Module 3 - Learning
(Group 3 - Pranav, Rutuja, Satvik)
A good starting point to start off with ‘Learning’ would be to ask the question: How do we
become the people we are?
Let us look at the example of Olive ridley. It has the capacity to navigate and identify sea
currents, which has been their routine for millions of years; it is instinctual. It has been evoked
by maturation processes and it is their genetic predisposition to do so. Unlike turtles, humans are
not born knowing how to swim or walk. ‘How exactly do we learn’ is what this module focuses
on.
Reflexes, Instincts and Learning
Reflexes are a motor or neural reaction to a specific stimulus in the environment. (e.g., the knee-
jerk reflex and the contraction of the pupil in bright light)
Instincts are innate behaviors that are triggered by a broader range of events, such as maturation
and the change of seasons. They are more complex patterns of behavior, involve movement of
the organism as a whole (e.g., sexual activity and migration), and involve higher brain centers.
What is Learning?
Learning refers to relatively permanent change in behaviour or knowledge that results from
experience. For learning to happen, there is a neurochemical change which occurs that is
irreversible in nature.
Learning involves inferring the nature of causality i.e learning often involves figuring out what
causes what. Some learnings are taken for granted and may even appear reflexive.What looks
like reflexive action (opposite of learned behaviour) may not be so. Eg: a newborn baby crying
when picked up by a stranger. At first glance, this might seem like a reflexive action. But in
reality, even very young infants begin to distinguish between familiar and unfamiliar people
within weeks.
B.F Skinner and the superstitious pigeons
Skinner's Superstition Experiment was a study conducted by B.F. Skinner in 1947, where he
placed pigeons in a Skinner box and found that even though the delivery of food was entirely
random, the pigeons exhibited superstitious behaviors and developed their own "superstitions"
that they believed were responsible for the delivery of food. This experiment showed that
animals (and potentially humans) may develop superstitious behaviors through association and
reinforcement, even in situations where there is no actual causal relationship between their
behavior and a desired outcome.
(Source: Skinner's Superstition Experiment | Brainkit)
Associative learning: A form of learning that involves connecting certain stimuli or events that
occur together in the environment.
Classical Conditioning
A process by which we learn to associate stimuli and consequently, to anticipate events.
Consider the famous example of Ivan Pavlov and experiments with dogs. Pavlov (1927)
observed that the dogs salivated not only at the taste of food, but also at the sight of food, at the
sight of an empty food bowl, and even at the sound of the laboratory assistants' footsteps
(Openstax, 196).
Ivan Pavlov’s experiment on classical conditioning involved dogs and their salivary response to
food. Initially, dogs naturally salivated when presented with food (an unconditioned response to
an unconditioned stimulus). Pavlov then repeatedly rang a bell (a neutral stimulus) just before
giving the food. After several times when the bell was rung, the dogs began to salivate upon
hearing the bell alone, even when no food was presented. This showed that the dogs had learned
to associate the bell with food. So the bell became a conditioned stimulus and the salivation a
conditioned response.
Glossary:
- Neutral Stimulus (NS) stimulus that does not initially elicit a response
- Conditioned Response (CR) response caused by the conditioned stimulus
- Conditioned Stimulus (CS) stimulus that elicits a response due to its being paired with
an unconditioned stimulus
- Unconditioned Response (UCR) natural (unlearned) behavior to a given stimulus. not
contingent on any other condition.
- Unconditioned Stimulus (UCS) stimulus that elicits a reflexive response This led
Pavlov to come up with two types of responses to stimuli:
1. Before Conditioning:
● Unconditioned Stimulus (UCS): Food
● Unconditioned Response (UCR): Salivation - a natural unlearned reaction.
● Neutral Stimulus (NS): Bell sound - causes no salivation initially.
2. During Conditioning:
● Bell (NS) is rung just before presenting the food (UCS).
● This pairing is repeated several times.
3. After Conditioning:
● The Bell (now a Conditioned Stimulus, CS) alone
● Triggers Salivation (now a Conditioned Response, CR), even without food. He called it
the dogs’ “psychic secretions”.
● The dog learned to associate the bell with food.
● This is called classical conditioning i.e. learning through association.
Principles of classical conditioning:
1. The conditioned stimulus (CS) must come before the Unconditioned Stimulus (UCS)
2. Temporal difference must be minimal
3. Neutral stimulus has to be paired with UCS several times.
4. CS should be distinctive in the environment.
A real life example:
After getting food poisoning from Pooris (UCS - illness), a person X develops a
conditioned response of nausea even at the sight or smell of Pooris (CS). The CS becomes a
substitute for the UCS in triggering the response.
Stimulus Generalisation: Tendency to respond to a stimulus that is only similar to the original
CS with the CR.
Eg: Fear of all dogs.
Stimulus Discrimination: Tendency to stop making a generalized response to a stimulus that is
similar to the original CS.
Eg: Fear of street dogs in particular.
Strength of CR and Time graph
Source: Classical Conditioning | Introduction to Psychology | Course Hero
1. Acquisition (CS + UCS)
● What happens: Conditioned Stimulus (CS) is repeatedly paired with Unconditioned
Stimulus (UCS).
● Example: Bell (CS) + Food (UCS) -> Salivation (CR).
● Graph shows: Strength of the Conditioned Response (CR) increases over time.
2. Extinction (CS alone)
● What happens: CS is presented without UCS.
● Result: The CR gradually weakens and disappears.
● Example: Bell rings without food -> dog stops salivating.
3. Pause and Spontaneous Recovery
● What happens: After a break (pause), CS is presented again.
● Observation: CR may suddenly return (but often weaker).
● Term: Spontaneous Recovery.
● Repeated CS-alone trials - CR extinguishes again.
Surrogate advertising - an example:
● Promoting restricted products (like alcohol) by advertising associated items (e.g., soda,
club soda).
● Taps into classical conditioning - brand names, jingles, models and visuals become
Conditioned Stimuli which evoke desire or response even without the actual product.
Higher-order Conditioning:
Strong CS paired with another NS.
Consider this example: Tiger the cat learns that the sound of the electric can opener means food
is coming.
● So, she gets excited when she hears the sound of the can opener.
Now, the cabinet starts squeaking just before the can opener is used.
● Over time, Tiger begins to get excited when she hears the squeak, even before the can
opener sound.
● This is called higher-order conditioning: the squeak (a new sound) becomes linked to the
original conditioned sound (can opener).
Change in associative strength between a CS and a UCS:
Associative Strength (Rescorla-Wagner Model)
● Associative strength is given by the following formula:
ΔV=αβ(λ−V), wherein:
● ΔV: Change in strength of associat okion
● α, β: Learning rates (dependent on CS & UCS salience)
● λ: Maximum possible learning (based on UCS strength)
● V: Current associative strength
● Used to explain: Why learning slows as association becomes stronger.
Note:
● Classical conditioning looks very reflexive, but it isn’t.
● Some classical conditioning responses are built within the first exposure to stimulus.
Theories of Extinction:
Gradual weakening and disappearance of conditioned response.
- Competition theory: Displacement of CR by alternative acquisitions. Another response
(like ignoring the bell in Pavlov’s experiment or doing something else) competes with the
original CR.
- Competition-frustration hypothesis: Removing rewards generate frustration and there
is space for other responses. Frustration itself might generate new behaviors (e.g. barking,
pacing) that displace the CR.
- Response inhibition theory: Result of learned conditioned inhibition. The stimulus
becomes associated with the absence of the expected outcome. The bell still rings, but the
animal learns to inhibit salivation because it’s conditioned that no food follows.
Conditioned Emotional Response:
Little Albert Experiment (1920)
● Conducted by: John B. Watson & Rosalie Rayner.
● Subject: 11-month-old baby named Albert.
Purpose:
● To show that emotional responses (like fear) can be conditioned in humans through
classical conditioning.
Experiment Setup:
● Before Conditioning:
● Albert showed no fear of a white rat (Neutral Stimulus, NS).
● He was naturally afraid of loud noise (Unconditioned Stimulus, UCS), which
caused crying (Unconditioned Response, UCR).
● During Conditioning:
Watson paired the white rat (NS) with a loud, frightening noise (UCS) multiple
times.
● After Conditioning:
- Albert began to cry at the sight of the white rat alone.
- The white rat became a Conditioned Stimulus (CS).
- Fear became the Conditioned Response (CR).
Stimulus Generalisation:
● Albert’s fear spread to similar objects:
- White rabbit, fur coat, Santa Claus mask - all triggered fear.
● This showed stimulus generalization:
When a conditioned response spreads to stimuli that are similar to the original CS.
Watson & Behaviourism:
● Watson was the founder of behaviorism:
● Focused on observable behavior over internal mental states.
● Believed that all behavior is learned through interaction with the environment
(conditioning).
● Claimed: “Give me a dozen healthy infants… I’ll train any one to become any type of
specialist…”
Operant Conditioning
Operant Conditioning: The learning of voluntary behaviour through the effects of pleasant and
unpleasant consequences to responses.
Edward Throrndike’s Law of effect: Behaviours followed by satisfying consequences are
more likely to be repeated, while behaviours followed by unpleasant consequences are less likely
to be repeated.
In Operant Conditioning, the following terms are used:
Reinforcer: A stimulus, response to which, increases the probability of occurrence of response.
Primary Reinforcer: Meeting a biological need. Eg: food or touch
Secondary Reinforcer: Pairing with primary reinforcer. Eg: Money, GPA, praise.
Punishment: To decrease the occurrence of a behaviour.
The picture below gives insight into types of reinforcements and punishments - positive and
negative.
The most effective way to teach a person or animal a new behavior is with positive
reinforcement.
Complex forms of operant conditioning:
- Shaping: Successive approximations by reinforcing simple steps that lead to a desired,
more complex behaviour. Eg: training therapy dogs.
- Chaining: Established a sequence of responses leading to the final response being
rewarded. Eg: giving treats to a seal after the performance.
Maintaining Trained Behaviour
● Once a behaviour is trained, it needs to be reinforced to be maintained.
● If you do what you were trained to do but don’t get a reward, the behaviour may stop
over time - this is called extinction.
● Some types of reinforcement schedules are more resistant to extinction than others (e.g.,
variable ratio – like gambling).
Example: Not Wearing a Helmet
● Why some don’t wear helmets:
● Every time they rode without a helmet, nothing bad happened.
● This creates a false belief that it’s safe.
● It’s a result of operant conditioning + flawed probability estimation.
Reinforcement Schedules
(Variable Ratio - most productive and most resistant to extinction)
Source: Openstax Psychology Textbook (page 212)
Operant vs Classical Conditioning
Classical Conditioning Operant Conditioning
Goal Associate 2 stimuli to elicit a Increase or decrease the
natural response. likelihood of voluntary
behaviour.
Type of Behaviour Reflexive or involuntary (e.g., Voluntary (e.g., pressing a
salivating, flinching) lever, studying to get
rewards)
Focus Antecedents (what comes Consequences (what happens
before the response) after the behavior)
Timing UCS and CS come before the Stimulus (reinforcer or
response punisher) follows the
response
Operant Conditioning & Probability Learning
● Operant Conditioning: Learning based on rewards and punishments following
behavior.
● Probability Learning:
● People learn to estimate how likely something is based on past outcomes.
● Often done inaccurately (especially when rewards are inconsistent).
Matching Law (Example from rats):
● Rats press two levers to get food.
● They tend to press each lever in proportion to how often it gives food.
● Matching Law:
The probability of a response matches the probability of reinforcement.
Real-Life Applications:
● Choosing routes, who to trust, gambling decisions, etc.
● Behaviour is influenced by past rewards, even if the strategy is not optimal.
Stimulus (S) and Response (R)
● Normally, learning assumes a stable link between Stimulus (S) and Response (R).
● Sometimes, this link breaks down:
- Changing S doesn’t change R, or
- Changing R doesn’t affect the outcome.
Gambler’s Fallacy Example of flawed stimulus-response learning.
● The Gambler’s Fallacy is an example of irrational learning from random outcomes.
● In gambling, people wrongly believe a certain outcome is “due” after a streak (e.g., red
hasn’t come up in a while, so it must come next).
● The brain wrongly imposes patterns on random events. Although there is no actual
reinforcement, there is a conditioned expectation.
Cognition in Operant Conditioning
1960 – Richard Solomon
● Studied escape conditioning: a form of operant conditioning where an organism learns to
escape an aversive stimulus.
● Example: a dog learns to jump a barrier to avoid electric shock.
Learned Helplessness (Seligman)
● Exposure to 60 unpredictable 5-second shocks over an hour.
● Responses and reinforcement are independent - leads to loss of perceived control.
● Effects: Loss of motivation, passivity, lack of escape behavior.
● Seen in humans too (e.g., loss of control over personal life).
● Little attention is paid to the subject’s belief system or internal state.
● Shows limits of behaviorism - can’t explain why the subject gives up.
Mismatch Between Stimulus and Response
● Animals learn that S and R are unrelated, leading to learned helplessness.
● The word “that” implies a direct cause-effect link, but in many cases the subject (e.g., the
animal) lacks this understanding.
● This makes it difficult to explain why the matching law fails in certain situations.
When the Matching Law Fails
● The matching law suggests that behavior should match the probability of reinforcement.
● But in learned helplessness or unpredictable environments, animals or people stop
matching their behavior to rewards.
● They may give up or behave randomly, even if rewards are still available.
Latent Learning (Edward Tolman’s Rat Maze Experiment)
● Introduced cognitive approach to operant conditioning.
● 3 groups of rats in a maze:
■ Always reinforced - steady decline in errors.
■ Never reinforced - no motivation to learn.
■ Reinforced starting Day 11 - sharpest drop in errors, best performance.
● Conclusion:
■ Rats had learned the maze without reinforcement (latent learning).
■ Learning occurred without immediate reward, but was expressed once
reward was introduced.
■ Shows that learning is not just motor behavior - rats formed mental maps.
Source:https://www.vrogue.co/post/latent-learning-tolman-theory-and-
characteristics-science-2023
Insights
● Learning involves mental representation of space, actions, and outcomes.
■ E.g., we don’t learn directions like “50 steps left, 20 steps right”; we build
cognitive maps.
● Challenges the “black box” model of traditional behaviorism.
■ Learning is not just S - R.
■ There’s mental processing in between.
Second Experiment (Cognitive Flexibility)
● Rats learned one path to reward.
● That path was blocked - 12 new paths opened.
● Rats quickly found new path, showing:
■ Use of mental structure.
■ Planning and adaptability, not trial-and-error behavior.
Reward Expectancy and Learning
● Learning depends on expectation of reward, not just response-reward link.
● Reinforcement works best when the learner expects it.
Conclusion
● Learning is not just about observable responses.
● It’s about how the stimulus and response are interpreted mentally.
● Cognitive processes like expectation, mental mapping, and belief systems play a vital
role.
Cognition in Insight Learning
Insight learning emphasizes problem-solving through internal mental processes rather than
through trial and error. This was demonstrated in the work of Wolfgang Köhler with
chimpanzees like Sultan.
● Köhler presented tasks of increasing difficulty which included placing bananas just out of
reach to Sultan.
● In one task, the chimp had to combine sticks or stack boxes to reach the food which was
something they had not been explicitly taught.
● In the Condais experiment, chimpanzees were placed in a cage with a banana outside and
no lever to open it.
● The chimps used stools creatively to bring the banana closer, showing that there was a
disconnect between the stimulus and their previous learned responses.
● The solution came through a sudden realization which was described as the “coming
together of all elements of a problem.”
● This supports the idea that animals (and humans) engage in active meaning-making and
can generate novel solutions, something Skinner’s rigid operant conditioning cannot
explain.
● Insight learning involves perception, reorganization of mental elements, and goes beyond
motor-level learning.
Observational Learning (Albert Bandura)
Insight learning leads us to the next major form of cognitive learning: observational learning
which was explored by Albert Bandura in the famous Bobo Doll experiment.
● The study asked: Is aggression learned through observation?
● Children were shown models (adults) behaving aggressively, non-aggressively, or not at
all.
● After being frustrated or angered, children entered a room filled with:
- Aggression toys: Bobo doll, mallet, dart gun, tether ball with a face.
- Non-aggression toys: tea set, crayons and paper, toy animals, cars.
● They were observed through a one-way mirror to assess how they behaved.
Measures of aggression included:
● Sitting on the doll, punching, or hitting it with a mallet.
● Imitating verbal aggression.
● Mallet aggression beyond the doll.
● Performing non-imitative aggression (new, unobserved forms).
Hypotheses and Conclusions from the Bobo Doll Study
The experiment tested several hypotheses:
1. Children who watched aggressive models would show more aggression than those who
saw non-aggressive or no models.
2. Non-aggressive models would reduce aggression (have an inhibitory effect).
3. Children would imitate same-sex models more than opposite-sex models.
4. Boys would show more aggression than girls due to social gender norms.
Hypothesis/Conclusion Suppo Details
rted?
Most significant difference
Aggressive models increase observed; strongest in boys with
aggression Yes male models
Lowest aggression in non-
Non-aggressive models reduce aggressive model group, effect more
aggression Yes robust for girls
Children imitate same-sex models Especially true for boys (physical)
more Yes and girls (verbal)
Highest for physical aggression;
Boys show more aggression than verbal aggression similar for both
girls Yes sexes
Aggression is learned through Laid foundation for Social Learning
observation, not reinforcement alone Yes Theory
Modelling and gender interact in Both model’s and child’s sex
effects Yes influence the behaviour imitated
Note: The experiment also highlighted the concept of vicarious reinforcement, where children
observed the consequences of the adult's actions. If the adult was rewarded for aggressive
behavior, children were more likely to imitate that behavior, anticipating similar rewards for
themselves Mimicking Aggression: Insights from The Bobo Doll Experiment - Achology
Cognitive Framework of Observational Learning
Bandura proposed that observational learning involves internal cognitive processes:
● Attention - learners must notice the behavior.
● Memory - learner must retain what was observed.
● Reproduction - learner must be physically capable of imitating the action.
● Motivation - learner must have a reason to imitate the behavior (e.g., reward or
identification with the model).
Lev Vygotsky’s Sociocultural Learning
Finally, learning is not only cognitive or observational. It is intertwined within cultural and social
contexts, as proposed by Lev Vygotsky.
● Learning is not just at the individual level, but embedded in social interaction.
● Situated learning happens within specific cultural environments.
● Cognitive processes like categories, concepts, and calculations are shaped by this context.
● The child and the cultural environment function as one unit: learning is co-constructed.
● All learning is mediated through language, tools, and interaction with more
knowledgeable others.
References:
- https://openstax.org/books/psychology-2e/pages/6-introduction (for some of the
definitions and examples)
- https://www.vrogue.co/post/latent-learning-tolman-theory-and-characteristics-science-
2023
- Classical Conditioning | Introduction to Psychology | Course Hero
- https://www.braink.it/principles/skinners-superstition-experiment
- https://www.simplypsychology.org/bobo-doll.html
- https://www.studysmarter.co.uk/explanations/psychology/approaches-in-psychology/
bandura-bobo-doll/
- https://achology.com/psychology/mimicking-aggression-insights-from-the-bobo-doll-
experiment/
- Slides presented in class
