Lecture 7 - Operant Conditioning

Operant Conditioning: - learning of voluntary behavior though effects of pleasant/unpleasant consequences to responses Thorndike’s Law of Effect: actions followed by pleasurable consequences tend to be repeated (without classical conditioning) and if an unpleasurable consequence, tends not to be repeated ● Trapped cat with food bowl outside. Pulls a pin and gets rewarded with food. Next time, walks near place last time that had pin Skinner & Reinforcement: ● Operant: voluntary behavior not elicited by specific stimuli ● Reinforcer: anything that increases the probability of repeating a prior response ● Primary Reinforcer: directly fulfills a basic need (food) ● Secondary Reinforcer: becomes reinforcing after being paired with primary ○ Ex. Bills have no intrinsic value, but it matters as it’s associated to buying things that are of value (Gains correlation through association) ○ Gets its power from classical conditioning ● Positive Reinforcement: adds something positive to increase probability of prior behavior (ex. Teaching dog tricks by using treats as reward) ● Negative Reinforcement: removes something negative to increase probability of prior behavior (NOT punishment, ex. Medicine to treat sickness as it removes the negative symptoms so likely to buy again) ○ Ex. Finishing homework (removes stress) before sleeping each time ● Both Pos and Neg reinforcement results in an increase of the target behavior ● Whereas punishment weakens a response ● Shaping Behavior: reinforcing successive approximations of the desired behavior until the goal is reached ○ goal behavior in mind, is rewarded with each progression towards final goal ○ Ex. Want rats to touch lever to get food. Skinner rewards with food even if the rat walks one step towards lever until finally touches lever for food ○ Ex. Professor moves toward side of room to engage more until he only taught in one corner of the room based on reinforcement ● Schedule of Reinforcement: ○ Reinforcement is intermittent that is based on... ● ● ● ● ● ● ● ○ Time (intervals) of time in between rewards ○ Behaviors (ratio) number of goal occurrences between rewards Partial Reinforcement Effect: tendency for a response that is reinforced for only some, but correct responses are very resistant to existention. allowance for chores either every day or once a week. Every day is more reinforced after every frequent single response. But every week is the partial reinforcement Continuous Reinforcement: reinforcement of each and every correct response Fixed Interval: reinforcer always presented after a fixed time interval ○ Slow, unsteady response rate with time that are evenly spread out ○ Reward is not immediate with goal; it just must fit into time window so only has to do goal once to get reward in that fixed interval. Creates a dead spot and slow response as learns the interval and only must do it once ■ Ex. Studying a certain number of hours fixed to test interval Variable Interval: reinforcer is presented after a varying time interval based on some average time, time b/n reinforcement varies ○ No drops or pauses in response rate, slow and steady ○ Reward is given quicker if do goal more in average time interval ■ Ex. Having many quizzes and tests so study time differs and must study every day due to variable interval ■ Ex. Waiting for mail to come every week with variable time/schedule Fixed Ratio: reinforcer is always given after a fixed number of nonreinforced responses, 1:1 ratio ○ More responses require higher response rate, can have high outcome ■ Ex. Spamming the button as gets food with every push ■ Ex. Peacework industry, money made based on production #/quota Variable Ratio: reinforcer is given after a varying number of nonreinforced responses based on average ratio ○ Highest rate, most resistant to extinction even when reinforcement stopsf ■ Ex. Pushes button a lot without reward as hope to get reinforced ■ Ex. Slot machines, gacha, video games (hit 1/100 spins on average) ■ Ex. Kid asking for stuff and cries, then gets reward (reinforced) Token Economy: getting rewarded with miles for a trip with credit card purchases ● Applied Behavior Analysis: Observes poor child behavior for replacement behavior. Implements training and reinforces simplest levels by gradually removing reinforcers as the child demonstrates the behavior independently ● Instinctive Drift (Kelly & Marian Breland): animals studied learned skills through reinforcement but eventually revert to their genetically controlled pattern behavior ○ An animal does NOT come to lab with tabula rasa “blank slate” and can’t be taught any behavior ○ Different animal species matter to see what behaviors can be conditioned ○ Not all responses are equally able to be conditioned to any stimulus ● Behavior Modification: use of learning techniques to modify/change undesirable behavior and increase desirable behavior by use of operant conditioning ○ Ex. Target behavior = making eye contact with teacher. Reinforcer = gold star every time the student makes eye contact with the teacher. The child then gets a reward based on how many gold stars earned ○ Token Economy: use of objects (tokens) to reinforce behavior, where tokens can be accumulated and exchanged for desired items/privileges ● Applied Behavior Analysis (ABA): form of functional analysis and behavior modification that uses a variety of behavioral techniques to mold for a desired behavior/response ○ Ex. Skills are broken down into simple steps and taught by reinforcement until prompts are gradually withdrawn and can-do skill independently ● Neurofeedback: form of biofeedback using brain-scanning devices to provide feedback about brain activity to modify behavior ● Discriminative Stimulus: provides a cue for making a certain response to obtain reinforcement while discriminating other cues ○ Red stoplight is a cue for stopping as if not, there is negative reinforcement of a ticket or getting hit. ○ Ex. Doorknob to open the door with reward of open door Tolman’s Maze-Running Rats: Latent Learning: learning that remains hidden until its application becomes useful ● 1 rat group reinforced with food to complete the maze ● 1 rat group not reinforced with food when exited maze, but was given on 2 day ● 1 rat group (control) was not given any food reinforcement when exiting nd ○ Results showed that the 2 nd group was the quickest from making a st “mental map” / “cognitive map” of the physical layout during the 1 day ● Ex. Figuring out how to cook byself even when thought previously couldn’t Kohler’s Smart Chimp: Insight Learning: sudden perception of relationships among various parts of a problem, allowing the solution to the problem come quickly ● Sultan the Chimp had to get an out-of-reach banana using simple trial-and-error learning like reaching, lying, stick, (had an “aha” moment) ● When banana was pushed more out, Sultan used 2 sticks as a tool to get it