OPERANT CONDITIONING -- learning in terms of rewards & punishments

Generally, operant condition works according to THE LAW OF EFFECT, which simply states that what FOLLOWS a behavior can influence likelihood that the behavior will occurring again.

B. F. SKINNER studied operant conditioning by using a SKINNER BOX (a simplified environment in which a rat pushes a level to dispense food pellets (rewards)). The point here is to get at the PRINCIPLES of operant conditioning.

SHAPING -- generally, the best way to get a creature to learn a complex behavior (X) is first to reward a behavior that approaches X, and then reward a behavior that gets a little closer to X, etc., etc.
 
 

First, here are some handy ways to distinguish classical from operant conditioning:

Classical conditioning involves RESPONDENT BEHAVIOR (an automatic response to a stimulus). Also, the new stimulus (CS) is presented BEFORE the behavior.

Operant conditioning involves OPERANT BEHAVIOR (a behavior that tries to operate on the environment to get rewards & avoid punishments). Also, the new stimulus (e.g., a reward) is presented AFTER the behavior.

Technically, rewards are called REINFORCERS, which by definition INCREASE the likelihood of the behaviors they follow. There are 2 basic kinds:

POSITIVE REINFORCER -- a pleasurable stimulus presented after a behavior.

NEGATIVE REINFORCER -- removal of an aversive stimulus after a behavior. (n.b. this also INCREASES the likelihood of the behavior).
 
 

A reinforcer can also be either a

PRIMARY REINFORCER -- inherently satisfying (e.g., candy)

SECONDARY REINFORCER -- learned (e.g., money, grades)

Technically, PUNISHMENT involves introducing an aversive stimulus after a behavior. By definition, it DECREASES the likelihood of the behavior it follows.

POSITIVE PUNISHMENT -- an aversive stimulus presented after a behavior.

NEGATIVE PUNISHMENT -- removal of an pleasant stimulus after a behavior. (n.b. this also DECREASES the likelihood of the behavior).

REINFORCEMENT SCHEDULES -- different ways of passing out reinforcers produce different patterns of learning.

CONTINUOUS REINFORCEMENT -- giving a reinforcer every time the behavior occurs -- produces fast learning, but also fast extinction.

PARTIAL REINFORCEMENT -- giving a reinforcer only some of time -- produces slower learning, but also more resistance to extinction.

There are four basic SCHEDULES OF PARTIAL REINFORCEMENT:

FIXED RATIO: a reinforcer after a fixed # of behaviors. Fastest rate of learning.

VARIABLE RATIO: a reinforcer after an unpredictable, varying # of behaviors (e.g., gambling).

FIXED INTERVAL: a reinforcer available only after time intervals of predetermined duration. Stop-start pattern of learning, with faster responding near interval's expiring (e.g., checking the mailbox).
 
 

VARIABLE INTERVAL: reinforcer available only after an unpredictable, varying intervals. Slowest learning, with slow, steady responding.

Graphically, it looks like this

# responses fixed ratio variable ratio

|

|

|

| fixed interval

|

| variable interval

|

|

|

|

|---------------------------------------------------------> time
 
 
 
 

Generally, learning via operant conditioning follows the same scheme for GENERALIZATION, DISCRIMINATION, ACQUISITION & EXTINCTION as we saw in classical conditioning.

----------------------------------------------------------------------------------------

COGNITIVE LEARNING

Generally, COGNITIONS enter into how we learn & behave

Creatures form COGNITIVE MAPS (mental representations) based on their simple experience, even if not reinforced.
 
 

BANDURA’S SOCIAL-COGNTIVE LEARNING

(from watching, imitating & modeling)
 
 

OBSERVATIONAL LEARNING, a.k.a. "modeling" -- learning by watching and imitating others.

BOBO DOLL EXPERIMENT - first demonstration of observational learning

4 primary factors: Attention, Memory, Imitation, Motivation
 
 

Also, BIOLOGICAL PREDISPOSITIONS enter into learning

Certain creatures tend to learn certain behaviors more easily than others.

Application: BEHAVIOR MODIFICATION -- the use of behavioral principles to modify behavior -- usually to eliminate problems.

Treatment of autism -- using operant conditioning to shape acceptable behaviors (looking in the eye, speaking, etc.).

Biofeedback -- using awareness of physiological responses (usually via machines) to produce relaxation responses.