2nd diagram on p. 43

The 2nd diagram on p. 43 bothers me because it comes too close to violating the Dead Man Test for my comfort when it says, in the Behavior box, Sue stops presenting the task. This seems more like a non-behavior than a behavior, & therefore something that a dead man could do. Instead of Sue's behavior being controlled by a reinforcement contingency, I think it's being controlled by a punishment contingency in which the Before condition is that she doesn't see Jimmy's discomfort, the Behavior is that she presents the task, & the After condition is that she sees Jimmy's discomfort.

Kinds of reinforcers, Part 3

Revised on 12/22/14

I suggest reading this post after you read the two posts called Kinds of reinforcers, Part 1 and Kinds of reinforcers, Part 2.

On p. 3 Malott tells us that a reinforcer is a stimulus that increases the frequency of a response it follows. Then on p. 4 he tells us that in order for a reinforcer to actually reinforce a particular response that it follows, the reinforcer must be delivered quickly. He says that one second or less is considered immediate reinforcement and up to 60 seconds is considered delayed reinforcement. So here are two more kinds of reinforcers – those that are delivered immediately following a behavior and those that are delayed.

So what about reinforcers that are neither immediate nor delayed? Can’t a reinforcer be delivered more than 60 seconds after a behavior occurs and still reinforce that behavior? Don’t we see this all the time? Suppose I say “Do me this favor and I’ll give you five bucks tomorrow.” Doesn’t the $5 reinforce the behavior of your doing the favor? No. Not according to what Malott says on p. 4.

Does that mean that the $5 is not a reinforcer? No, it doesn’t mean that. Go back to the definition of Reinforcer on p. 3. It’s a stimulus that increases the frequency of a response it follows. There’s nothing in that definition about following the response quickly. If the target behavior (that is, the behavior we’re focused on right now) is your doing me a favor, then if you do it and tomorrow I give you $5, AND (this is very important) you do favors for me more often in the future when I ask you to, then the $5 functions as a reinforcer.

Getting confused? Then read this very carefully or else you’ll get even more confused. A stimulus is a reinforcer if a behavior increases in frequency as a result of being followed by that stimulus, whether the stimulus followed the behavior quickly or not. But if the stimulus DOES follow the behavior quickly (preferably within 1 second, but at most within 60 seconds), then it DIRECTLY reinforces that particular behavior. If the stimulus follows the behavior by more than 60 seconds, and yet the frequency of the behavior still increases, then that stimulus is still a reinforcer, according to the definition. But it doesn’t DIRECTLY reinforce that particular behavior. Instead it INDIRECTLY reinforces the behavior. We’ll learn more about direct and indirect reinforcement in later chapters (like in Ch. 22).

This leads us to another very important principle, which is that behaviors increase in frequency ONLY when they’re quickly followed by delivery of a reinforcer. So if that’s true, you’re thinking, then how come a behavior that’s followed by a reinforcer more than 60 seconds later (maybe even as much as 15 minutes, a day, 2 months, ...) will still increase in frequency? It’s because in between that behavior and the delivery of that reinforcer, there must have been some other stimulus that quickly and directly reinforced the behavior. That’s the answer to the mystery. But it presents us with another mystery, which is “What is that other stimulus that quickly and directly reinforces the behavior?” Stay tuned.

Kinds of reinforcers, Part 1

On p. 3 Malott provides the definition of Reinforcer. Notice Positive Reinforcer in parentheses. This suggests that there are other kinds of reinforcers, which is correct. Later you’ll learn about negative reinforcers. What you need to know is that most of the time when reinforcer is used by itself, without positive or negative in front of it, that’s shorthand for positive reinforcer.

When the sight of the thing reinforces looking at it

Revised on 1/4/14

Malott's 2nd question on p. 6 (middle of 1st column) asks what reinforces the behavior of giving attention to Eric as he throws a tantrum. Watching something - Eric throwing a tantrum, a beautiful sunset, a good-looking person - is a behavior. And all behaviors require explanation.

Malott says that sometimes the sight of the thing we're looking at reinforces the behavior of looking at it. This is an example of a stimulus functioning as a reinforcer (remember the 4 kinds of reinforcers discussed on p. 3). The sight of something, the smell of something, the taste of something - these are all stimuli, & in the right circumstances, any of them might function as a reinforcer for a behavior that they immediately follow.

Anyway, Eric is throwing a tantrum & you watch him as he howls & flails about. Your watching is a behavior. Why do you do it? Because it's reinforced by the sight of the thing you're looking at.

Reinforcer is as reinforcer does

Revised on 1/4/14

Malott points out several places in Ch. 1 that whether or not something functions as a reinforcer in a particular instance depends on whether the frequency of the behavior it immediately follows increases when circumstances are similar in the future. That's kind of a tangled sentence, so you should re-read it as many times as you need to in order to understand what I just said.

One of the places he makes this point is on p. 8 (the "Reinforcer Assessment" section). We can't simply make the blanket statement that candy is a reinforcer, because for some people, if you give them candy immediately following a target behavior, the future frequency of that behavior doesn't increase. And even if candy often works as a reinforcer for someone, there may be times when it doesn't. So the only way to know for sure if something functions as a reinforcer for a given organism performing a given behavior in a given type of situation is to see if the frequency of that behavior increases when that organism is in a similar situation in the future.

Despite all this, it's certainly true that some things almost always function as reinforcers for most normal people in most normal situations. A good example is money. So as long as we realize the limitations of what we're saying, it's OK to refer to those things as reinforcers. Can you think of some examples?

What are the functions of the stimuli in a behavioral chain?

Revised on 3/29/14

The dual-functioning stimuli that link the responses in a behavioral chain function as reinforcers for the behaviors that precede them. The 1st stimulus or condition that starts the chain doesn't link two behaviors, so in analyzing a chain, we normally don't label that one as a reinforcer. The 2nd function of these stimuli may be S^D, EO, operandum, or opportunity to respond. This last option is not discussed in Ch. 20. It's briefly discussed in Ch. 22 (p. 354) but it's relevant here too. Without having it as an option for the 2nd function of these linking stimuli, we're handicapped in some cases. Anyway, figuring out which of these functions applies to a particular dual-functioning stimulus is one of the trickier things to do when analyzing a behavioral chain.

In this chapter, Malott's only discussion of a stimulus functioning as an EO is in the footnote at the bottom of p. 315, where he assigns that function to the stimulus that starts the behavioral chain he's describing. Can a dual-functioning stimulus in the middle of a chain also have the EO function? I think it can because the introduction of an EO increases the likelihood that the next behavior in the chain will occur, and that's the sort of thing these linking stimuli do to keep the behaviors happening one after the other. But that's not my main concern in this post.

My main concern is this opportunity to respond option. In Ch. 22 we're going to learn about analogs to discriminated avoidance. These are situations in which there's a deadline and the target behavior has to happen before the deadline or else you lose the chance to receive a reinforcer. We'll learn that in lots of these scenarios, the time before the deadline functions as an S^D, meaning that if the target behavior happens before the deadline, it's reinforced. And in those scenarios, the time after the deadline functions as an S^Δ, meaning that if the behavior happens after the deadline, it won't be reinforced. But, Malott explains to us, the time before a deadline doesn't always function as an S^D. It's not an S^D in cases where the target behavior, for various possible reasons, cannot be performed after the deadline has expired. If the target behavior can't be performed, this means the time after the deadline is not an S^Δ, which means that the time before the deadline is not an S^D. So if the time before the deadline is not an S^D, what is it? According to Malott, that stimulus condition is best labeled as an opportunity to respond (again, see p. 354).

Are you still with me? OK, then here we go, back to those dual-functioning stimuli in behavioral chains. When my students think up examples of behavioral chains, they sometimes have trouble figuring out the function of the stimulus/condition described in the 1st box as well as the other linking stimuli. I now realize that there are 4 possibilities: S^D, EO, operandum, or opportunity to respond. As an example of opportunity to respond, consider the example provided by one of my students. The 1st box says "Car is parked." The following behavior box says "Start car." The following dual-functioning stimulus box says "Car running." Is that first stimulus condition (Car is parked) an operandum? I don't think so. Is it an EO? I don't think so because if it was, that would mean that when the car is parked, the reinforcement value of "car is running" would be greater than if the car was not parked, that is, if the car was running or being driven. This doesn't make sense because in the latter condition (car is running or being driven), the behavior of starting the car can't be performed because the car is already started. Is it an S^D? I don't think so because the corresponding S^Δ would then be "car is running/being driven." And again, in that condition the behavior of starting the car couldn't be performed because the car is already running. So what's the function of the condition described in that first box, "car is parked?" I think it's an opportunity to respond.

So some quick guidelines for figuring out the functions of the initial stimulus/condition and the linking stimuli:

Is it an S^D? Figure out what its corresponding S^Δ would be. Then ask yourself two questions: (1) In the S^Δ condition, could the following behavior be performed? If not, then this condition is not an S^Δ, which means that the stimulus/condition in question is not an S^D. (2) In the S^Δ condition, if the following behavior could be performed, would it be followed by the reinforcer that's described in the next stimulus box? If so, then then this condition is not an S^Δ, which means that the stimulus/condition in question is not an S^D.

Is it an EO? Look at the reinforcing stimulus that results from the behavior. Is it a more powerful reinforcer for the behavior because of the presence of the stimulus/condition in question? If so, then it's an EO.

Is it an operandum? Does the stimulus/condition constitute making available some "thing" that you must have in order to perform the following behavior? An example is "fork in hand" when the following behavior (such as "put fork in potato") can't be performed without that "thing." If so, then it's an operandum.

Is it an opportunity to respond? Similar to operandum in that without either - operandum or opportunity to respond - the following behavior can't be performed. Here, tho, the inability to perform the behavior is not because some implement or object is not available, but for some other reason it's not possible to perform the behavior, like in the "car is parked" example above.

Have fun figuring out the dual functions of these linking stimuli in behavioral chains!

Choosing target behaviors for contingency analyses

Revised on 1/4/14

Take a look at an earlier post with the "Principles of Behavior: Ch. 02" label for tips about preparing the reinforcement CA. This post will elaborate on some of what was written there.

One of the most common errors that students make with CAs is in identifying the target behavior. Because the courses that use Malott's book focus on the basic behavioral contingencies, the examples of contingencies that you use in your CAs should use simple, discrete behaviors. "Discrete" means not consisting of multiple parts. So you should not use behaviors that are actually behavioral chains consisting of a sequence of connected behaviors. An example of a behavioral chain is baking a cake, which is not a discrete, unitary act, but several acts or behaviors performed one after the other until you're finished. You should not use any other "collection" of behaviors either. Doing your homework or cleaning up your room, for instance, are collections of separate behaviors, consisting of any number of individual behaviors. Often these collections of behaviors have other, unrelated, behaviors interspersed. So you might work for a while on your homework, or pick up a few things in your room, then take a break, get a bite to eat or go to the bathroom, then resume working on your homework or cleaning your room. Collections of behaviors like this don't work for the kinds of CAs we're doing. One of the reasons is that in this kind of situation, it's not possible to specify what is meant by "immediately following the target behavior."

The target behaviors you choose should have a clear starting point, a clear ending point, and between the starting and ending points there should be no interruption and no interspersed behaviors.

The one possible exception is the reinforceable response unit (RRU), which you can read about on p. 374. An RRU meets the criteria I just spelled out in the preceding paragraph, so it's acceptable. But if you specify an RRU as the target behavior in one of your CAs, be careful, because they're rare in real life, and your CAs need to be faithful to real life.