Learning II.doc

Learning II: Reinforcement Principles Partial Reinforcement Reinforcement Schedules Shaping Premack Principle Partial Reinforcement ?Humphrey?s paradox? behaviors that we reinforce only occasionally are much harder to extinguish than those we reinforce continuously Why ? continuous reinforcement = quicker learning partial reinforcement = harder to go extinct e.g. Abusive relationships (1) Fixed Ratio Fixed ratio means that reinforcement is provided every X items. e.g. - Rat gets 1 pellet for every 3 presses. - You get $5 for every signature you get. (2) Fixed interval. Fixed interval provides reinforcement for producing a response at least once after X amount of time has passed. e.g. - Pigeon gets a seed for the first peck after 1 min. - Salary work (3) Variable ratio Variable ratio provides reinforcement at some average rate, but the specific amount of response required during any given period varies randomly. e.g. - Pigeon on VR=10 might get a seed after 5, 17, 3, 15 pecks. - The lottery / slot machines (4) Variable interval Variable interval reinforces for producing a response after a random time interval (with some average across times). e.g. - Pigeon on VI = 5 gets seed on first trick after 1, 8, 4, 7 minutes - Asking your parents for money only works at random time periods (4, 2, 6, 12) Schedule-based Behaviors Schedules of Reinforcement Fixed Ratio ? Reward after certain number of responses Fixed Interval ? Reward after certain amount of time Variable Ratio ? Reward after certain number of responses on average Variable Interval ? Reward after average amount of time Shaping reinforce behaviors that aren?t exactly the target behavior but are progressively closer start by reinforcing anything close, then fading out reinforcement for not-exactly-right behaviors Premack Principle reinforcing a less-frequently performed behavior with a more-frequently performed one eat your veggies before you have dessert Superstition: Conditioning Gone Awry Skinner (1948) used a FIXED INTERVAL reward schedule on pigeons doing random things RECALL: this means their behaviors were rewarded randomly actions linked to reinforcement by sheer chance Superstition & Reinforcement ?lucky pen? & other objects ? recipes ? sporting traditions? Mawrer?s Two Process Theory When classical and operant conditioning work together Phobias: Phobias & Fetishes may develop (in part) due to classical conditioning may be maintained by operant conditioning Sight of pickles produces fear (unpleasant state) Running away reduces fear (removes unpleasant state) Running away is negatively reinforcing Can easily lead to more extreme ways to reduce fear that significantly hamper normal functioning Mawrer?s Two Process Theory CLASSICAL CONDITIONING leads to the development of a phobia or some other contingency OPERANT CONDITIONING explains the behavior that maintains this fear Other Applications of Operant Conditioning Token economies: provide tokens/chips to reinforce good behavior Tokens are secondary reinforcers that can be exchanged for primary reinforcers (naturally pleasurable things) Common in daycares / psych hospitals / jails Latent Learning learning that isn?t directly observable (or hasn?t yet been observed) relates to competence vs. performance knowing vs. showing rats in trolley cars can construct cognitive maps Insight Learning not based on trial and error but instead an ?aha? moment Observational Learning Learning by watching others Observational Learning Implications: Violence in the media Interactive violence in video games? Biological Influences on Learning Food Aversions Preparedness & Phobias can be learned very rapidly (only one exposure) flavors/foods tend to be very specific can have very long delays between CS and UCS ( 6+hrs) not all aversions are equal (no equipotentiality) ? some make more sense, seem to belong more rats more readily associate feeling nausea with food aversion Preparedness evolutionary predisposition to fear some things more than others Consider: Cars vs. snakes? Guns vs. lightning? Models of Learning moving away from simple Stimulus-Response (S-R) approach to allow a place for thoughts Stimulus-Organism-Response (S-O-R) accounts for the impact of the individual?s thoughts, interpretations and emotions cognitive conditioning: thoughts can change learning Two Takes on the Same Story S-O-R theorists ( your interpretation (based on your thought process) dictated your response Behaviorists ( your learning history (how you were trained to react) dictated your response Methodological behaviorists (Watson) ( focus on observable measures (behavior) Radical behaviorists (Skinner)( thinking, emotions are just different kinds of behaviors (unobservable) many (most) modern theorists now incorporate thoughts/intentions into their models