Behavior 101 #1: The four quadrants of reinforcement and punishment

Behavior controls all that we do, yet I find that most people don’t have a solid understanding of exactly how it governs our lives, or the lives of the animals which share our world. Knowledge of the laws of behavior can help you manipulate the environment in such a way as to elicit the behaviors you want to see and make them maintain. It doesn’t matter if your ‘subject’ is a dog, a person, an elephant, a dolphin, or a pigeon, behavior is behavior, and the laws of behavior apply to all. I trained dogs long before I went to school for behavior, so I always find it easier to think of behavior in terms of dogs first, but that’s more difficult for some people, and they need human examples first. Because of this, I’ve tried to include both dog and human examples, so you can visualize whichever is easier for you to understand.

All behavior is controlled through the environment, even our own. What causes a behavior to maintain, increase, decrease, disappear, or change, all depends on what happens immediately after that behavior occurs. We call this the consequence. I’m sure we’re familiar with this idea from childhood. When Mom would scream at us, “If you continue to do that you will suffer the consequences!!” (Right? It couldn’t have just been my mom.) Behavior occurs for several different reasons as well, but before getting ahead of ourselves, let’s learn, or review, the basic terms used in the science of applied behavior and see how these fit into our life.

Most people who have dabbled in any sort of dog training are aware of the basic principles of reinforcement and punishment, although not everyone gets the definitions correct. Positive/Negative Reinforcement and Positive/Negative Punishment are the most commonly tossed around terms I hear in dog training communities, so we’ll start there.

Homer Simpson knows the four quadrants!

Reinforcement is a stimulus change, immediately after the behavior, which causes the future rate of that behavior to increase. Punishment is the opposite- that’s when the stimulus change that happens after the behavior causes the behavior to decrease. The positive and negative on the front of that word just means that you are either adding or removing the stimuli from the situation.

So when you hear ‘positive reinforcement’ that means you added a stimulus (typically referred to as a reinforcer) and the rate of the behavior went up. How do we know that it went up? Well, ALL behavior analysts take data on the rate of behaviors they’re training or trying to change. Without data, it’s not applied behavior analysis. But, of course, not all trainers are behavior analysts, so they usually go with- does the dog sit more frequently when asked? Does it seem to be learning what I’m teaching? Is the dog becoming more reliable or responding quicker or seem to understand what you’re asking? Anecdotal observation probably says yes. When teaching a new puppy to sit, you probably give food or toy rewards immediately following the dog’s completion of the behavior. Through this the dog learns that sitting when they hear the “Sit” means they will be rewarded. The “sit” becomes a discriminative stimulus (an Sd), but we’ll get into that later.

What about negative reinforcement? Almost sounds contradictory, doesn’t it? Well, negative means you’re removing a stimulus (or preventing one), and reinforcement means that behavior is going to increase. Suppose your child doesn’t like broccoli. You set a plate down in front of them and it has broccoli on it. They see the broccoli and scream and cry. The parent removes the broccoli from the plate to stop the crying. The child’s screaming has been negatively reinforced- the screaming caused the removal of the stimulus, and this means that in the future, the odds that the kid will scream when presented with something they don’t like will increase. (Removing the broccoli is also negatively reinforced for the parent. They remove the broccoli and the god-awful wailing from their child stops. In the future, they’ll probably get rid of that broccoli faster, the get their kid to hush, or prevent the behavior altogether, by not placing broccoli on the plate). Preventing a consequence can also be negatively reinforced. If you burn yourself on a hot pan while getting it out of the oven, you’ll probably remember to put on an oven mitt the next time you go to grab a hot pan out. By preventing the burn, the rate of your oven-mitt-wearing behavior will most likely increase. Even though you’re not directly experiencing the painful stimulus every time, you’re still removing (negative) that painful sensation preemptively, by putting on that oven mitt.

Biting dogs are usually inadvertently negatively reinforced. Something might be causing them pain, such as a person roughly handling them. The dog bites, the person will generally stop whatever they were doing, and the pain to the dog stops. This is also negative reinforcement. Don’t be fooled by the reinforcement part of the word. I hear a lot of trainers say they ‘only train with reinforcement.’ Well, shock collars are quite often used as negative reinforcement, and I’m sure most people saying they are reinforcement trainers don’t mean it that way! If the dog if being taught to force-retrieve, often a shock collar is placed on the dog. The shock is triggered, and a dumbbell is forced into the dog’s mouth. As soon as the dumbbell is in the mouth, the shock stops. The dog learns that to remove (negative) the pain, he must pick up the dumbbell, and the rate of picking up the dumbbell goes up (reinforcement).

So if that’s reinforcement, then how does punishment come into play? Positive punishment is the style of punishment and correction that is most hotly debated in dog training forums. Again- positive, we’re adding something to the equation. And punishment means the rate of that behavior is going to go down. Say we’re walking down the street and our dog is pulling. We pop the dog hard with a leash and give what is commonly referred to as a leash and collar correction. This is positive punishment. Adding a chain collar or pinch collar to the mix doesn’t change anything other than the intensity to the dog. We’ll talk about intensity later. Shock collars are, of course, also used as a form of positive punishment. A dog barks and receives a shock, and his rate of barking goes down. This is positive punishment.

By now you may be able to guess what the fourth quadrant, negative punishment, would look like. Once again, it’s negative, so we’re taking (or preventing) something from the organism and the rate of behavior will go down. Say you have a dog with a terrible jumping problem. If you are petting your dog, and he jumps up, and then you withdraw the attention and walk away, and the dog learns that jumping up ceases the flow of attention, you are using negative punishment. You’re removing the attention, and the rate of jumping up goes down. Ever get grounded as a teenager (or ground your own children?). This is also negative punishment. You’re removing privileges and the teenager is question will stop what ever caused them to get grounded, or at least, be less likely to do that in the future. Say an off-color joke at work and get suspended without pay (or even fired)? Negative punishment.

Now some of you parents may have learned the hard way: “But wait! I tried this on my own kids, and it didn’t work!!” If the behaviors do not go up or down depending on what type of reinforcement or punishment that you’re using, then you’re simply not punishing or reinforcing that behavior. And here we reach the crux of a problem that it is difficult for people to understand or sometimes they never think of this to begin with. If the behaviors are not going up or down, then you’re not using a reinforcer or a punisher. In the case of the grounded teenager- if the rate of the behavior doesn’t go down (and I don’t mean cease completely in one application, behavior very often doesn’t work that way, unless the reinforcer or punisher is extremely powerful) then whatever your using is not a reinforcer, or the one maintaining the behavior is stronger.

My primary line of work is with children and adults with autism, downs syndrome, prader-willi, fragile-x, mental retardation, and other severe intellectual disabilities who exhibit some of the most extreme behavior, and many of these individuals exhibit extreme aggression, or self-injurious behavior (SIB). I’ve been bitten more severely by a 7 year old boy than I ever have by a dog. And on more then one occasion. (And that’s saying something because I had my top lip nearly bit off by a dog once. On accident).  In many cases I’ve seen individuals with SIB that bite themselves so hard they draw blood, and they do it repeatedly. Or hit themselves in the chin so hard they fracture their jaw. I worked with one boy who would slam his fingers in the kitchen cabinet drawers and jump up into the air before throwing himself down onto the tile floor on his knees, causing his kneecaps to have hairline fractures in them. Wouldn’t they be positively punishing themselves and then automatically stop the behavior? There’s the infliction of pain, which could be a punisher, but the rate of behavior doesn’t go down? Why not? Well, the answer is actually very simple, and one most people don’t think about. To these individuals, pain is not a punisher. Or whatever they are receiving after engaging in these behaviors is a more powerful reinforcement.

The number one rule of using these four quadrants applied to behavior you are working with is that just because it’s a reinforcer or punisher to you, doesn’t mean it is to the individual you are working with!! In the examples above, just because the behavior elicits pain in you, doesn’t mean it will elicit pain in the individual, or in many cases, your pain tolerance may be vastly different from the pain tolerance of someone with special needs. In many of the cases I mentioned above, the behavior functioned for access to desirable items. We’ll talk about function in a later installment, but for a quick down and dirty lesson in function I’ll say this. All behavior serves a function. Finding out that function is key to altering the behavior. In a majority of the individuals I mentioned, when they would engage in these severe self-injurious behaviors, their caretakers would often run around, even turn their homes upside down, trying to find out what the individual wanted. One older woman with a severe intellectual disability I worked with in a group home, would bite herself until she bled, and would continue until someone brought her McDonald’s French fries. The boy who slammed his fingers in the cabinets? When he did this, his parents would run around the house presenting things to him until they figured out what he wanted. They were inadvertently reinforcing the slamming behavior by giving him reinforcers when he did this behavior.

Commonly I see these applications used incorrectly with training dogs. Many people assume food treats will be a great reinforcer for training dogs. And usually, yes, it is. But you can never assume that because something is a reinforcer for one dog, or even a majority of dogs, that it will be a reinforcer for the dog you are presently working with. My female German Shepherd, Tiki, is a great example of this. She likes food…ish. She’ll eat her dinner, albeit it slowly (compared to the other three who are scarfers at dinner time). She could care less about food treats. She enjoys them, when I hand her a treat, she’ll often take it from me (after first sniffing it suspiciously) then take it somewhere in the house, where she’ll put it down, lick it a few times, and then it’s hit or miss whether she’ll eat it, or leave it for the other 3 to find later. Food is just not a powerful reinforcer for her. Training her with food wouldn’t get me anywhere quickly. What is a reinforcer for her? Praise. She loves to be praised in that roughed up way that involves vigorous rubbing, high-pitched voices, and butt scratches. She’ll do anything for it. I never need to carry food with me when we’re training. But my male German Shepherd? Forget it. He could care less about your silly, puny praise. You’d better have some delicious hot dogs or meatballs, or you’re not getting any decent progress with him. The yellow Labrador in our house would prefer to have a neon sign above her head that states, “Will Work For Fetch.” Our little guide dog puppy? She’ll work for plain old kibble. Doesn’t matter that she had a bowlful that morning, or will get another bowlful that night. If I offered kibble to my male shepherd in exchange for completing a behavior, there’s a very strong chance he wouldn’t do it again the next time. He’s actually taken the kibble and spat it out at my feet before. Definitely not a good reinforcer for him.

Definitely a punisher for me, especially if I found this say… in my bed? I’d have to have a long look at my life choices to ensure this wouldn’t happen again. For your dog though… I’m going to go out on a limb and guess it’s probably a reinforcer for him or her. It’s all a matter of perspective!

In people, and in dogs too, reinforcers can and will change daily or even by the minute. Know what is actually reinforcing to the individual you are working with, not what you think is reinforcing or has been effective as a reinforcer with others. It’s usually most helpful to have access to a variety of high-quality reinforcers. The more you present one reinforcer, the closer you get to satiation. This is where a reinforcer looses its value due to repeated presentation. In essence, the person or dog you’re working with gets tired of it. The best way to combat that is to deny them access to it for a time. This will result in deprivation; the reinforcer is more powerful because they haven’t had access to it in awhile. Many people do this without realizing it. Does your dog have a favorite toy that they only get to play with while you are training? By not allowing them to have constant access to it, you’re creating deprivation between training sessions. Sometimes deprivation can turn mediocre reinforcers into slightly more powerful ones. My male German Shepherd that I said wouldn’t work for kibble? If it’s late in the evening and he hasn’t eaten dinner yet, I can often motivate him to track for a bowl of dinner at the end of the track. He hasn’t had kibble in a while, and his tummy is most likely getting rumbly. An entire bowl of food is also quite a bit more than just a few pieces of kibble, and the quantity of the kibble increases its motivating factor, also known as MO. We’ll talk about that later, too. Seasoned trainers know this is considered a ‘jackpot’ reward, even if it is not a very highly motivating reinforcer. I usually place a few meatballs or a hot dog or two in there as well, just to ensure that the reward has been worth the effort to him.

So where does these styles of behavior control fit in? As a positive trainer, of course, I urge people to use positive techniques and never to resort to positive punishment. When working with people, ethically, you have to start with positive reinforcement techniques. And in most cases, this is more then enough to alter the behaviors. But there are times when it’s not enough, or the behavior is too severe, or too dangerous. There are times when we must resort to positive punishment to alter dangerous behavior in individuals with intellectual disabilities. As a master’s level, board certified behavior analyst, I am one of the few allowed to use positive punishment in practice with people, but only after everything else has been exhaustively tried and met with no success. I definitely do not take that responsibility lightly, and only do so in the most extreme cases. My own plans must go to be reviewed by an ethics committee, be peer-reviewed by other behavior analysts, and then often a Ph.D.-level behavior analyst also looks them over. Positive punishment is a serious thing, with serious implications and side effects that must be weighed and considered. It is generally only used when there is an immediate danger to life and limb, or when nothing else has worked (and we must present data, graphs, plans, and detailed information about what has been tried and what hasn’t worked). I can only wish that as much care would be taken when trainers insist on using positive punishment with dogs, as often a behavioral review would show that positive reinforcement, discrimination training, interval ratios and other correct manipulation of behavior analytic principles either hadn’t even been tried, or were used incorrectly. I feel very strongly that if a master’s level board certified behavior analyst (a board exam which carries a 42% fail rate for first time exam takers, and a 74% fail rate for subsequent attempts) must use caution, be peer reviewed, and be overseen by an ethics committee prior to using a positive punishment intervention in which they’ve been trained and certified to use, why would a layman, no matter how much experience you have, feel that it is something they should use as a first-line style of training.

So go forth with your new knowledge, and try to identify how these four quadrants maintain your own behavior in your daily life. Forget your keys and can’t start your car? Negative reinforcement for remembering your keys. Put a dollar in the vending machine and receive a snack? Positive reinforcement! Having a bad day and snap at your co-workers, resulting them leaving you alone? Positive punishment for them, negative reinforcement for you. Everything we do throughout the day in controlled by the environment around us and the rules of behavior. Next time we’ll talk about extinction, discriminative stimuli, s-deltas, motivating operations, functions, various styles of reinforcement, and more! So be sure to come back!

 

Interested in becoming a board certified behavior analyst (BCBA)? Check out www.bacb.com for more information

One thought on “Behavior 101 #1: The four quadrants of reinforcement and punishment

Leave a Reply