Only if the Behavior Increases!

This post is paired with “Only if the Behavior Decreases!

Pop quiz:

Zani can sit on a crate

Zani sits all sorts of places. 

Q: If I give my dog a piece of kibble whenever she sits, is that positive reinforcement?

A: Only if sitting increases.

It sounds simple, doesn’t it, but the second part is so easy to forget! We casually say, “I reinforced that behavior” or even worse, “I reinforced the dog.” (Thanks Eric Brad, who the other day reminded several of us that you can’t reinforce animals, only behaviors.)

A definition of positive reinforcement:

The presentation of a stimulus following a behavior that maintains, or increases, the strength of the behavior. Positive reinforcers tend to be pleasant stimuli, at least valued, from the learner’s perspective. — Susan Friedman, Living and Learning with Animals Professional Course

So most of us remember the part about adding or presenting something. But that’s only half the definition. The other half is that the behavior must maintain or increase.

It makes sense that when one first learns about positive reinforcement, one tends to focus on the added thing because that’s the thing we are learning to do. Add the cookie, the toy. A beginner (including me) might define positive reinforcement as, “Adding something good after a behavior,” or even “Adding something that the animal likes to the environment after a behavior.” Those definitions focus on our action.

There are two problems here. One is that it isn’t always something generally thought of as “good” or even something the animal likes.  For instance, yelling at an animal can be a reinforcer if very little attention is paid to that animal otherwise. And even if they don’t like it it’s often not a punisher. See below. But the more insidious problem is that that definition leaves out the consequence: that the behavior must increase or maintain.

It is not just an aphorism when a behavior analyst says that behavior is defined by its consequences. Look again at definition above. It’s all about the consequences.

Why Wouldn’t Sitting Increase?

OK, so with my little example above. You give your dog a piece of kibble every time she sits, but sitting doesn’t increase. Why might it not? And let’s say that the dog does like the kibble well enough to eat it.

I had a section here with three examples, but I’ve decided instead to be coy (and buy some time to check my terminology). Let’s have a discussion about it. Why might the behavior not increase?

Consequences

Full disclosure: I was inspired to write this pair of posts after I had left out consequences when discussing reinforcement for the umpteenth time. Just a friendly reminder to myself–and all you out there–to pay attention to future consequences, and remember to include them when we even think about reinforcement and punishment. It’s not just about what we do. It’s about what happens after that.

Check out the other post in this pair: Only if the Behavior Decreases!

Coming up:

Eileenanddogs on YouTube

Share Button

About eileenanddogs

Passionate amateur dog trainer, writer, and learning theory geek.

Eileen Anderson on Google+

This entry was posted in Behavior analysis, Operant conditioning, Positive Reinforcement, Punishment, Reinforcement and tagged . Bookmark the permalink.

7 Responses to Only if the Behavior Increases!

  1. Pingback: Only if the Behavior Decreases! | eileenanddogs

  2. Sue Alexander says:

    You need also to ensure that you define your target behaviour. Without a defined target, you don’t have anything to analyse.

    I will also add that yelling at a dog might increase behaviours not because the dog lacks attention, but because of something else. Really. Think about this; the target behaviour is barking. the human and the dog do lots of things together, but the human doesn’t like it when the dog barks, and when the dog barks, the human yells. The doorbell rings (eliciting stimulus), the dog barks and the human yells. The barking increases. The human yells more. The barking maintains or increases. This goes on for a long time; and eventually, both the human and the dog stop yelling and barking. In this case, the dog isn’t getting more attention or a higher quality of attention when the human yells, but he may be engaging in a socially facilitated behaviour. He may continue barking because the human is also making a racket, and he may perceive the human’s behaviour as an alarm behaviour that he can particpate with. Or he may feel threatened by the yelling and be barking to try and drive the human away. Or he may be so focused on the doorbell that he doesn’t perceive the human’s yells and the yelling is incidental. He may even find the yelling to be appetitive or desirable, and he may have learned that the best way to get the human to make noise is to initiate barking himself. The important thing about behaviour analysis is to stop looking for reasons why the reinforcer is reinforcing, but look at if it causes an increase or maintenance in the behaviour.

    You may be interested in some of the definitions for reinforcement that are put out by organizations such as the AVSAB; they have gone through a review process and they have been looked at by a commitee, so they tend to be better accepted than the definition of an individual. You can find the definition for reinforcement according to the AVSAB in their position paper on punishment.

    • All good points. Great study of yelling and barking. And you are absolutely right about not dwelling on reasons. The question is what happens.

      And

        thank you

      for the AVSAB reference. I started out with three definitions of positive reinforcement in this piece: Chance’s; Powell, Symbaluk, Honey’s; and Friedman’s. Then I decided three overlapping definitions really wasn’t helpful. I would have used Chance’s but I really needed positive punishment as well, and couldn’t grab that out of the “Search this book” function on Amazon, grin. Then ended up not quoting anybody in the second piece. I’ll edit the AVSAB definitions in later on. Symmetry between the two pieces would be much better than what I have now, and I appreciate it.

  3. Christina says:

    As a student of this stuff, I find it fascinating! Thank you for taking the time and energy to analyze all the details.

    To me, what you are talking about here, the definition of reinforcement, is the one point where there is a gap between theory and (dog training) practice. According to the definition, you don’t know until after everything is done whether the consequence you provided was reinforcing. If I tried to teach my dog to sit by patting him on the head several times every time he sat, for two weeks, I could at the end use this definition to conclude that because the behavior did not increase, head-patting is not a reinforcer. But it wouldn’t tell me at the beginning that I shouldn’t use head-pats to try to train my dog to sit. In other words, the abstract definition is descriptive of how learning works, but not instructive of how to train a dog (a specific dog in a specific instance). A dog trainer ideally wants know before beginning the training program what is going to increase the occurrence of the behavior they are trying to train – that’s the whole point, right!? That’s I think why we so easily slip into thinking of positive reinforcement as “adding something the animal likes after the behavior”. Of course, in reality, there are certain things that are usually reinforcing (treats, toys), and importantly, once one understands the definition, one can critically evaluate a training program and make adjustments if needed.

    It’s an interesting glitch – not a big deal in real life, but a tiny mismatch that catches my brain every time I think about it.

  4. Pingback: Weltweites Wissen am Wochenende – Wie Tiere lernen – Verstärkung | Chakanyuka

  5. Pingback: Can We Determine Whether Training with Food Is Positive or Negative Reinforcement? | eileenanddogs

  6. Pingback: Choice isn’t a reinforcer. So why is the behavior increasing? – nala the wonder dog

Comments are closed.