equine clicker training

using precision and positive reinforcement to teach horses and people

Capturing Behavior

busterlaydown_smallIn Kay Laurence’s IDTC class last year we took a closer look at capturing behavior.  I’ve always been intrigued by capturing behavior because it’s one of the advantages of clicker training and there are some behaviors that are easier to capture than to shape.

A behavior is “captured” when the trainer marks and reinforces the animal for doing a complete behavior, instead of shaping the behavior in small steps.  In theory, capturing behavior is simple because a clicker trained animal has learned to repeat behavior that has been marked and reinforced.  Then once the behavior is being repeated, then you can add some stimulus control.   In practice I find it’s not always that easy, but under the right circumstances, it can be a very useful way to get some behaviors.  Here are some tips to help you become more successful at capturing behavior.

One thing to keep in mind is that while the word “capturing” usually refers to marking and reinforcing a complete behavior (a jump, spin, laydown, etc…), captured behavior can be further refined through shaping just like any other behavior.   The only difference is that you are starting with a larger “chunk” of behavior.  Let’s look at the advantages and disadvantages of capturing.

Advantages to capturing behavior:

There are some behaviors that are easier to capture than to shape. These are usually natural behaviors that occur on a regular basis. Some examples are stretching, yawning, large body movements, spinning, the flehmen response, vocalization and lying down.

It’s also quite easy to teach a horse to go to or stay in a certain location just by reinforcing them for being there.  I’ve taught some stationing behaviors in the field by reinforcing a horse when it stood in a certain spot and over time, the horse learned to go there and wait for reinforcement.

Capturing can be faster than shaping a behavior from the very beginning. I tend to think of capturing as jumping into the training progression farther along in the process.   If I have 5 steps: A,B,C,D, and E and I want to get to my final behavior of F, I can get to F faster if I start at D than if I start at A.

Because there are fewer steps in the process, there are fewer “previous versions” of the behavior that the animal can offer instead of the one you want.  Note that I’ve listed this as a disadvantage too!

It’s a great way to put novel behaviors under stimulus control. Sometimes our animals are more creative than we are, and I love being able to capture a behavior that I would not have thought of training or known how to train.

Disadvantages to capturing behavior:

The behavior has to happen a few times, and preferably on a regular basis with some kind of predictability, in order to be captured.

Because the behavior is captured in a larger chunk, it’s harder to go back and rebuild the behavior if you lose it. With a shaped behavior, I can always go back to the original shaping process if the animal stops doing the behavior.   If the behavior is captured, I don’t have any way to remind the animal by reviewing the original training process.

It can be hard to identify the antecedents and get the behavior reliably on cue so that I can ask for it in other situations. Captured behavior is often the strongest under the conditions in which it was originally reinforced. It can certainly be done, but I find that it can sometimes take longer to get reliable stimulus control compared to a shaped behavior.

If the animal is not aware of what it’s doing, the click and reinforcement can end up initiating a training session, not identifying and reinforcing the behavior it just did.   This is actually a fairly common experience because animals are not always paying attention to what they are doing. This happens with people too. Have you ever had someone ask you what you just did or said and you have no idea? You weren’t thinking about what you were doing as you were doing it, so you have to try to remember what you were doing in order to answer the question.   The same thing can happen with your horse if you unexpectedly interrupt its normal activity by clicking.   Then what often happens is that the horse offers a different behavior that it already knows, instead of the one you just marked and reinforced.

Tips to make capturing more successful:

Observe your animal. Behavior always has antecedents. If your horse does a behavior on a regular basis, spend some time observing when he does it and see if you can identify what happens right before or the conditions under which it happens.  This makes it easier to predict when the behavior will happen (so you’re ready) and in some cases, you can even set up the environment so the behavior you want is more likely to happen. Antecedents can be something specific that happens (a behavior you do or the presence of a specific item), or they can be a set of context cues (wet horse, hot day, sandy spot) or even things like time of day, feeding time, or some other part of the daily routine.

I have successfully captured a few behaviors that would happen predictably when I first approached my horse or dog. I captured “nickering” because Willy would nicker when I entered the barn in the morning. I captured a stretch with Bandit, our dog, because he would do it when I returned home and he got up from sleeping. If you can identify and control the antecedent, capturing the behavior will be  much easier.

Pay attention to your timing.  When I shape behaviors, I am looking for tiny changes that I can click and reinforce.  With capturing, I want to wait and mark the entire behavior, so I usually click near the end of the behavior, or past the point at which the animal is committed to doing it. Clicking too early can interrupt the behavior and make it harder to capture it.

If clicking and treating the behavior seems counterproductive because the animal then offers other behaviors, try just reinforcing the behavior. You can also try using other reinforcers that are not as closely associated with training sessions. Bandit would offer to sit when I clicked and reinforced stretching so I just reinforced stretching (no click) and tried several different reinforcers (food, attention, patting). Over time I did see an increase in the number of times he stretched when I got home and he started to offer it at other times too. Once it was happening more and also happening out of the original context, I could start to click and treat for it, just like any other behavior.

I said capturing can be faster and I think it can be for some behaviors, but in other cases it requires a lot of persistence.  It may be that the actual training time is shorter (if you count minutes spent on a particular behavior), but if you only have a few opportunities each day when the behavior is likely to happen, then it will take longer overall.   So don’t be discouraged if it seems like progress is slow, just remember that if you keep reinforcing a behavior, it is more likely to happen again.

If you can’t control or identify the antecedent, you can try adding something into the environment when you expect the behavior to happen. The idea is to introduce an antecedent that you can control. An object or a long duration cue often works well because it allows the animal more time to respond. I think of this as introducing the idea that if you do the behavior under these conditions (when this object is present), I’ll reinforce you for it, not as much about adding a cue to which I expect you to respond. I am just trying to create a new association. I’ve only experimented with this a little bit but I found it worked well for a few behaviors that were based on environmental cues.

A behavior is not truly captured until it’s on cue and you can use the cue to ask for it at other times.   It may take you a while to get there.  The first step is to click and reinforce (or just reinforce) the behavior until it happens reliably under the original conditions and/or starts to happen at other times. Then I usually add a cue. Once I can ask for the behavior a few times on cue in the original setting, then I start to work on asking for the behavior on cue in other situations, choosing them carefully so the horse continues to be successful.

Don’t forget that you can capture behaviors during training sessions when the horse offers something unexpected. Rosie learned to canter in-hand because she offered it when I was doing some trot work. I was not actively cueing a canter and didn’t expect it, but when she offered it, I clicked.   In some ways capturing  an  unexpected behavior during training is easier because the horse is already in training mode and in this case it was easy for me to repeat the steps I had just done.

An example of a captured behavior:

I don’t have any video that shows capturing a behavior because it’s not necessarily something I plan in advance and early sessions are often short because I might only get to click and reinforce the behavior once.  So instead I’ll show a fun video of Buster lying down on cue, which is a behavior I captured when he decided to lie down in his stall while I was working in the barn.

It took me three days to go from the first clicked and reinforced laydown to having it on cue, but I think that’s unusual. There were a number of things that worked in my favor.  One was that he had already done a lot of clicker training so he knew that he should repeat something when I clicked.  Another was that he had already learned to bow and kneel so offering those behaviors, which often precede lying down, was something he tried.  It also helped that he is small so this behavior is easy for him, and I was able to be present at a time of day when he wanted to lie down.  Just for comparison, it took me 3 months to shape my TB Willy to lie down.  But he was older and bigger, so while he might lie down once, he would not necessarily repeat it.  That meant it took longer to get it on cue.

 

 

 

 

Categories: Uncategorized

Tags: , , ,

2 replies

  1. I captured my horse’s sequence of parking out and then stretching down (like downward dog in yoga). It was a cute behavior and I also wanted to incorporate it into my pre-ride stretching routine. It took us years to really get it associated with a cue because it occurs so infrequently. He tends to do it in cross-ties but not often. After a year or so of just clicking and rewarding every time I saw it I finally got the light bulb moment with him again and he offered it twice. After that we could go through our normal routine of reinforcement and then adding a cue etc. Took a lot of patience to train like this but it was really fun to see the final ah-ha!

    Like

    • It does take persistence, doesn’t it? That’s great that you are able to get the stretch on cue. I know a few people who are still working on it with their horses. I’ll have to check and see how they are doing.

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s