Topic 6.2: operant conditioning

Operant conditioning

 Learning wherein the frequency of a behavior is controlled by its consequences

• Also known as instrumental conditioning

• The organism “gets something” because of its response (food, avoids punishment)

The law of effect

stating that behaviors followed by satisfying results (reinforcers) are strengthened and more likely to recur, while behaviors followed by unpleasant results (punishers) are weakened and less likely to happen again, essentially shaping learning through consequences

• The concept contrasts with the predictions of the insight hypothesis

– This theory argues performance only changes once an organism “grasps” the underlying nature of the problem

Thorndikes Puzzle Box

A series of experiments conducted by psychologist Edward Thorndike to study how animals learn through trial and error.

BF Skinner

 Followed up on Watson and Thorndike’s work on behaviour

• Designed the Skinner box to more effectively record operant behavior unsupervised.

a laboratory apparatus called an operant conditioning chamber, created by B. F. Skinner to study animal behavior

Reinforcer

outcomes that strengthen the probability of a response

– Can be positive or negative

• Positive reinforcement involves giving a stimulus

• Negative reinforcement involves taking away a stimulus

Punishment

 any outcome that weakens the probability of a response

• Like reinforcement, can be positive or negative

• Disciplinary actions are punishments only if they decrease

the chance of the behaviour happening again

Does Punishment Work?

 Not as well as reinforcement, say many

• Has several disadvantages

– Tells a person what not to do, but doesn’t inform what they should do

– Creates anxiety

– Encourages subversive behaviour (refers to actions that secretly undermine, resist, or work against authority, rules, expectations, or a system — often in a subtle, indirect, or hidden way)

– May provide model for aggressive behaviour

Discriminative Stimulus

signals the presence of reinforcement. A signal that a reward or punishment is available if a certain behavior is performed)

Schedules of Reinforcement (1)

Refers to the pattern of delivering reinforcers

• In continuous reinforcement, the response is rewarded every time

• Partial reinforcement occurs when we reinforce responses only some of the time

– More resistant to extinction

Schedules of Reinforcement (2)

 Vary along two dimensions

• Consistency of administering reinforcement

– Fixed or variable

• The basis of administering reinforcement

– Ratio or interval

Schedules of Reinforcement: Fixed Ratio

Reward after a set number of responses.

Example: A salesperson gets a bonus for every 10 sales.

Applications of OC

Animal training using shaping by successive approximations and chaining

• Encouraging behavior via the Premack principle

– Positively reinforce a less performed behavior (less desirable) with a more often performed behavior (more desirable)

Primary reinforcers

 naturally rewarding— they satisfy biological or basic survival needs (food, sleep, relief from pain)

Secondary Reinforcers

 get their value by being associated with primary reinforcers— they are learned rewards (money, grades, praise)

Skinner and radical behaviourism

Early behaviourists did not believe that thinking played much of a role in learning (i.e. Radical behaviourism)

• Argued that thinking and emotions are behaviours, just covert ones

• Today, most psychologists acknowledge at least some role for cognition

Stimulus Response vs Stimulus-organism-response

The way an organism responds to a stimulus depends on how that organism interprets the stimulus (meaning)

• Most scientists now believe that C C and O C depend on thinking

Latent Learning

 Refers to learning that is not directly observable

– Competence v s. performance

• Implies that reinforcement is not necessary for learning to occur

• Tolman & Honzik’s studies

– Rats could learn a maze even without reinforcement

– Rats would not show their learning until they received reinforcement

Observational Learning

Learning by watching others (no direct experience

required)

• Don’t have to engage in trial and error to learn something new

• Bandura’s research on observing aggression

Mirror Neurons

Become activated when an animal observes or performs an action

• May play a role in observational learning and having empathy for others

Insight Learning

Koehler’s chimpanzees and “Aha!” moments

– Köhler studied how chimpanzees solved problems — not through trial and error, but by suddenly realizing the solution.

– Chimps were placed in situations where food (like bananas) was out of reach. They were given tools such as sticks or boxes. Rather than random attempts, chimps often: Looked at the situation quietly, Thought about the problem, Suddenly acted with a clear solution

• Suggests humans and some other animals may gain insight

Biological Influences on Learning

Conditioned taste aversion.

• Develops after only one trial.

• Learning is still possible even with very long delays

(6-8 hours)

Problems with Biological Influence

Shows little generalization

– Contradicts equipotentiality

Biological influence impacts

We are biologically predisposed (“prepared”) to learn certain fears more easily because they were dangerous in our evolutionary past.

– Snakes and spiders vs. cars and guns.

• This preparedness for certain phobias also contradicts equipotentiality.

• May make us more likely to form certain illusory correlations: We sometimes believe two things are connected even when they're not — just because our brains are prepared to look for that connection.

– Assume certain animals or people are dangerous even without real evidence

– Over-notice threats that match our expectations

– Create fear associations where none exist

Instinctive Drift

Tendency for animals to return to innate behaviours following repeated reinforcement.

 Biological influences place limits on what kinds of behaviours we can train through reinforcement.

Learning Fads

Many techniques are purported to make you learn better

or faster

– Sleep-assisted learning

– Accelerated learning

– Discovery learning

• None of these programs seems to be more effective than traditional learning

• Direct instruction is almost always better

Equipotentiality

a theory in psychology and neuroscience suggesting that many parts of the brain have the same potential to carry out functions. If one area of the brain is damaged, another area can sometimes take over the lost function.

Fixed Interval

Reward for the first response after a set amount of time.

• Example: Bi-weekly paychecks

Variable ratio

Reward after an unpredictable, average number of responses.

Example: Gambling (slot machines), fishing.

Variable interval

Reward for the first response after an unpredictable, average amount of time.

Conditioned taste aversion

organism develops a strong dislike and avoidance of a specific food or taste after becoming ill from it, even if the food was not the actual cause of the illness.