Transcript here

NSCI 5702
PRINCIPLES OF
TRAINING
LECTURE OUTLINE
1.
2.
3.
4.
What is training?
Why is it useful to train animals?
How do animals learn?
Training techniques used with animals
WHAT IS TRAINING?
“The shaping of an animal so that it behaves in a way that
humans desire.” UFAW (1992)
WHY IS IT USEFUL TO TRAIN ANIMALS?






Husbandry and health purposes
Safety of handler
Treatment of problem behaviour
Human assistance
Entertainment/Education
Enrichment
TRAINING TECHNIQUES USED WITH
ANIMALS
1.
2.
3.
4.
Classical Conditioning
Desensitisation
Counter Conditioning
Operate Conditioning
I.
II.
III.
IV.
Shaping
Positive Reinforcement
Negative Reinforcement
Combinations of positive and negative reinforcement
5. Flooding
6. Punishment
To understanding how to train animals we must first have an
understanding of how they learn – Learning Theory
HOW ANIMALS LEARN
HOW ANIMALS LEARN
 There are a number of dif ferent forms of learning. We will
discuss the following four:
1.
2.
3.
4.
Sensitisation
Habituation
Imprinting
Associate Learning
Classical Conditioning
Operant Conditioning
SENSITISATION
“The increasing of a response to a repeated stimulus”
(Broom and
Fr a s e r : D o m e s t i c A n i m a l B e h av i o ur a n d We l f a r e , 2 0 07 )
 The animal learns to respond to a stimulus
 Adaptive for survival.
 E.g. Gazelles reacting to the sound of a twig breaking (signals
approaching predator?)
 E.g. A rat that has just experiences an aversive stimulus, such
as a bright light will immediately afterwards be extra
sensitive to other cues, such as noises or lights, that it would
not normally respond to.
DESENSITISATION
 A decrement in response that is produced by gradual exposure
to a stimulus that elicits the response.
 Commonly used in training.
 E.g. Using a tape recording of a particular sound which a dog
is fearful of. Tape is played very softly at first and only
gradually increased in volumes at increments designed to
elicit no response.
HABITUATION
 The animal learns not to respond to irrelevant stimuli.
 This decline in response is specific to a given stimulus.
 Animals will not habituate to relevant stimuli e.g. those
associated with predators, food or mates.
 Advantageous in that it saves energy that would be wasted on
repeated response to trivial stimulus.
 E.g. Zoo animal habituate to the presence of visitors.
 E.g. Sheep habituate to the sound of passing
traf fic.
IMPRINTING
 Phase-sensitive learning that is rapid and apparently
independent of the consequences of behaviour
 Phase sensitive learning = learning occurring at a particular age or a
particular life stage.
 E.g. Chicks hatch with an innate tendency to approach and follow
their mother.
 They have already imprinted on her vocalisations.
 After hatching (24-36hrs) they imprint on her visual appearance.
 Other young animals imprint on olfactory cues from their
mothers.
 Also has an impact upon the animals future choice of a sexual
partner
 http://en.wikipedia.org/wiki/File:Anas_platyrhynchos_ Boston_Harbor,_Massachusetts,_USA -_parent_and_chicks -8.ogv
IMPRINTING
ASSOCIATIVE LEARNING
 There are 2 types of associative learning:
1. Classical (Pavlovian) Conditioning
2. Operant (Instrumental) Conditioning
CLASSICAL (PAVLOVIAN) CONDITIONING
CLASSICAL (PAVLOVIAN) CONDITIONING
 When an animal learns to associate a conditioned stimulus
(bell ringing) with an unconditioned stimulus (food) and
eventually elicits a conditioned response (salivation)
 Definitions:
 Primary or Unconditioned Stimulus (US) = Stimuli that animals react
to without training
 Secondary or Conditioned Stimulus (CS) = Stimuli that have been
associated with a primary (unconditioned) Stimulus
CLASSICAL (PAVLOVIAN) CONDITIONING
 When an animal learns to associate a conditioned stimulus
(bell ringing) with an unconditioned stimulus (food) and
eventually elicits a conditioned response (salivation)
CLASSICAL (PAVLOVIAN) CONDITIONING
CLASSICAL (PAVLOVIAN) CONDITIONING
Examples:





Knock on the door and bark – people
Keys and run to you – leaving/car
Tin opening and cat meowing – food
Hay basket and squeaking GP’s food
http://www.youtube.com/watch?v=WfZfMIHwSkU
CLASSICAL (PAVLOVIAN) CONDITIONING
Application
 In training to be used prior to a reinforcer
 E.g. a Bridge
 Clicker
 Whistle
CLASSICAL CONDITIONING: EXTINCTION
 If the food is no longer presented with the bell, causing the
dog to salivate less in response to the bell
OPERANT CONDITIONING
 Thorndike (1898) put hungry cats into a ‘puzzle box’ with a
lever mechanism that opened a door which lead to a food
reward.
OPERANT CONDITIONING
 Thorndike concluded that it was merely a process of trail and
error.
 The box invoked a series of trial and error voluntary actions. The
cat learnt to press the lever to escape (a rewarding experience).
 “a response that is followed by a reward is more likely to recur
whereas one that is followed by an unpleasant experience is less
likely to occur again? (laws of effect)
 Learning is the result of associations forming between stimuli
and responses. Associations are weakened or strengthened by
the nature and frequency of stimulus -response (S-R pairings)
 The animal learns to associate its own behaviour with a
particular outcome. If the outcome is rewarding e.g. access to
food, the animal learns to repeat the behaviour that resulted in
food access previously.
OPERANT CONDITIONING
 Operant conditioning is “the
type of learning in which the
probability of a behaviour
recurring is increased or
decreased by the
consequences that follows.
 This includes
positive/negative
reinforcement and
positive/negative punishment.
 Forms an association between
a behaviour (voluntar y) and a
consequence.
REINFORCEMENT AND PUNISHMENT
Definitions:
 Anything that increases a behaviour = Reinforcer
 Anything that decreased a behaviour = Punisher
Consequences:
1. Something good can START or be presented = behaviour
increases (Positive Reinforcement)
2. Something bad can END or be taken away = behaviour
increases (Negative Reinforcement)
3. Something bad can START or be presented = behaviour
decreases (Positive Punishment)
4. Something good can end or be taken away = behaviour
decreases (Negative Punishment)
REINFORCEMENT AND PUNISHMENT
REINFORCEMENT AND PUNISHMENT
 Examples
1. Positive Reinforcement (R+)


Adding something good to increase behaviour
Food
2. Negative Reinforcement (R -)


Removing something bad to increase behaviour
Elephant training, horse reins
3. Positive Punishment (P+)


Adding something bad to decrease a behaviour
Shock collar, physical punishment
4. Negative Punishment (P -)


Removing something good to decrease a behaviour
Time out.
REINFORCEMENT AND PUNISHMENT
 Definitions are based on their actual ef fect on the behaviour
in question.
 They must reduced or strengthen the behaviour (to be defined
as a punishment or reinforcer).
 Pleasures meant as rewards but that do not strengthen the
behaviour are indulgences not reinforcement.
 Aversives meant as a behaviour weakener but which do not
weaken behaviour are abuses, not punishment.
REINFORCERS AND MOTIVATION TO
LEARN
 Rewards mean dif ference things to dif ferent animals (e.g.
food, toys, af fection, other animals).
 You must first establish what motivates the animal.
 This could be:





Food
Social contact
Toys
Praise
clicker
REINFORCERS
 Primary Reinforcer:
 (e.g. food)
 A stimulus or event that is
inherently rewarding to the
animal
 Secondary Reinforcer:
 (e.g. clicker)
 Initially meaningless stimuli or
event becomes inherently
rewarding after repeated
association with primary
reinforcer
http://www.youtube.com/watch?
v=hgDHWLyztCI&feature=related
REWARDS
 Ti m ing i s eve r y t h ing! Th e rewa rd
m us t o c c ur w i t h in 1 - 2 s e c on ds o f
t h e be h av iour.
 Th e fre q ue n c y o f t h e rewa rds i s
a l s o i m por t a n t . D uri n g t ra i n ing by
rewa rdi ng fo r eve r y c o rre c t
be h av iour t h e n g ra dua l ly s w i t c hing
to i n te rmit ten t va ri able ra te s .
 D o n ’ t ph a s e o ut rewa rds to o
q ui c kly.
 H e l p to m a ke t ra i n ing s e ssion
e n j oyable a n d to s t re n gt h en t h e
h um a n a n i mal bo n d.
 h t t p: / / w w w. yo ut ube. c om/wat c h ?v=
bD Z Cy ObM fk A
PUNISHMENT
 What is punishment?
 “an aversive action or unpleasant sensation (not necessarily physical)
applied either during or within on e second of a particular behaviour
that reduces the likelihood of that behaviour being repeated in the
future.”
 Differs from negative reinforcement (where the aversive stimulus is
applied before the behavioural response)
 E.g. Hitting a Zebra a few seconds after they have bitten you.
WHY PUNISHMENT MIGHT NOT WORK
1. Pain, fear, anxiety, learned helplessness and stress which
are all welfare concerns.
2. Pain, fear, anxiety learned helplessness and stress also
interfere with the animals ability to learn and focus.
3. It can intensify the occurrence and severity of behaviour
problems.
4. Dif ficulty in getting the timing right causing association
with the wrong things.
5. It becomes meaningless (desensitisation).
6. Breakdown in the trainer-animal relationship.
SHAPING
 Specific behaviour may be “shaped”
 Involves teaching the desired behaviour pattern one step at a
time through operant conditioning.
 The animal needs to be rewarded for behaviour that resembles
the eventual behavioural goal.
 Initially reinforcement is given to an approximation of the
behavioural goal.
 Reinforcement continues as the animals behavioural
approximations develop to resemble the final behaviour more
closely.
 Eventually, only the more precise behaviour is rewarded.
http://www.youtube.com/watch?v=g6F0bRTurPk&feature=relate
d
COUNTER CONDITIONING
Animal is encouraged to engage in another behaviour that is
more pleasurable and which cannot be performed
simultaneously with fear responses in the presence of the
triggering stimulus.
E.g. Feeding a vet phobic Giraf fe
whilst earing a vets uniform.
FLOODING
 Prolonged exposure to a
negatively perceived
stimulus at a level that
provokes the response so
that the animal
eventually gives up.
 VERY STRESSFUL AND
POTENTIALLY DAMAGING
 E.g. confining a dog in an
area and playing the tape
at a louder than
appropriate level until the
dog no longer reacts
fearfully
 Used as a last resort and
executed in the most
humane way.
HOW ANIMAL TRAINERS USE LEARNING
THEORY
 Trainers should use a
variety of different
learning methods.
 Animal must be
motivated to learn.
 Minimal use of negative
reinforcement because of
stressor/& fear – not a
good learning state for
the animal.
 Shaping behaviours –
helping the animals to
learn.