Randomly and all the time

These two phrases — “randomly” and “all the time” — are ones that I hear quite often when I am working with my training clients.

Often, dog owners call in a trainer because their pet is exhibiting behaviors that are annoying, frustrating, or even downright dangerous. My job as a trainer is to identify the behavior, figure out why the dog is engaging in this behavior, and then coach the owner as we teach the dog new, more appropriate patterns of behavior. (And usually, the human gets just as much or more training than the dog!)

One of the first questions I have to figure out is “Under what conditions does this behavior occur?” Before we start any training program, I need to have a pretty solid idea of when and where the behavior occurs and why it might be occurring. During what situations does the behavior occur? Are there certain factors that make the behavior more (or less) likely to occur? Why might it be beneficial for the animal to engage in this behavior?

So, I start by asking the owner “When does your dog do this behavior?”

The two very common answers that I get to this question are “randomly” and “all the time.”

Well, that’s not much help!

A similar answer that I sometimes get is that the dog “never” does the behavior. For example, my clients often complain because the dog never comes when called. This spring I worked with a client with a young lab mix. The owner was exasperated because the dog “never” came when called. However, after some questioning, I figured out that the dog actually did great when she took him to the park and let him off leash and would come back every time. However, he often ignored her if she called him in the house. We approached this training situation very differently than we would have if the dog had been good at coming in the house, but had needed help when outside.

What if behavior was really random?

With some behaviors, it can sometimes be tough to figure out exactly when the behavior occurs. As a result, it is easier to say that the behavior is “random,” rather than to try to figure out the precise conditions. I’ve worked with clients with puppies that are mostly potty trained, but that “randomly” have accidents inside or adult dogs that “randomly” bark at strangers.

However, imagine what a mess the world would be if many of your behaviors did truly happen at random! You might lie down and go to sleep while riding up an escalator. Or you might stand up and start reciting the Gettysburg address during a meeting at work. Or any number of equally strange possibilities.

As a behavior analyst, one of the basic assumptions that guides all of the work that I do is that behavior is not random. Behavior happens for a reason. And by “reason” I mean that behaviors happen in certain situations and not in other situations because of the consequences (reinforcers and punishers) that have followed different behaviors in those situations.

For example, if surrounded by a hundred other people, a person will yell loudly if he’s at a baseball game, but whisper softly if he’s at church. Most people have a long history of reinforcement (and often punishment!) that teaches them when it is appropriate to yell, when to talk in a normal voice, and when to be sure to use the tiniest of whispers. (Remember that grumpy librarian from elementary school who was always glaring at you if you talked too loudly?) Of course, there’s probably been times when someone whispered at the baseball game or screamed at church, but I bet the person did so for a particular reason.

Animals (including humans) do what “works” to gain access to the things that they want and to avoid the things that they don’t want. Although certain behaviors can seem odd or out of place, they usually start to make sense if we can figure out more about the animal’s past experiences.

Unfortunately, behavior can seem “random” because most people are not very good observers of their own behavior and even poorer observers of the behavior of people and animals around them. I often have clients tell me that their dog does a certain behavior “all the time.” When I inquire if there are specific times or situations when he does or does not do the behavior, they tell me “no.” The dog does the behavior in all situations all the time.

So, then when I’ve been at the client’s house for about 20 minutes, I’ll sometimes point out that the dog hasn’t been doing the behavior for the past 20 minutes. So, this tells me that the dog sometimes doesn’t do the behavior when new visitors come over. Interesting! This can sometimes be eye opening and then I can often get the client to start listing situations or certain contexts where the dog does or does not do the behavior.

During training, we sometimes try to solve a behavior problem without completely understanding all of the different factors and triggers that relate to the behavior. This can sometimes be successful, but it can sometimes backfire on you. The more you know about exactly when, where, and why the behavior occurs, the better prepared you will be to design and implement a successful training program.

Time to be a detective

So, you have a behavior that you want to change. And, you’re working under the assumption that the behavior is not happening randomly. So, when does the behavior occur?

Remember what I said above, about how most people are not usually good observers of behavior? More often than not, my clients can give me a fairly good description of when a behavior occurs, but not a complete description of all of the relevant factors, conditions, and variables.

For example, imagine a young German shepherd that is “randomly” growling at friends who come over to visit. The owner is perplexed, because the puppy only seems to do it at about one out of every five people and sometimes the pup will be nice to someone one week, but growl at the same person the following week. It will be much easier to teach the puppy to be friendly and relaxed around strangers if you can figure out why he’s getting upset around certain people.

I often have my clients take data. If it’s a behavior that happens all throughout the day (barking at things out the window, potty training issues, stealing things) I have the client start by taking data on when the behavior occurs. This can be very revealing!

For example, I had one client with a young cocker spaniel puppy that was doing pretty well with potty training, but that was still having 1-2 accidents at “random times” throughout the day. The client understood to take the puppy outside on a frequent and consistent schedule and was either crating the puppy or watching him very closely when he was loose in the house.

After keeping a log for about a week, the client determined that the dog was often having accidents in the kitchen in the morning or evening when she was cooking meals. She realized that she “thought” she was keeping an eye on him because they were so close together in the same room, but she usually wasn’t actually watching him because she was distracted by her cooking. She added in an extra potty break around these times and started keeping a closer watch on him. If she was cooking something that really required a lot of concentration, she would put him in his crate or play pen so that he wouldn’t have an accident in the house.

Under what conditions?

I often start by having my clients keep a log of “when” the behavior occurs. This requires some effort, but not too much effort and can result in meaningful data. When clients see the patterns in the data (with some help from me) and how this will be useful for training, it’s usually not too difficult to ask them to start collecting data on additional variables.

For example, let’s return to the hypothetical young German shepherd that is sometimes growling at friends who come to visit. The owner knows the dog is more likely to growl at men and thinks that the dog is also growling more often when the friend has a beard. However, she’s perplexed because the dog sometimes will be friendly with someone one week, and then growl at the person the following week.

For about two weeks, you have the client takes data on:

• Who the dog growls at
• When the episode occurs (time of day, day of week, etc.)
• Where the episode occurs
• Who else was present
• Where the person approached from
• What the dog and people were doing before the new person approached
• What each person did after the dog started growling
• Any other things going on that might be important

After taking data, you and the client realize that the growling is closely related to whether or not the client’s brother is present and that it just so ends up that many of her brother’s friends have beards. (The growling could be because of something the brother is doing, or because the client behaves differently toward the dog when her brother is present, or a number of other reasons.)

You might still have some detective work to do to figure out what is going on when the brother is present that is not happening when the brother is absent. However, you are now a lot closer toward being able to design a really super training program for this dog, than if you had gone ahead and started a training program based on the assumption that the dog didn’t like men with beards. Because of data, you’ve identified new, relevant factors.

Being a dog trainer is a bit like being a detective. You have to like puzzles and you have to be able to look at a problem from all sorts of different angles. Sometimes, even after you’ve collected some good data, it really does still seem like a behavior is happening “randomly.” In these situations, it can be helpful to try to approach the puzzle from a completely new angle or to talk over the situation with another trainer who will have a fresh set of eyes.

I often do see trainers start a training program before they really have gotten a good enough sense of what is going on. I know I occasionally have done this too. Solving tough training puzzles can be difficult if the client needs immediate help, you have a limited amount of time, or you have difficulty getting the client to gather or provide enough information.

Imagine, though, if you had not collected enough information about the German shepherd and you had jumped into setting up a training program to help the dog learn to like men with beards. Maybe you get lucky and you do have some success with this training program. Very likely, however, this training plan is not going to be successful because you are targeting the wrong variables. Being able to pinpoint important factors and variables will allow you to design training programs that will be the most effective and that will have long lasting results.

One more example. Imagine you’d like to lose a bit of weight and you know you often snack too much at work. You start bringing healthy snacks with you and keep them at your desk. However, you still find yourself often going down to the vending machine to get a candy bar or going across the street to get a cookie or muffin from the local bakery.

So, you start taking data on your snacking behavior, including what usually happens before you leave your desk to go buy a snack and what happens afterward.

You have a pretty demanding job and, from taking data, you start to realize that you generally have one or two high stress conversations with your boss mid-afternoon. It’s usually after this that you need to get up, walk around, and get away from your desk for a few minutes. And, this usually leads you to the vending machine.

So, your idea of bringing healthy snacks didn’t work because what you are really craving is not a midday snack, but a short break from work that will help relieve some of your stress. Some possible solutions you might try could include improving your relationship with your boss, going and eating your healthy snack in the break room (so that you can get away from your desk for a few minutes), or finding some other way to take a break.

Final thoughts and a challenge

Behavior is not random. However, for perplexing behaviors that we do not understand, it can be tempting to label the behavior as “random.” This is largely because most people do not have good observational skills. Also, for behaviors that happen infrequently (once or twice a day, or even once or twice a week), it can be challenging to pick up on relevant variables and patterns. Keeping a log and collecting data can help you identify important and sometimes quite surprising information about when and where the behavior occurs.

Want a challenge for this week? Pick a behavior that you think usually happens at random (or just a behavior that you don’t feel that you know much about). It could be a behavior of one of your animals or even one of your own behaviors. Pick an interesting behavior or even just a “boring” behavior, such as how often you get up to get a glass of water. Keep a log of the behavior for at least 3 days. See if you find out anything interesting!

Photo credit: “Kim at 14 weeks” by Marilyn Peddle – originally posted to Flickr as Kim at 14 weeks. Licensed under CC BY 2.0 via Wikimedia Commons.