Schedules of Reinforcement There are several alternate ways to arrange the delivery of reinforcement A. Continuous reinforcement (CRF), in which every appropriate response is reinforced B. Intermittent schedules of reinforcement, in which reinforcement is unpredictable. There are two sub-categories of intermittent reinforcement schedules. 1. Ratio Schedules a. Fixed Ratio (FR) b. Variable Ratio (VR) 2. Interval Schedules a. Fixed Interval (FI) b. Variable Interval (VI) Cumulative Frequency Graph • Used to illustrate the effects of different schedules of reinforcement • Each response is added to the previous one(graphed cumulatively) • The steeper the slope of the data path, the more rapid is the rate of responding R e s p o n s e r a t e Time Continuous Reinforcement (CRF) • Use during acquisition stage of learning • Reinforce each occurrence of desired behavior • Low resistance to extinction (behavior doesn’t persist when reinforcement is withheld • Risk of reinforcer satiation, esp. with primary reinforcement Intermittent Reinforcement • Reinforce some, but not all, desired responses • Use to maintain established behavior • Resistant to extinction • Satiation less likely Intermittent Schedules Ratio • Reinforcement based on number of responses √ Fixed (FR) FR 10--every 10th correct response is reinforced √ Variable (VR) VR 10--on average, every 10th correct response is reinforced Interval • Reinforcement based on passage of time √ Fixed (FI) FI 2min.--1st correct response after passage of 2 min. reinforced √ Variable (VI) VI 2min.--1st correct response after an average of 2 min. has passed is reinforced FIXED RATIO SCHEDULES Under fixed ratio schedules, the reinforcement is contingent on a set (fixed) number of responses. (FR-2, FR-4, FR-17) FR schedules result in a high rate of responding, with pauses after reinforcement. Increasing the ratio results in longer pauses. FIXED RATIO (FR) •Post-Reinforcement Pauses •Low to Moderate Response Rate •Poor Maintenance Under Extinction Considerations • Behavior must occur fixed number of times for reinforcement • Larger ratio = ratio strain. If too large, student will cease to engage in behavior • Therefore, should not thin reinforcement abruptly from a dense schedule (e.g., CRF) to a very lean FR schedule. Instead, proceed in smaller steps (FR 2, FR 4, etc.) • Example: assembly line labor Example: FR20 SR+ 20 words read SR+ every 20 words= 2 min on computer STO - Given 20 basic first grade sight words on note cards, Jason will orally read the words within 1 minute with fewer than 5 errors. VARIABLE RATIO SCHEDULES On a VR schedule (VR-3, VR-10, VR-200) reinforcement may come at any time. On the average, however reinforcement comes after so many responses. VR reinforcement results in high steady response rates. Extinction usually results in a high number of responses in a short time. Responses come as rapid bursts of behavior, followed by increasing pauses and then abrupt cessation of responding. VARIABLE RATIO No Post-Reinforcement Pauses High Rates of Responding Good Resistance to Extinction Average number of times behavior occurs Considerations • Reinforcement can’t be predicted • The more responses one makes, the more likely that reinforcement will occur • Therefore, supports high rates of responding with no postreinforcement pauses • Because reinforcement can occur at any time, extinction isn’t likely • Thinning from CRF to a lean VR schedule is easily accomplished • Example: slot machine Example: VR6 9 hand raises SR+ 2 hand raises 8 hand raises 5 hand raises SR+ on average of every 6th hand raise = puzzle with a peer STO - Given a teacher prompt to raise his hand during math seat work periods, Eric will raise his hand on 12 consecutive opportunities across 5 math classes. FIXED INTERVAL SCHEDULES Reinforcement under a fixed interval schedule is contingent on a correct response that occurs after the passage of time (FI-20 sec., FI-2 min., FI-1 hr.) On a FI schedule, the first correct response after a fixed length of time is reinforced. This results in scalloping-pauses after reinforcement with increased responding at the end of the interval. FIXED INTERVAL (FI) Post Reinforcement Pauses (scallops) Low to Moderate Response Rate Poor Resistance to Extinction SR+ end of interval if behavior occurred or ASAP after end of interval Considerations • Reinforcement at end of interval if behavior occurred in that interval, or as soon as the behavior occurs after the interval has expired • Extinction occurs if schedule is thinned too quickly (post-reinforcement interval is too long) • Example: waiting for bus FI1 Min. SR+ SR+ every 1 min= 1 min on computer STO - Given the instruction to remain in his seat, Jason will be in his seat each time the timer rings* on 20 of 30 opportunities for 5 consecutive sessions. *set for 1 min intervals VARIABLE INTERVAL SCHEDULES VI reinforcement results in sustained responding at a low rate. The reinforcement is contingent on the first correct response that occurs after an interval averaging a certain length of time (VI-20 sec., VI-3 min., VI-2 hr.) Extinction after interval schedules of reinforcement results in low, sustained responding that gradually tapers off. Extinction on interval schedules takes longer than it does on other schedules. VARIABLE INTERVAL (VI) No Post-Reinforcement Pauses Low to Moderate Response Rate Best Resistance to Extinction Considerations • Relatively low but stable response rate • No post-reinforcement pauses, because intervals between reinforcement are not predictable • Most resistance to extinction of any schedule • Example: timer game to promote in-seat behavior Example: VI5 SR+ SR+ on average every 5 minutes STO - During 35 minute reading activities without prompts, Jim will be on-task for 30 consecutive minutes 4 classes in a row. Post-test Which schedules: • Result in high rates of behavior? • Result in low rates of behavior? • Result in pauses in responding? • Result in steady responding? • Are best for acquisition of behavior? • Are best for maintenance of behavior? • Lead to most rapid extinction? Answers • • • • Ratio schedules result in high rates of responding Interval schedules result in low rates of responding Fixed schedules result in pauses in responding Variable schedules result in steady responding FR VR FI VI • A CRF schedule is best to use during the acquisition of a behavior • An intermittent schedule is best to use to maintain an established behavior • Extinction is most rapid on a CRF schedule