Transcript Convert
Reinforcement & Punishment: What is an SR? Lesson 8 What is an SR? Thorndike’s Law of Effect Satisfiers & annoyers Skinner determined by how B changes reinforcer: B punisher: B Primary reinforcers & punishers biologically important stimuli ~ What is an SR? (continued) Secondary reinforcers & punishers money praise How do they become an SR? Classical Conditioning Higher order learning ~ Drive Reduction View (50s & 60s) Similar to Law of Readiness Relative state of deprivation required for a basic drive thought to always be true Drive motivation B reduction of drive state (SR) ~ But... Sometimes hard to identify drive What drive is this? ~ Sensory reinforcement Sensory stimulus unrelated to biological drive monkeys learn response reward is watching toy train rats learn to bar press reward = turning on a light or turning off light ~ Premack Principle Commonly used in educational setting impractical or unethical to use food Thought of reinforcers as responses press bar eating response wider application of I/O conditioning Differential probability principle High probability responses reinforce low probability responses ~ Premack Principle Homme et al (1963) Unruly 3 year olds High probability behaviors ignored teacher screaming pushing furniture Low probability behavior sitting quietly ~ Premack Principle: Homme et al Rewarded sitting quietly with... 3 min of running around screaming Results: sitting quietly increased Particular behaviors observed by different kids different responses effective reinforcers for different kids ~ Premack Principle Charlop, Kurtz, & Casey (1990) autistic children High probability behaviors echolalia perseveration Low probability behaviors adding up coins judging objects: same or different ~ Premack Principle: Charlop et al 100 % correct responses 80 echolalia RFT 60 food RFT 40 # of sessions Premack Principle: Problems Fluctuation of response probabilities e.g., sometimes kid would rather play outside than play video games Solution: token economies Does not explain how reinforcer increases response probability ~ Behavioral Regulation Approach Response deprivation limit access to a response does not require high vs. low probability Behavioral homeostasis preferred distribution of activities operant conditioning imposes limits behavioral bliss point e.g., time spent studying vs. video games ~ Behavioral Regulation Approach A behavior is limited below bliss point disturbance of behavioral homeostasis analogous to increased biological drive Contingency set during I/O procedure establish relationship between responses B move toward bliss point (baseline) ~ Behavioral Regulation Approach Low probability behaviors as reinforcers observe baseline rate of behavior limit activity below baseline Require a response to engage in deprived behavior contingency Increase toward bliss point cost vs. benefits determines how much ~ What Becomes Connected? Skinner? refused to consider associations Thorndike: S-R view (SD-B) association b/n stimulus context and response R NOT the outcome (S ) no representation of reinforcer ~ S-R-O (SD-B-SR) view: Tinkelpaugh (1928) Goal-oriented responding respond with idea of getting reward The monkey and the hidden banana 2 cups, put banana under 1 task: choose cup with banana Secretly substituted rotten lettuce monkey became agitated Expected banana reward (outcome) ~ S-R vs. S-R-O Adams & Dickinson (1981) Taste aversion paradigm Associate sucrose (sweetner) w/ lithium chloride (LiCl) illness Will rats press bar to get something that makes them sick? ~ S-R vs. S-R-O Phase 1: Trained rats to bar press for sucrose Phase 2: associate sucrose w/ illness Phase 3: Will rats press bar now? No sucrose delivered ~ S-R vs. S-R-O : Results Predictions? If S-R-O If S-R Results Rats did not press bar Supports S-R-O ~ S-R vs. S-R-O Use different levels of training Phase 1: Same procedure but… some get 100 RFTs some get 500 RFTs ~ Results & Conclusions Less training low response rate Little training outcome important S-R-O Extensive training high response rate outcome less important response is well established S-R ~ Parallel learning in humans Learning a skill e.g., to drive a car Early trials consider consequences must think about what you are doing After extensive experience becomes automatic after many trials ~ Extrinsic Reward vs Intrinsic Motivation Early trials expectation of reinforcer extrinsic reward CER = positive affect Well-established behavior no expectation of reward intrinsic motivation CER = positive affect ~