Exam 2 – Tuesday November 5th
Week 4 – Basics of Learning
LEARNING
Learning fundamentally shapes who we are and how we behave
Learning: change in behavior or knowledge/skill that is due to
experience.
All animals learn, it is necessary for survival.
Learning is a continuous process from in-the- womb onward, takes
place many ways:
o Non-associative learning: very simple learning from repeated
exposure to a stimulus (e.g., habituation, sensitization) In
textbook section 6.2
o Learning through association –classical conditioning (drooling
dogs, phobias)
o Learning through consequences – operant conditioning (habits)
o Learning through imitation – social or observational learning
Learning through association
Classical Conditioning
Classical Conditioning: learning through forming associations
between experiences
Extremely simple form of learning, can even take place in the womb!
Pavlov’s dogs
o Pavlov: studied saliva production in dogs. Noticed dogs would
begin to salivate even before the meat powder was presented.
Designed an experiment to verify this observation
o Meat --> Saliva
o Bell -- > Meat
o Bell -- > Saliva
Terminology:
o (UCS) = unconditioned stimulus: a stimulus that provokes an
unconditioned response without previous conditioning (e.g.,
meat)
o (UCR) = unconditioned response: an unlearned reaction to
that occurs without conditioning, a reflex (e.g. salivating)
o (CS) = conditioned stimulus: a previously neutral stimulus that
has acquired meaning through conditioning and the capacity to
evoke a response (e.g. bell)
o (CR) = conditioned response: a learned reaction that occurs
because of previous conditioning (e.g. salivating to a bell)
, Classical conditioning takes place in much of daily life, often
determines unique emotional responses based on our learning
histories
o nostalgic childhood associations: smell of playdough or crayons,
opening songs to childhood cartoons
o romantic associations: ‘our’ song, cologne/perfume
o anxious associations: phobias are often learned response:
conditioning can even occur in a single context –such as in
horror movies, they often condition the audience to
associate the ‘monster’ with a set of sounds. Example:
JAWS
But many stimuli can be paired, not all become easily associated.
Strongest association when:
o Contiguity (timing): onset of CS a bit before or at same time
as UCS, ends with or slightly after UCS
o Novelty: novel CS lead to faster associative learning, because
no other associations
o Biological preparedness: prepared stimuli are those pairings
that we are evolutionarily prepared to learn more easily
Preparedness 1: Taste Aversions
Not all stimulus pairings can create conditioned responses (CR) as easily
Garcia’s rats – showed it is easier to condition taste or smell with nausea
(few trials to learn) than it is to condition light or sound with nausea (more
trials to learn).
Biologically prepared to pair internal threats with internal stimuli –and
external threats with external stimuli.
Stronger with NOVEL tastes.
Preparedness 2: Learned fear
Mineka’s monkeys-- easier to teach a lab reared monkey to fear a snake
(fewer trials) than a flower, even though had never seen either (so both
novel and neutral)
Presumably because primates (including us!) are ‘hard-wired’ to fear snakes.
Evidence in humans in classical conditioning to a snake vs a flower. For
example, “irrational” persistence of the response when the shock electrodes
are removed for snakes, but not flowers. . .
Classical Conditioning Continued
How do conditioned responses change across time?
o Extinction: with repeated presentation of only the CS without
the UCS, eventually the CR will extinguish (ring bell continuously
without meat, dog will stop drooling at the bell)
o Spontaneous Recovery: when the CR rebounds after a time of
no exposure to the CS
, implies we don’t really ‘unlearn’ original association,
rather the initial extinction means we learned a new
response to it
o Discrimination: when learning becomes more specific (e.g. not
all bells, *this* bell means meat)
o Generalization: when learned CR to one CS generalizes to
other, similar CS objects
Watson, Little Albert and the white rat (btw, Little Albert
was not disabled and did not die young, that’s interweb
rumor)
Learning through consequences
Operant/Instrumental Conditioning
Thorndike and Skinner reasoned that we also learn from
consequences of our actions, in addition to associations in the
environment.
Operant conditioning: learning controlled by consequences of a
behavior
Reinforcement: occurs when the consequences of a behavior
*increases* tendency to respond the same way
o positive reinforcement: response increases because get a
positive (e.g. give a reward or praise if kid makes the bed)
o negative reinforcement: response increases because remove a
negative (e.g. stop nagging if kid makes bed)
Punishment: occurs when the consequence of a behavior *decreases*
tendency to make that response in the future
o Positive punishment: response decreases because get a negative
(e.g. spank if bed is not made)
o Negative punishment: response decreases because remove a
positive (e.g. take away cellphone if bed is not made)
o Skinner believed reinforcement generally preferable to
punishment, especially compared to positive punishment:
Positive punishment may carry unintended consequences
(e.g., associate punisher with fear and anxiety). Explains
why you should not use positive punishment during puppy
housetraining, cannot be sure what the puppy is learning
(to fear the action, or to fear the owner)
It is important to note that when discussing reinforcement or
punishment –positive and negative refer to the presence or absence
of a stimulus, NOT whether that stimulus is positive or negative in the
way it makes you feel
o Positive reinforcement –presence of good thing – increases
behavior
o Positive punishment–presence of bad thing–decreases
behavior
, o Negative reinforcement–absence of bad – increases behavior
o Negative punishment –absence of good – decreases behavior
What about training behaviors that might not spontaneously occur?
Shaping: the reinforcement of closer approximations of a behavior
o Shaping: the reinforcement of closer approximations of a
behavior until you get the behavior you want. Initial learning of a
behavior (like bringing slippers in response to a command)
o “Chaining” reinforcing multiple previously learned behaviors
into one behavioral sequence (chain), by reinforcing when the
entire “chain” of previously learned behaviors are emitted in the
right order
To build the chain –reinforce each behavior as it occurs (in
order), then withhold reinforcement until two behaviors are
done in the chain, then three, etc until you build the full
chain)
Example, chain 3 different “tricks” your dog already
knows into one sequence, if your dog knows “lie
down” “roll over” and “pull blanket” as 3 separate
tricks–then you could reinforce each in turn when
done together following a command “go to bed”,
then withhold reinforcement until done in one chain.
“Go to bed” is then a command that uses all 3 prior
tricks in a sequence
Extinction: if a behavior was previously reinforced, then becomes
unreinforced, behavior will extinguish after a time
o Overjustification effect – if you reward a behavior that
otherwise was inherently interesting or rewarding, can reduce
interest in engaging in the behavior without reward
The “marker” study: expecting and receiving a reward for
a drawing task decreased preschool children’s interest in
drawing during free play
Implication: rewarding a child for playing the
piano/reading can reduce his/her natural enjoyment of it
o Resistance to extinction: when an unreinforced behavior
continues to be emitted –often due to types of reinforcers or
schedule used
o Types of Operant Conditioning reinforces and schedules:
Primary reinforcers: events that are inherently
reinforcing because they satisfy basic or biological
needs (e.g., sweet tastes, social approval, loud noises,
pain, etc)
Secondary reinforcers: events that acquire reinforcing
qualities through association (e.g., money, grades, the
‘clicker’ for training animals, etc)
Week 4 – Basics of Learning
LEARNING
Learning fundamentally shapes who we are and how we behave
Learning: change in behavior or knowledge/skill that is due to
experience.
All animals learn, it is necessary for survival.
Learning is a continuous process from in-the- womb onward, takes
place many ways:
o Non-associative learning: very simple learning from repeated
exposure to a stimulus (e.g., habituation, sensitization) In
textbook section 6.2
o Learning through association –classical conditioning (drooling
dogs, phobias)
o Learning through consequences – operant conditioning (habits)
o Learning through imitation – social or observational learning
Learning through association
Classical Conditioning
Classical Conditioning: learning through forming associations
between experiences
Extremely simple form of learning, can even take place in the womb!
Pavlov’s dogs
o Pavlov: studied saliva production in dogs. Noticed dogs would
begin to salivate even before the meat powder was presented.
Designed an experiment to verify this observation
o Meat --> Saliva
o Bell -- > Meat
o Bell -- > Saliva
Terminology:
o (UCS) = unconditioned stimulus: a stimulus that provokes an
unconditioned response without previous conditioning (e.g.,
meat)
o (UCR) = unconditioned response: an unlearned reaction to
that occurs without conditioning, a reflex (e.g. salivating)
o (CS) = conditioned stimulus: a previously neutral stimulus that
has acquired meaning through conditioning and the capacity to
evoke a response (e.g. bell)
o (CR) = conditioned response: a learned reaction that occurs
because of previous conditioning (e.g. salivating to a bell)
, Classical conditioning takes place in much of daily life, often
determines unique emotional responses based on our learning
histories
o nostalgic childhood associations: smell of playdough or crayons,
opening songs to childhood cartoons
o romantic associations: ‘our’ song, cologne/perfume
o anxious associations: phobias are often learned response:
conditioning can even occur in a single context –such as in
horror movies, they often condition the audience to
associate the ‘monster’ with a set of sounds. Example:
JAWS
But many stimuli can be paired, not all become easily associated.
Strongest association when:
o Contiguity (timing): onset of CS a bit before or at same time
as UCS, ends with or slightly after UCS
o Novelty: novel CS lead to faster associative learning, because
no other associations
o Biological preparedness: prepared stimuli are those pairings
that we are evolutionarily prepared to learn more easily
Preparedness 1: Taste Aversions
Not all stimulus pairings can create conditioned responses (CR) as easily
Garcia’s rats – showed it is easier to condition taste or smell with nausea
(few trials to learn) than it is to condition light or sound with nausea (more
trials to learn).
Biologically prepared to pair internal threats with internal stimuli –and
external threats with external stimuli.
Stronger with NOVEL tastes.
Preparedness 2: Learned fear
Mineka’s monkeys-- easier to teach a lab reared monkey to fear a snake
(fewer trials) than a flower, even though had never seen either (so both
novel and neutral)
Presumably because primates (including us!) are ‘hard-wired’ to fear snakes.
Evidence in humans in classical conditioning to a snake vs a flower. For
example, “irrational” persistence of the response when the shock electrodes
are removed for snakes, but not flowers. . .
Classical Conditioning Continued
How do conditioned responses change across time?
o Extinction: with repeated presentation of only the CS without
the UCS, eventually the CR will extinguish (ring bell continuously
without meat, dog will stop drooling at the bell)
o Spontaneous Recovery: when the CR rebounds after a time of
no exposure to the CS
, implies we don’t really ‘unlearn’ original association,
rather the initial extinction means we learned a new
response to it
o Discrimination: when learning becomes more specific (e.g. not
all bells, *this* bell means meat)
o Generalization: when learned CR to one CS generalizes to
other, similar CS objects
Watson, Little Albert and the white rat (btw, Little Albert
was not disabled and did not die young, that’s interweb
rumor)
Learning through consequences
Operant/Instrumental Conditioning
Thorndike and Skinner reasoned that we also learn from
consequences of our actions, in addition to associations in the
environment.
Operant conditioning: learning controlled by consequences of a
behavior
Reinforcement: occurs when the consequences of a behavior
*increases* tendency to respond the same way
o positive reinforcement: response increases because get a
positive (e.g. give a reward or praise if kid makes the bed)
o negative reinforcement: response increases because remove a
negative (e.g. stop nagging if kid makes bed)
Punishment: occurs when the consequence of a behavior *decreases*
tendency to make that response in the future
o Positive punishment: response decreases because get a negative
(e.g. spank if bed is not made)
o Negative punishment: response decreases because remove a
positive (e.g. take away cellphone if bed is not made)
o Skinner believed reinforcement generally preferable to
punishment, especially compared to positive punishment:
Positive punishment may carry unintended consequences
(e.g., associate punisher with fear and anxiety). Explains
why you should not use positive punishment during puppy
housetraining, cannot be sure what the puppy is learning
(to fear the action, or to fear the owner)
It is important to note that when discussing reinforcement or
punishment –positive and negative refer to the presence or absence
of a stimulus, NOT whether that stimulus is positive or negative in the
way it makes you feel
o Positive reinforcement –presence of good thing – increases
behavior
o Positive punishment–presence of bad thing–decreases
behavior
, o Negative reinforcement–absence of bad – increases behavior
o Negative punishment –absence of good – decreases behavior
What about training behaviors that might not spontaneously occur?
Shaping: the reinforcement of closer approximations of a behavior
o Shaping: the reinforcement of closer approximations of a
behavior until you get the behavior you want. Initial learning of a
behavior (like bringing slippers in response to a command)
o “Chaining” reinforcing multiple previously learned behaviors
into one behavioral sequence (chain), by reinforcing when the
entire “chain” of previously learned behaviors are emitted in the
right order
To build the chain –reinforce each behavior as it occurs (in
order), then withhold reinforcement until two behaviors are
done in the chain, then three, etc until you build the full
chain)
Example, chain 3 different “tricks” your dog already
knows into one sequence, if your dog knows “lie
down” “roll over” and “pull blanket” as 3 separate
tricks–then you could reinforce each in turn when
done together following a command “go to bed”,
then withhold reinforcement until done in one chain.
“Go to bed” is then a command that uses all 3 prior
tricks in a sequence
Extinction: if a behavior was previously reinforced, then becomes
unreinforced, behavior will extinguish after a time
o Overjustification effect – if you reward a behavior that
otherwise was inherently interesting or rewarding, can reduce
interest in engaging in the behavior without reward
The “marker” study: expecting and receiving a reward for
a drawing task decreased preschool children’s interest in
drawing during free play
Implication: rewarding a child for playing the
piano/reading can reduce his/her natural enjoyment of it
o Resistance to extinction: when an unreinforced behavior
continues to be emitted –often due to types of reinforcers or
schedule used
o Types of Operant Conditioning reinforces and schedules:
Primary reinforcers: events that are inherently
reinforcing because they satisfy basic or biological
needs (e.g., sweet tastes, social approval, loud noises,
pain, etc)
Secondary reinforcers: events that acquire reinforcing
qualities through association (e.g., money, grades, the
‘clicker’ for training animals, etc)
