,with little delay between responses. The length of the
following pause is related to the size of the ratio. In the past employers have used such piece-rate
methods to pay workers, who show response styles similar to those found in animals.
The "variable ratio" is the most efficient way of maintaining a behaviour. Like the fixed ratio,
there is a high rate of responding except that there is no pause after the reinforcement. Many
natural reinforcers, e.g., success, profit and achievement, are usually on a variable ratio schedule,
and maintain high rates of responding. Practically all gambling behaviour is maintained by a
variable ratio schedule, with the fruit machine the most obvious example. The brush salesman
who sells on average once every fifty houses is also operating on a clear variable ratio. It is the
chance that the next coin or house might succeed which maintains the behaviour.
One related strategy to maintain purchase behaviour is used by some small shopkeepers. A coin
is tossed for the purchase - "double or quits"; if they lose they will want to try again to regain
their losses, if they win they want to repeat the success. Thus, the purchase behaviour becomes
controlled by a variable ratio reward, like the fruit machine. Similarly, if people are occasionally
given rewards or prizes for purchasing a product, this should act as a maintaining variable ratio
schedule. One experiment found that by giving small rewards on a variable ratio schedule for
riding a bus, the same amount of bus riding could be achieved as when the rewards were given
on a continuous schedule. For approximately one third the cost, the same amounts of behaviour
were sustained.
Under a "fixed interval" the organism must make the response after the interval has ended, only
then will reinforcement be given for a response. There is a variable rate of responding after
reinforcement, starting slowly with gradually increasing rates just before the next reinforcement,
giving a 'scalloped' pattern. A human example is the number of times a kettle is looked at before
it is boiled, or any situation where a deadline is set. If a reinforcement is given regularly, whether
or not a response has been made, the frequency of whatever accidental behaviour preceded the
12
initial reinforcement will be increased. This leads to the development of the stereotyped
"superstitious" behaviour patterns previously mentioned.
A variable interval schedule gives a constant and steady rate of responding, although this is at a
lower rate than the variable ratio schedule. The shorter the times between opportunities for
reinforcement, the faster the responding will be.
In behaviourist terms an advertisment (or other promotional tactic) might be considered a reward
for choosing the product. The most effective schedule for launching a new product, as discussed
above, would therefore be an initial continuous phase of advertising during which the
conditioning of purchase behaviour occurs, followed by an intermittent schedule which serves to
maintain purchase. Thus, in the long-term, a behaviourist analysis predicts that the optimum
strategy would be to use low levels of advertising to maintain sales in a mature market. Indeed,
econometric analysis shows that this appears to be the case (6).
Intermittent reinforcement makes the response more resistant to "extinction", i.e., a reduction in
frequency when the reinforcement is removed. It is found that a response "extinguishes" faster
when it has been reinforced all the time than when it has been reinforced only part of the time.
Similarly, a punished response recovers more slowly if during the punishment phase only
occasional responses are punished than if every response was punished. An increased price might
be considered a punisher -if it reduces sales.
A demonstration of the practical application of some the above ideas is in the treatment of a child
who continually beat his face with his hands, causing severe injury. The child had been in a
straightjacket for several years to stop this behaviour -no treatment had been considered possible.
The behaviourist interpretation was that the original behaviour had been reinforced and
maintained on a "variable ratio" schedule by attention, as with the head-banging child mentioned
earlier. The behaviorist treatment was first to "punish" the behaviour by giving an electric shock
every time he beat his face. This procedure quickly reduced the beating response to zero.
Secondly, a buzzer was turned on and a shock applied, though he could turn both off by holding
a toy truck; thus a truck holding response, i.e., the "escape behaviour", was increased. Thirdly,
the escape procedure was turned into an "avoidance procedure". As long as hands were on the
truck, no shock was received. If he took them off the buzzer sounded, warning of a shock three
seconds later. By these procedures he was trained to keep hold of the truck. The truck could then
be successfully replaced by any soft toy, without the need for a shock or punishing consequence.
However, if the soft toy was removed, the child would start hitting his face again. Skills training
was then successfully started, during which the soft toy was not required, since the child was
now otherwise occupied; the child could be seen holding it only in quiet moments. The point is
that something seen as intractable under a theory of inferred explanatory variables was able to be
changed through a structured behavioural approach.
13
5.9 Discriminative Stimuli
The above operant conditioning concepts have dealt with the effects of consequent
reinforcement. What about the effects of stimuli presented before the response? If a response is
only reinforced in the presence of a stimulus, in consequence the response will always occur
when the stimulus is presented. If the response is never reinforced in the presence of a stimulus,
in consequence the response will never occur when the stimulus is presented.
In the first case, the cueing stimulus for the reinforcement is called a "discriminative stimulus" or
"Sd". It acts like a switch or trigger for the response; "setting the occasion" for the response. In
contrast, a stimulus which is NOT associated with reinforcement is called an "S-delta". When the
response is made with one stimulus and not when the other is present, the response is said to be
"under stimulus control". In effect, this discrimination of responses between the two stimuli is a
combined conditioning and extinction of a response.
In a wartime example, pigeons were trained to aim missiles by pecking at the image of the target
on a screen whenever it went off centre. The pecking caused the missile to redirect itself until it
was on target again. Here the off-centre target became the discriminative stimulus for pecking.
In a supermarket, there may be an area in which more purchases occur, perhaps because more
frequently purchased objects are there. This area might become, with training, a discriminative
stimulus for the purchase behaviour, in which case less purchased products might become
purchased more frequently if they are also placed there. New products might only be bought in
one supermarket but not in others, in which case some aspect of the setting of this supermarket
cues novel purchases. A consumer might only purchase certain items if accompanied by a certain
other person, e.g., a child, where the child serves as the Sd for the particular purchase behaviour.
Similarly, an employee might only work hard when his boss is present. Note that it is the
behaviour that is observed and the situation, not what reasons the people making the responses
may give for their behaviour.
Instead of just an Sd and an S-delta, there can be two classes of stimuli the members of which
have similar properties in common. When a response is to all members of a class then "stimulus
generalisation" has occurred. For example, pigeons were initially trained to discriminate between
painted and non-painted diodes using continuous food reinforcement. They had to peck at the
viewing window at one place to reject and at another to accept. This was then gradually altered
until they could discriminate between transistors with scratches on them and perfect ones. The
behaviour was maintained on a (reduced) variable ratio reinforcement schedule of 50, i.e., food
was only given on average after every 50 correct responses or discriminations.
14
The reinforcement of a response increases the probability of similar responses. In addition,
alteration of one response can inadvertently influence other, different, responses. For example, if
a person is reinforced for smiling, the frequency of laughing and talking might also increase. This
effect is called "response generalisation". If consumers are rewarded for buying one product,
say, through some incentive scheme, purchases of other products may increase as well.
5.12 Fading
Fading techniques allow individuals to perform discriminations which they could otherwise
never perform. The accuracy of this method leads to "errorless discrimination" in that no errors
need to be made to learn a discrimination, an important aspect of learning. The procedure has
now been widely adapted to learning techniques.
5.13 Prompts
Events that can help the development of a response, such as instructions, gestures, directions,
examples and models, are called "Prompts". Prompts precede a response. When a prompt
initiates a response, and the response is reinforced, the prompt becomes an Sd for the response.
For example, if a child is told to return from school early and the child is reinforced for doing so,
the instruction or prompt becomes an Sd. As a general rule, when a prompt consistently precedes
reinforcement of a response, the prompt becomes an Sd and can control behaviour. Gradually,
prompts may be "faded" out after training has been completed.
15
A stimulus-response chain is a sequence of discriminative stimuli and responses. A reinforcer
only alters or sustains the response immediately preceding it. Each response produces some
change in the environment which acts as a discriminative stimulus for the next response. The
chain is finally terminated by some major reinforcement, e.g., food.
This has been used to train long chains of habitual behaviour in both humans and animals. The
last response in the chain is trained first, then the response that precedes that, etc., until the initial
response is conditioned. This is called "backward chaining".
Interpreting a chain of behaviour in this way gives a model of how a routine habit may be
controlled and altered by technical factors. It may therefore be important in understanding the
chains of responses within which consumer products are used. For example, in dishwashing, the
level of dirt or number of plates may act as an Sd for the squeezing of the bottle, and this is
directly reinforced by the warmth of the water on the hand. This in turn serves as a cue for
picking up the brush and beginning washing. Identification of these cues, and their importance
may be done by experimentally varying factors in the chain, right down to the final
reinforcement, to understand how dosing, for example, is controlled. It may also be used to
understand how the toothbrushing habit is created and maintained in children, and thus what
interventions may be made to alter the habit.
This states that if one activity occurs more frequently than another, it will be an effective
reinforcer of that activity. Thus, if there is a response that you want increased, then it will be
driven by arranging for a more frequent behaviour to follow it. It also alters the interpretation of
a reinforcer. Rather than stating that food is a reinforcer, the act of eating food itself may be the
reinforcing activity, if this activity occurs more than the response to be conditioned.
For example, if a less frequently purchased article is arranged to precede a more frequently
purchased article, or perhaps another frequent behaviour, e.g., tea drinking, then purchases of the
first article should increase. In home, if ironing were to be made contingent on a more likely
behaviour, e.g., relaxing, then the frequency of ironing should increase. The notion of a
"Preferred" behaviour is avoided following the behaviourist principle of studying behaviour itself
rather than invoking inferred variables from the behaviour.
5.16 Autoshaping
In general, the phenomena of operant conditioning are distinct from those of classical
conditioning. However, there are cases of behaviour apparently determined by both. One is
Autoshaping. A hungry pigeon is placed in a Skinner box (i.e., a cage containing a light and a
lever which gives food pellets) for the first time. The lever is lit for 6 seconds every minute.
When the light is turned off (automatically) a food pellet is provided. The key remains unlit for
another 54 seconds and is then turned on again, and so on. Note that food is delivered
independently of the pigeon's behaviour towards the key. Nonetheless, after a few trials the
pigeon begins to peck the key when it is lit. This procedure is called "Autoshaping" because this
novel response was created without experimenter intervention, which normal "shaping" requires
16
(Section 5.7). Classical conditioning is involved; by repeated pairings the light is thought to
becomes a conditioned stimulus for the response to food, i.e., the classical conditioning of a
skeletal response, pecking, has occurred. Once the pigeon pecks the key, it is given food almost
immediately, as in standard operant conditioning, which reinforcement then serves to maintain
the response. Closer study shows that if water is used instead of food, the pigeon pecks the lever
as if it was water (closed beak); if food is used, the pigeon pecks the lever as if it was food (open
beak). Thus, stimuli paired with the reward come to elicit the same response as the reward itself -
as in classical conditioning. It is now generally accepted that in situations involving classical
conditioning, an animal will approach and contact the conditioned stimulus, this is called "sign-
tracking".
There are commercial implications. In supermarkets, it appears that much of the behaviour is
habitual and "low-involvement" , i.e, not much thinking is going on. It is under these
circumstances that classical conditioning may occur. If a packet has a pleasant feel, this will be
associated with the prior sight of the packet through constant pairings in use at home -the sight of
the pack predicts the feel. Thus, in the supermarket, autoshaping implies that the consumer will
be conditioned to "sign-track", i.e., to approach and contact, the packet itself.
- There are two reinforcement procedures that increase the frequency of a response; the
application of a positive reinforcer and the removal of a negative reinforcement.
- There are two procedures that decrease, or extinguish, the frequency of a response; the
application of a negative reinforcer (punishment) and the removal of a positive reinforcer.
- Shaping refers to reinforcing small steps or approximations towards a new terminal response.
- Chaining refers to the linking together of responses already within an individual's repertoire, it
is done in a backward direction.
- Antecedent events, such as discriminative stimuli or prompts may also affect behaviour, and
when this is occurs the response is considered to be under stimulus control.
- Stimulus generalisation refers to the transfer of a response to situations other than those in
which training has taken place.
- Response generalisation refers to the altering of responses besides the trained response.
- The Premack principle is that low frequency responses are increased when they are followed by
high frequency responses.
17
- Autoshaping is the unaided initial classical conditioning and subsequent operant conditioning of
an approach to an object.
The Behaviorist approach provided a valuable service by making psychologists aware of the need
for objectivity, measurement and the use of the experimental method. To recap, Behaviourists
believe that the environmental contingencies are all that is needed to be known to specify
behaviour. Inferences about hidden or hypothetical intervening variables are irrelevant, e.g.,
intelligence, mind, mental operations, imagination, reasoning, induction, understanding, thinking,
imagery, symbolic behaviour, and intended meanings.
In contrast, the cognitivists state that behaviourist interpretations ignore the fact that our minds
are not just passive receptors of stimuli, but are constantly processing information into new forms
and categories. To explain behaviour the information processing and mental representations
within the brain need to be understood as well as environmental factors. Here is some of the
background to the cognitive position.
By the 1960's, some psychologists had started taking the ideas of computer science as models for
information processing in human cognition. Cognition refers to the mental processes of
perception, memory and information processing by which the individual acquires and structures
knowledge and can use it to solve problems and to plan. The way incoming information is
selected, compared and combined with information already held in memory, and rearranged and
transformed, will determine the nature of a person's response -the response output depends on
these internal processes and their state at the moment. Behaviorism, with its input-output stance,
ignores such cognitive processes, as a result cognitive psychologists regard it as providing an
incomplete or "empty" account of human functioning.
The notion of a mental model of reality is central to the cognitive approach: "If the organism
carries a small scale model of external reality and of its own possible actions within its head, it is
able to try out various alternatives, conclude which is the best of them, react to future situations
before they arise, utilize the knowledge of past events in dealing with the future, and in every
way to react in a much fuller, safer and more competent manner to the emergencies which face
it" (Craik, 1943).
18
Cognitivists therefore dispute that behaviour can be understood by considering only external or
environmental factors. The crux of intelligence lies in an organism's ability to represent mentally
aspects of the world and then to operate on these mental representations rather than the world
itself. In some cases the operations are like mental trial and error.
Evidence for these mental representations have been found in animal studies. Tolman was an
early cognitive psychologist who studied rats in mazes. He viewed rats as developing a cognitive
map of the maze layout rather than a sequence of left and right handed turns, which would be the
behaviourist view. He showed that rats could learn about locations even though they had not
made any response to them, which would be necessary in the behaviourist approach. Kohler was
another early researcher who showed that chimpanzees could solve problems through insight. To
get bananas out of reach they would pile up boxes into a platform, which would be generalised to
adding or using other objects. The solution would be sudden, would be available afterwards, and
would be transferable to abstractly-related problems. Instead of being a behavioural trial and
error learning, it appeared to be a mental trial and error. Rats when offered food at the ends of
maze consisting of a centre start with eight radiating arms visit the arms randomly rather than
going through a sequence of steps (as a human might do!). After about 20 trials they learn never
to visit an arm they have already been to. The rat has probably developed a "mental map" which
specifies the spatial relations between the arms, and makes a mental note of which arm it has
visited. These facts run counter to a simple behaviourist interpretation.
Rescorla and Wagner's research (7) showed, in a variation of Pavlov's classical conditioning
experiment, that if the meat was presented alone on some occasions and preceded by the bell on
others, the bell did not acquire a salivation response. However, if the bell was presented alone on
some occasions, and on others was followed by the meat, the salivation response was
conditioned. Their interpretation was that in the first group the bell had no predictive power for
the meat, whereas in the latter the meat was always preceded by a bell. Subsequent experiments
showed that this predictive relation was more important than either the closeness of presentation
or the number of pairings, variables which had previously been seen as the most important.
19
The importance of this finding is that "predictability" implies a cognitive act on the animal's part,
it is not responding in some automatic way to the stimulus.
There are human examples of this principle. If a particular signal reliably predicts that pain is
coming, then with no signal the animal can relax. If the signal is not predictive then continuous
stress occurs. If a doctor tells a child that a procedure "will hurt" then the child will be anxious
only during the procedure. If he says "it won't hurt" every time, even though it sometimes does,
then it is not predictive. There being no danger signal, the child will become nervous whenever
in the doctor's surgery. Unpleasant events are unpleasant by definition, but unpredictable
unpleasant events are intolerable.
This change in thinking about the subject matter of psychology away from the study of behaviour
was called the "Cognitive revolution" since it led to the overthrow of Behaviourism as the
predominant force in psychological thinking. Cognitive psychology has been the predominant
force in academic psychology to the present day.
Despite this, behaviorists continue to argue that cognitive psychologists are just as guilty of
unfounded conjecture as the old 'mentalist' psychologists, by giving states of mind as causes of
behaviour rather than the products of causes. Skinner himself (5) comments: "Cognitive
psychologists, like psychoanalysts, observe causal relations between behaviour and genetic and
personal histories and invent mental apparatus to explain them....In reactivating the dream of the
central initiating control of behaviour, cognitive science has drawn attention away from the
accessible variables needed in an effective technology". In behaviourist style, he criticises
cognitive science of:
1. Misusing the metaphor of usage. The brain is not an encyclopaedia, library or museum. People
are changed by their experiences; they do not store copies of them as representations or rules.
2. Speculating about internal processes with respect to which they have no appropriate means of
observation. Cognitive science is often only premature neurology.
4. Reviving a theory in which feelings and states of mind observed through introspection are
taken as the causes of behaviour rather than as parallel or collateral effects of causes.
5. Claiming to explore the depths of human behaviour, of inventing explanatory systems which
are admired for a profundity which is more properly called inaccessibility.
6. Relaxing standards of definition and logical thinking and releasing a flood of speculation
characteristic of metaphysics, literature, and daily intercourse, perhaps suitable enough for such
purposes but inimical to science.
20
A different type of memory appears to be involved in the remembering of facts from the memory
required to retain skills. Studies of amnesia have shown striking differences in disruption of the
two memory systems. Thus, while amnesiacs may be unable to remember old facts or new ones,
they retain motor skills (such as tieing shoelaces or riding a bike) and perceptual skills (such as
reading). Amnesiacs improve with practice at developing a skill the same as normals, but may
have no recollection of the practice trials, these being facts.
Skill knowledge is "knowing how" and fact knowledge is "knowing that". The knowledge in a
skill appears to be represented by the procedures needed to perform the skill, and such
knowledge can only be retrieved by executing the procedures involved.
Recent work in the neurophysiology of memory indicates a different location for the two
memory systems. There appear to be two separate brain systems at work, which have been
identified by various workers as:
Mishkin (3) uses the "habits verses memory" distinction, and this is the one which is most
appropriate to the present discussion. In the habits system a sequence of responses is directly
controlled by the environmental stimuli, i.e., the "stimulus-response" chaining of the
behaviourists. All animals show such direct responding to the environment. As such, the habits
system is considered the most primitive system. On the other hand, more advanced animals have
a highly developed cortex, within which is constructed a "mental model" or organised memory of
the outside world. Being able to use a mental model frees animals from the "tyranny" of
responding mechanically to what is happening moment to moment around them as in habitual
responding. The use of language by humans is the most sophisticated model of the world
constructed by the cortex.
Therefore the Behaviourists' view is analogous to behaviours controlled by the habits system,
whereas the Cognitivists view is analogous to behaviours controlled by the memory system, i.e,
more intellectually based. There is not necessarily always a conflict between the Behaviourist
and Cognitivist approaches; they are simply looking at different aspects of what controls
behaviour.
The behaviours involved in interactions with many consumer products, whether in the
supermarket, watching TV or in the bathroom and kitchen, have a routine or habitual character.
This suggests that a behaviourist analysis, based on the above concepts, with its emphasis on
elucidating the factors controlling such routine chains of responses, is appropriate. However, it is
21
likely that any routine behaviour is going to involve aspects of both the habitual and cognitive
systems. The cognitive approach of choice is Script Theory, which looks at how people think
about and organise stereotyped sequences of behaviour.
The purpose of the present report is to introduce the basic principles and thinking behind this
influential branch of psychology. The experimental methods devised to understand and
manipulate habitual behaviour deserve a report to themselves (8) and have much wider
application than in behaviour modification. These methods use small numbers of people or
systems and involve baseline measurements followed by systematic interventions, during which
behaviour is monitored, to infer causality. For example, the ABAB design, where A is a baseline
measure and B is the intervention. There are three changes in intervention, so if behaviour
changes each time, then you can be surer it is what you are doing that is having the effect.
Another method is to measure a baseline across several people or systems, and then to introduce
the interventions at different times. If behaviour changes each time the intervention starts, you
can be surer that it is what you did that caused the change.
22
REFERENCES
1. Foxall GR, The role of radical behaviourism in the explanation of consumer choice, Advances
in Consumer Research, vol.XIII, 187- 191, 1986 (Foxall has written many articles and books on
the subject).
2. Nisbett RE and Wilson TD, Telling more than we can know: Verbal reports on mental
processes, Psychological Review, vol.84(3), 231-259, 1977.
3. Mishkin M, Malamut B and Bachevalier J, Memories and Habits: Two neural systems, in "The
Neurobiology of Learning and Memory", Eds. McGaugh JL and Weinberger NM, Guildford
Press, NY, 1984.
5. Skinner BF, Cognitive science and behaviourism, British Journal of Psychology, vol.76, 291-
301, 1985.
6. Unilever report: The Measurement of the Effect of Marketing Variables and Product Quality
on Sales, Implications for Research and Advertising. G Hill.
8. Barlow DH and Hersen M, Single case experimental designs, Pergamon Press, London, 1984.
BIBLIOGRAPHY
1. Whalley DL and Mallot RW, Elementary Principles of Behavior, Meridith Corporation, New
York, 1971.
2. Honig WK and Staddon JER (Eds.), Handbook of Operant Behavior, Prentice-Hall, New
Jersey, 1977.
3. Craighead WE, Kazdin AE and Mahoney MJ, Behaviour Modification: Principles, issues and
applications, Houghton Mifflin, Boston, 1976.
23