Principles of

1
Principles of
Learning Theory
Vaughan Bell vaughan_at_backspace.org
2
Outline

• Roots of behaviourism and its legacy
• Classical conditioning
• Operant conditioning
• Behaviour modification
• Social Learning Theory

3
Roots of Behaviourism

• Largely a reaction to the subjective and
  introspective psychology of structuralism
  (Wilhelm Wundt) and psychoanalysis (Sigmund
  Freud).
• Behaviourists saw psychology as the science of
  behaviour, not the mind.
• And rejected the idea of internal mental states
  such as beliefs, desires or goals, believing all
  behaviour to be a reaction to the environment.

4
Watson (1913)

• From Psychology as the behaviorist views it

Psychology as the behaviorist views it is a
purely objective experimental branch of natural
science. Introspection forms no essential part
of its methods The behaviorist recognizes no
dividing line between man and brute.
John B. Watson
5
Legacy

• Behaviourism is no longer the dominant theory in
  psychology (we will see why later).
• However, it has left us with some important
  empirical findings and theories.
• Particularly
• Classical conditioning
• Operant conditioning
• Which are still used to understand certain
  aspects of learnt behaviour and to develop
  therapies.

6
Classical Conditioning

• In the early C20th Ivan Pavlov was working on the
  physiology of the digestive system in dogs.
• He discovered salivation could be caused by the
  sound of a bell if it had previously been sounded
  when food arrived.

• This is called classical conditioning and
  involves the pairing of a stimulus which already
  elicits a response, with a stimulus which
  initially does not.
• After conditioning both stimuli will elicit the
  same response.

7
Classical Conditioning

• Initially there may be an Unconditioned Response
  (UCR) to an Unconditioned Stimulus (UCS).

8
Classical Conditioning

• The UCS can be presented with an arbitrary
  Conditioned Stimulus (CS).

9
Classical Conditioning

• Which causes the CS to elicit a Conditioned
  Response (CR) when only the CS is presented.

10
Common Example
We eat

• Advertising attempts to associate a UCS (picture
  of attractive person) with a CS (product) to
  produce a CR to the product only.

11
Classical Conditioning

• The more times the CS UCS pair is repeated the
  stronger the CS ? CR association.

12
Extinction

• However, if the UCS is repeatedly omitted the
  association between the CS and the CR weakens.

• This is known as extinction.

13
Extinction
Measure of CR
Acquisition
Extinction
CS UCS
Just CS
14
Generalisation

• If we only conditioned to an exact CS we would
  have to relearn every time we encountered a
  slight difference.
• As it happens, we generalise, so we will show a
  similar CR to a stimulus similar to the CS.
• e.g. we might show a similar CR to Heinz
  Spaghetti and Heinz Beans even though we have
  never seen Spaghetti and the original UCS paired.
• The CR gets weaker the less similar the CS it
  becomes.
• As the following graph shows.

15
Generalisation Graph
Measure of CR
Measure of stimuli difference from original CS
16
Discrimination

• In some situations it is possible to make fine
  discriminations between two (or more) similar
  CSs.
• Heinz may decide it does not want people to
  associated the same CR with both its Spaghetti
  and Beans.
• Initially, generalisation may mean we show a
  similar CR for similar CSs.
• But over time, we can learn the discrimination if
  we encounter each CS paired with different UCSs.

17
Discrimination Graph
Responses to CS1
Measure of CR
Responses to CS2
Number of trials
18
Application Latent Inhibition

• Is the effect where a UCS originally presented as
  irrelevant, takes longer to associate with a
  novel CS.
• LI demo here
• In other words, when you have learnt that a
  stimulus is irrelevant
• learning a new association between the ignored
  stimulus and something else is inhibited.
• This is hypothesised to be useful, as it allows
  irrelevant things to not demand as much attention
  in the future.

19
LI and Psychosis

• Some people show reduced or absent LI
• People with acute schizophrenia
• People high in schizotypy
• It has been suggested that psychosis may result
  from being unable to keep irrelevant stimuli out
  of consciousness.
• However, this effect may be part of a continuum,
  as LI can be absent or reduced in
• People given low doses of amphetamine
• People high in creativity

20
Operant Conditioning

• Most associated with the work of Thorndike and
  B.F. Skinner
• Operant refers to the fact that an organism is
  operating or acting on its environment.
• Feedback from the success of these actions
  determine if they are reinforced or discouraged.

B.F. Skinner

• The feedback happens after the action, so it is
  not a simple pairing of two stimuli as in
  classical conditioning.
• This can be used to shape behaviour.

21
Operant Conditioning

• The likelihood of a behaviour can be increased by
  the use of reinforcement.

behaviour
reinforcer
hard work
sweets

• The likelihood of a behaviour can be decreased by
  the use of punishment.

22
Punishment / Reinforcement

• Of course, the removal of something pleasant /
  unpleasant can act in the same way.

23
What reinforces ?

• A primary reinforcer is something that is
  inherently pleasant (food, rest etc).
• Many reinforcers used in everyday life have
  nothing inherently reinforcing about them.
• e.g. I could give cinema tickets to encourage
  good work but there is nothing about a small bit
  of paper that is inherently pleasant.
• A secondary reinforcer is something which has
  been associated with a primary reinforcer and
  works in the same way.
• Association can happen by classical conditioning.

24
Schedules of Reinforcement

• Continuous reinforcement is where a behaviour is
  reinforced every time.
• A fixed-ratio schedule is where behaviour is
  reinforced consistently on a ratio basis (e.g. 1
  in 4).
• A variable-ratio schedule is where reinforcement
  varies or is unpredictable, although may, on
  average, still occur at a certain ratio (e.g.
  20).

25
Variable Ratio Reinforcement in Action

• Fruit machines are a good example of a variable
  ratio reinforcement schedule.
• Behaviours learnt through variable-ratio
  reinforcement are difficult to extinguish.
• Skinner argued this was how gambling becomes
  compulsive.

26
Behaviour Modification

• These theoretical developments have been widely
  applied in clinical settings to change behaviour.
• Either to encourage beneficial behaviour, reduce
  unwanted behaviour or change a response to a
  previously conditioning stimulus.

27
Desensitisation

• This is often used in the treatment of phobias.
• A phobia can be thought of as an inappropriate or
  out of proportion response to a stimulus.
• Desensitisation involves gradually exposing the
  person to the phobic object and waiting until the
  anxiety subsides.
• i.e. present the CS repeatedly until the CR is
  extinguished.

28
Reciprocal Inhibition

• The concept was originally created by Sherrington
  who noted that some muscles inhibit others. i.e.
  their use is mutually exclusive.
• This can be applied to emotions and feelings, for
  example anxiety and relaxation may reciprocally
  inhibit each other.
• As a therapy this would involved trying to
  associated an emotion or feeling to an event
  which is already associated with a unwanted
  reciprocal emotion.
• e.g. associating something pleasurable with a
  stimulus that evokes fear.

29
Cuing

• Behaviour can be conditioned to cues or prompts.
• This is particularly useful for people who have
  trouble with self-prompting.
• For example after certain types of memory loss.
• It may also be possible to slowly remove the cues
  so the behaviour occurs spontaneously or with
  minimum cueing.
• This is known as fading or the vanishing cues
  method.

30
Social Learning Theory

• Most associated with the work of Albert Bandura.
• Bandura realised that much learning in humans
  (and some animals) was based on observation of
  others.
• Rather than the trial-and-error learning of
  conditioning.

• If the individual is learning without any pairing
  of stimulus or direct interaction with their
  environment, they must have a cognitive
  representation of the process.
• So, behaviourism cannot explain all learnt
  behaviour.

31
Required Conditions

• Bandura called the process of social learning
  modelling and gave four conditions
• Attention must pay attention to the model.
• Retention must be able to remember the observed
  behaviour.
• Motor reproduction must be able to replicate the
  behaviour being observed.
• Motivation / Opportunity

32
Effects of Social Learning

• Social learning may effect behaviour in the
  follow ways
• Teaches new behaviours
• Increases or decreases the frequency of which
  previously learnt behaviours are carried out
• Can encourage previously forbidden behaviours
• Can increase or decrease similar behaviours.

33
Social Learning and Copycat Suicide

• Copycat suicide is where 2nd hand experience of
  suicide (particularly through the media) is
  linked to an increase in suicidality.

34
Social Learning and Copycat Suicide
Pop babe Britney Spears' shocking new video
could spark copycat suicides, experts warned
yesterday Derek Chambers of the National Suicide
Group said "There is a risk of imitation by
people who are vulnerable. The Mirror,
16/04/2004

• Social learning theory has been cited as one way
  of explaining this effect.
• Particularly as media exposure encourages
  attention and retention.

35
Conclusions

• Classical and operant conditioning explain the
  way we associate and learn some behaviours.
• They cannot fully explain all behaviour.
• Among others, Banduras work has shown that we
  can learn without any change in our behaviour.
• And so must have some internal cognitive
  representations to account for this.
• Cognitive psychologys information processing
  view of the mind posits internal representations
  and so has become the dominant model in
  psychology as it has a larger explanatory power.

36
Conclusions

• Behaviour modification can successfully use
  conditioning techniques to change behaviour.
• Or change a response to a situation.