Home › Blogs › Aidan's blog

Conditioned Reinforcers - Part 1

Aidan's picture
Wed, 08/01/2007 - 11:19 — Aidan

What is a Conditioned Reinforcer?

A conditioned reinforcer is a stimulus that has taken on the properties, by association, of a reinforcer.

Whoa! Let's back up a bit...

What's a Reinforcer? A reinforcer is anything that maintains or increases operant behavior - said more plainly, a reinforcer is something that an animal will work for. Any behavior that your dog does voluntarily has been reinforced.

A stimulus is simply something in the environment that you can sense - a sound, an object, a color, something you can feel etc.

So when we say that a conditioned reinforcer is a stimulus that has taken on the properties, by association, of a reinforcer what we are really saying is that something in the environment has been associated with a reinforcer so much that it itself becomes a reinforcer.

Perhaps an example might help?

Take the humble clicker from clicker training. What we do when we use a clicker is pair the click sound with food, or toys, or games - something that our dog will work for. We use the conditioned reinforcer to mark the behavior.

You can actually "charge up" a clicker and use it without following up with a reinforcer. It can be demonstrated that this will continue to reinforce behavior - for a while. Eventually, it loses it's power. A better use of the clicker is to always follow up with a reinforcer. Make your click a promise that something worth working for is coming pronto!

Experiments show that the timing of a reinforcer is vitally important for the animal to learn exactly what they should do more of. The reason for using a clicker is to improve the timing by marking the exact behavioral response that leads to the reinforcer.

In some training situations you can't deliver a reinforcer at the precise moment that the animal performed the behavior you want to see more of. Thankfully, nature gave us a mechanism for linking behavior with reinforcer and we exploit that opportunity deliberately when we use a clicker.

There are still lots of training situations where you don't need to use a clicker and humans are endlessly adept at getting a job done without technology. But this is cheap, simple technology that offers a clear advantage and is therefore worth taking advantage of.

Now, what nature did not give us is a "mind wipe" device that somehow magically leads the animal to ignore everything between the click and the reward. Behavior happens between "click" and "treat", and this behavior is sometimes reinforced so deliver the treat as soon as possible after you click to minimise any learning after the click.

If you're a really advanced trainer you can even take advantage of this by deliberately delivering your treat in an advantageous position where such an opportunity exists. Personally, I'm not usually that clever and I like to keep things simple. I stick to these rules:
1. click then treat as soon as possible
2. always treat after click

Comments

Wed, 08/01/2007 - 13:14 — Jane Higgins
Jane Higgins's picture

Reinforcer

This is exactly what I have done (without knowing the 'science' behind it) when I have taught Frankie to shut the back door after he comes in. Frankie hates noise (especially sudden ones) so I taught him to close the door & he was clicked & treated for a successful shut. He used to 'wince' when the door banged shut but he now does it of his own free will. I can let him out at night for toileting, I leave the door ajar & when he's 'done his business', he comes back inside & shuts the door (while we're still sitting watching TV or reading). I usually treat him when he's back on his bed so now the treat doesn't come immediately although when we hear the door close we always say "GOOD boy!" It's working for him and it's working for us because we don't have to get off our lazy backsides to shut the door again!