PSY 402 - CPP

PSY 402Theories of LearningChapter 7 – Behavior & Its ConsequencesInstrumental & Operant Learning

Two Early ApproachesReinforcement Theory

Thorndike’s “Law of Effect” for cats in the puzzle box.Skinner boxes – rats pressing bars

Contiguity TheoryGuthrie – association is enoughEstes – Stimulus Sampling Theory

Problems with Contiguity TheoryGuthrie proposed that no reinforcement was needed – just contiguity (closeness) in time and place.If learning is immediate and one-trial, why are learning curves gradual?

Only a few stimulus elements are associated on each trial, but more build up with each trial.

His view was wrong but influential (Estes).

7.1 Learning as Guthrie saw it

Guthrie & ReinforcementThe reinforcer is salient, so it changes the stimulus (environmental situation).Reward keeps competing responses from being associated with the initial stimulus.

Competing responses are instead associated with the presence of the reward.

Fixity of cat flank-rubbing supported Guthrie but was later shown to be related to the presence of the experimenter instead.

7.2 The effect of reward as Guthrie saw it

Grooming becomes associated with start area because rat did it last.

Grooming is associated with start area that is associated with reward

Grooming is not associated with start area, so rat goes to goal area instead of grooming.

This is grooming

Tolman’s Operational BehaviorismHis theories relied on “intervening variables”not mechanistic S-R associations.

Behavior is motivated by goals.Behavior is flexible, a means to an end.

Rats in mazes form cognitive maps of their environment.

Animals learn about stimuli, not just behavior.

Evidence of Cognitive MapsChanging the maze layout resulted in running toward the same “goal.”

A light could have been used as a cue in both situations.

Using a “plus maze,” some rats were trained to always turn a certain direction, while others were trained to reach a consistent place.

The consistent place was easier to learn.

7.3 Bird's eye view of the apparatus used by Tolman, Ritchie, and Kalisch

7.4 Maze with two start and two goal locations for place and response learning

Latent LearningRats were given experience in a complex maze, without reward.

Later they were rewarded for finding the goal box.Performance (number of errors) improved greatly with reward, even among previously unrewarded rats.

Reward motivates performance, not learning.

7.5 Latent learning experiment: The 14-unit T-maze (Part 1)

7.5 Latent learning experiment: The results (Part 2)

Skinner’s ContributionSkinner was uninterested in theory – he wanted to see how learning works in practice.Operant chambers permit behaviors to be repeated as often as desired – more voluntary.Superstitious behavior – animals were rewarded at intervals without regard to their behaviors.

Animals related whatever they were doing to the reward, and wound up doing odd things.