 All right, everybody, welcome back. So we're gonna be talking a little bit about the principles that make reinforcement effective. And what we're talking about here is under what circumstances are stimuli most effective as reinforcers and as punishers. The reading that you may be assigned for this unit really focuses just on reinforcers, but bear in mind that these principles also apply to punishment as well. So before we begin, I want to introduce you perhaps one of Skinner's most influential quotes or at least one that's most relevant here to the content that we're gonna be discovering. Skinner said that the way that positive reinforcement is carried out is more important than the amount of positive reinforcement used. So what he's emphasizing here is that there really is an artful application to reinforcement that we have to be attending to. The way in which we use positive reinforcement is more important than the amount. It's not enough to just give big bonuses. It's not enough to just jackpot behavior and expect that that's gonna change behavior. If you want durable, effective behavior change, you have to be very carefully, you have to analyze closely and use reinforcement effectively. So how do you do that? Well, one stimulus you wanna pay attention to is this idea of deprivation. Remember that a reinforcer or a punisher is only effective if the learner has not had too much of that stimulus. Kind of the flip side of that is satiation. If you've had too much of something and you're satiated and you don't really want anymore. Well, so how do you do that? How do you know when an organism is sufficiently deprived? This one is pretty difficult because you only know if reinforcement is effective if the learner will continue to work for the stimulus that you're offering them. This is one where you have to gauge based on experience whether or not the stimulus is losing its effectiveness. One way of kind of determining this is determine whether or not the learner even wants to continue earning a stimulus. If you start to see that you've delivered a lot of the stimulus, a lot of the reinforcer and then it's no longer effective. One of the first places to look is to see whether or not perhaps the learner has become satiated or full of the reinforcer. If they're no longer deprived, it won't be effective. Another concern to attend to when you're using reinforcement is how quickly are you using that reinforcer? And Miller refers to this as the principle of immediacy. Reinforcement tends to be most effective when the reinforcer is delivered immediately contingent on the response. However, we know that humans are pretty good at cognitively appraising what's going on or establishing these kinds of rules. So we know that if you deliver the stimulus at any point within the minute of response or if you communicate somehow that there's gonna be a delayed reinforcer perhaps by using like a clicker or some sort of token that we can mediate a very long delay. So remember that we talked last time about token economies, we talked a little bit about clicker training and those are some ways that we can help mediate that delay. Let's look at some of the basic research on immediacy. This is a really wonderful study by Raymuth and Oryth in 1971. It was evaluating what happens when social reinforcers like touch or telling a baby, oh who's a good boy, is effective for infants. And they found that they're really only effective if they were delivered contingent after a vocalization. So let me describe this for you. You've got a researcher and you've got some young infants and whenever the baby would make kind of a noise, like baby noises, the researchers would deliver some sort of touch or giggles or cuddles or tell the baby, oh who's a good baby, there'd be a response to the infant. And we look to see what happens contingent on that response. So in baseline, the researchers were essentially giving kind of no response whatsoever whenever the baby would make noise. And then they, in the next conditions, would look to see what happens when we provide those consequences, when we touch or cuddle or kiss or whatever the baby. And then we do this, but for some conditions, for the white data that I'm about to show you, there's no delay in the response. We see this dramatic increase in the number of vocalizations. But for the other two data paths, the data path here shown in yellow, what we see is that when there's a three second delay, the rate of vocalizations does not increase. So a three second delay would, for instance, be a baby going, one, two, three. Surgical baby, right? Or in the six second delay, the baby would make a noise like, ah! And then one, two, three, four, five, six. Surgical baby! Under those conditions, there was absolutely no change in the rate of vocalizations. But in that no delay condition, when the baby would make some sort of vocalization, the researcher would respond immediately. And we see that under those conditions, the rate of vocalizations, the rate of the baby speaking or making noises dramatically increased. So we see under these conditions, we actually see that time matters and being there, being present, responding immediately, as associated with effective reinforcement. But remember that we're very, very skeptical about data. So under these conditions, what we're looking at right now is a kind of comparison design. We cannot make any conclusions about what happens. Raimi and Oryth went back to the baseline condition to see what would happen if the researchers no longer responded to the vocalizations. And they demonstrated that there was a dramatic decrease in the rate of the response, which shows us here that, yes, it was the researcher's response that was increasing the rate of infant vocalizations. And when that response is omitted, when it's withheld, we see a decrease in the response. So this is a very effective reversal design. Another stimulus we want to attend to or another feature we wanna be careful of is this idea of size. So the amount of reinforcement. And this one's kind of difficult because it's a double-edged sword here. When we talk about size, we want the reinforcer to be of sufficient magnitude. We want it to be big enough to be meaningful. However, if we choose too large of a reinforcer, we also have to go back to deprivation. We have to worry about satiating the learner. So when we talk about size, remember magnitude is sometimes used. Magnitude means the size of the reinforcer. And most of the time we tend to prefer, we tend to choose large reinforcers or we're small reinforcers. For instance, if I were to offer you the choice between a $50 bill and a $5 bill, I feel pretty good in guessing that most folks would choose the $50 bill. Or if you were to be offered the choice between, say, a tiny candy bar or a normal-sized candy bar or just a big old honking candy bar, some may choose the biggest candy bar possible. What's interesting is actually when I present this to people, I have a fair number of folks who prefer the bite-sized snickers because the ratio or combination of sort of features there, the ratio of peanuts to nougat to whatever. Some people actually prefer the small one and this one, it's a little bit eclectic, but kind of fun. And like I mentioned before, you do wanna bear in mind that size and the issue of size of reinforcement is kind of a double-edged sword. So reinforcers are only effective if the amount is worthwhile, right? Like if the juice is worth the squeeze, but you gotta be careful because while you have to make the amount of reinforcement worthwhile, you also have to worry about the effect that it will have on deprivation. So this is kind of a teeter-totter where you have to make it large enough to be effective, but not so large that the client becomes super satiated very, very quickly. If you give too little of the reinforcer, it's not going to be effective, but if you get too much of the reinforcer, then suddenly the client is no longer deprived, so it will also be ineffective. And again, you determine whether your reinforcer is effective through practice and through careful analysis. So let me show you one basic study, one really amazing study done by Wolfen colleagues that shows us what happens when we alter the size of the reinforcer. Wolfen colleagues, what I'm showing here, let me take a sec to orient you to the graph. Wolfen colleagues were looking at a dependent variable of the number of sessions that kids were reading in a class and, or the number of stories that they would read in this class. And they were evaluating that across class sessions. And in the baseline condition, they were giving 90 points. These points could be exchanged for rewards and treats from a kind of backup menu. So again, a kind of token economy. Kids could earn 90 points per story. Then they switched to the next condition. They dropped the number of points that kids could earn to 52. And we see that there's a dramatic decrease in the number of stories that the students read. When they went back up to 90 stories, we see an increase again in the number of stories being read, but it's a little bit below that initial baseline. And then they dropped back down to 52 points per story. And we see again a decrease in the number of stories read. And when returning to a condition where kids could earn 90 points per story, we see again an increase in the number of stories read. This shows us that size matters. And we can also apply the same to ourselves. How do you feel if you walked into work and your boss suddenly said, you're gonna earn half as much as you did before? Well, I mean, some people would just be thankful for the opportunity to work. Some wouldn't mind because maybe that's not their primary reinforcement from work, but I think the vast majority of us would be very upset to find out that suddenly we're gonna be paid half as much. Some of us would work half as much. Some of us would just tender our resignation immediately. The size of the reinforcer matters. We also see this is true when we're using interventions, clinical interventions like contingency management. Contingency management is a kind of treatment that's used for folks who are trying to stop using drugs or alcohol. And the way that it works is you provide an amount of reinforcement contingent on clean. It's kind of a differential reinforcement of other behavior, which means you earn the reinforcer for doing literally anything that is not drugs. Silverman and colleagues do a lot of this contingency management research. And typically we've shown that contingency management can be incredibly effective for folks who are not successful in other forms of treatment. Narcotics Anonymous working with their doctor, getting support, even in patient rehab, folks who fail out of all of those other forms of treatment, they tend to be very well supported by contingency management. And Silverman and colleagues, again, going back to the idea of incentive size or reward size, they looked at what happens when we provide larger reinforcers for abstinence from drugs and alcohol. And they found that for cocaine dependent individuals, when the reinforcer was $54 per day, most of them could maintain six or more weeks of cocaine abstinence. And $54 per day, it sounds like a lot. And it feels a little bit counterintuitive, like why would you pay a person to be abstinent? But bear in mind that inpatient treatment, inpatient psychiatric treatment or treatment facility, rehab treatment can be thousands of dollars. So this is actually quite a big cost savings. Last, let's talk about contingency. This is often the most difficult to understand because when we talk about contingency, remember, generally we mean a kind of if then statement. We mean that if a response occurs, then something will happen. When we're talking about contingency as a principle of effective reinforcement, what we mean here is that we have to be very, very careful that the learner earns the reinforcer only for this response, only for the target response we want to occur. For contingency, the reinforcer has to be delivered only for the target behavior. So it does go back to our definition of if then. We have to make sure that the reinforcer is delivered only if they engage in the target response. And this idea of breaking or violating effective reinforcement in terms of contingency, it would be sort of like giving the reinforcer for everything or giving it for nothing at all. If a reinforcer, if an established reinforcer is given for free, that reinforcer is not going to increase the probability of a specific target behavior because you can kind of do anything to get it. For example, if a vending machine were just to deliver a product, just to drop a product every 25 seconds and you didn't have to do anything, you didn't have to put in any money or press any buttons, would you actually pay for your stacks? Would you put money in it or would you just wait 25 seconds for it to be delivered? So contingency means you have to make sure that you're giving the reinforcer only for the target response, not giving it for other things, not giving it for free. A basic study that talked about this was Brigham and colleagues. So they were looking at the number of correct handwriting responses. They were trying to teach a person how to write effectively in these different sequences of conditions involved with the token economy. In the baseline condition, there were no tokens, no tokens could be earned. In the next condition, the learner was just given tokens that they could exchange for backup reinforcers for free. And then in the third condition, the researcher said, okay, we still got the tokens, but now you have to earn the tokens by improving your handwriting, right? By giving us good handwriting samples. And we see that the amount of correct handwriting increases when the tokens are made contingent on correct handwriting. In the next condition, they went back to free delivery of the tokens and we see the quality of handwriting again decreases. And then in the subsequent condition, when the tokens are made contingent on good handwriting, we see again an increase in good handwriting. So this demonstrates, this shows us that you have to make sure for a reinforcer to be effective that it's given contingent on a response and you're not just giving it for free because it's going to decrease the rate of the response upon which the reinforcer is delivered. So how do you keep these straight? How do you remember all of these different little principles? The best way I found to remember Miller's four principles of effective reinforcement is with the mnemonic disk. Deprivation, immediacy, size, contingency. Deprivation is make sure that they haven't had too much of the reinforcer, immediacy is, deliver that reinforcer immediately as quickly as you can within a minute is best but the faster, the better. Size is it has to be large enough to be worthwhile but not so large that we're violating deprivation or making the client satiate on the reinforcer and contingency. Make sure that you're delivering the reinforcer only for the target response. All right, thanks so much. I'll see you guys next time.