A Training Strategy for Building Duration: Backchain it by using a terminal behavior.

Duration with a short chain.jpgBuilding duration is often a sticky point for many trainers. I know that I find it fun and easy to get behaviors started, but when it comes to building duration and adding stimulus control, it starts to feel a bit more like work. Part of this is because those two processes often involve withholding reinforcement for an existing behavior (which seems slightly unfair) and keeping better track of what you are doing (which requires better record keeping).   This is a little different than the shaping part where I have a training plan, but my job is to work with what the learner offers and the learner typically has more freedom and choices about how we do the training.

But that doesn’t mean that building duration has to be difficult. It’s really just a matter of having a variety of options for ways to build duration and choosing the right one for the animal and the behavior that you are training. I was originally taught to build duration by slowly delaying the click so the animal has to do the behavior for longer before it gets the click and treat. There are lots of additional details and tips that can make this successful and if you are new to building duration, you might want to read the article on building duration on my website.

While the method I described above is one I still use at times, I’ve also learned some other ways that can make building duration easier for certain types of behaviors. One of my favorite ones is to increase duration by using a short chain of two behaviors.  The first behavior in the chain is the one that I would like to get more of (increase the duration) and the second behavior is what Kay Laurence calls the “terminal behavior.” It’s the behavior that tells the animal when it is done with the first behavior. You can also do this with more than two behaviors, but your chain will always include the duration behavior and the terminal behavior.

I like this method because it means that I can set up the training so the animal learns a predictable end to the behavior, and this often works better than having it be the more open ended “do it until I click.” It also takes advantage of the power of backchaining which is when you build a chain by starting with the last behavior and lengthen the chain by adding new behaviors to the beginning.

I first started playing around with this idea when I was teaching young horses to pick up and then continue to hold up their feet. Since they have often gotten a lot of reinforcement for picking the foot up and immediately putting it back down, it’s a big criteria shift to ask them to hold it up for longer. I always felt like they were a bit confused about why they didn’t get clicked for picking it up, and also when they did hold it up, they were waiting anxiously for the click. Since relaxation is important when handling feet, I started looking for ways to give them information about what I was doing so they knew when they were getting close to the end of the behavior.

Why does this matter? I think it matters because it changes the horse’s focus. I think of it as creating a “forward thinking” horse so that instead of the horse thinking “Am I done? Am I done? Am I done?” while it is doing the behavior, and looking for any sign it might be done, the horse is waiting for, or moving toward the next behavior. It might seem like this is the same as waiting for the click, but in practice I find that there is often a difference in the emotional state of a horse waiting for a click vs. a horse waiting for the next behavior.

An additional advantage is that using a terminal behavior can make the behavior clearer to the horse by providing a concrete end point. I have always felt that Rosie does better when she can see the point of what we are doing or there is a clear goal behavior.   I have also found that horses will usually try and look for information that predicts when they are getting close to the click. By choosing and deliberately using a terminal behavior, they are less likely to start to use other behaviors (that I might be unaware they are using) and try to end the behavior early because they are using information that I didn’t intend for them to use.   Not every advantage applies to every use of this technique, but hopefully you’ll get a feel for how and when to use it as I share some different examples.

Back to hoof handling. It turns out that I am pretty systematic about how I teach hoof handling and once I started thinking about it, I realized that I was already set up to create a little chain to build duration for holding the foot up. My usual progression is:

  • Can you pick up your foot and hold it up for 1 second?
  • Can you pick up your foot and hold it up for 2 seconds?
  • Can you pick up your foot and hold it while I brush it with my hand?
  • Can you pick up your foot and hold it up while I brush it with my hand for longer? (I do a few repetitions of this until I have enough duration to match one stroke with the hoof pick)
  • Can you let me use the hoof pick once?
  • Can you let me use the hoof pick twice? And so on… until I can pick out the entire hoof before putting it down.

Normally I progress through these steps, but what if I continued to do the same behavior right before I put the foot back down? Would that provide some consistency so that the behavior was “hold up your foot until I do this behavior” and not “hold up your foot for a variable amount of time.” So I tried it using the behavior of “brushing the foot with my hand” and it worked very well. Even once I was up to cleaning the entire foot, I always finished by brushing it with my hand before I put it back down. It looked like this:

  • Can you pick your foot up hold it while I brush it with my hand?
  • Can you pick your foot up and hold it for one swipe of the hoofpick followed by a brush with my hand?
  • Can you pick your foot up and hold it up for two swipes with the hoofpick followed by a brush with my hand?

The horses learned that I might do various thing with their feet but that I always ended with brushing the foot with my hand. I found this made them calmer while I was doing the other hoof behaviors because they were not waiting with such anticipation for the click. This made it easier to build duration by slowly increasing the number of things I did before I brushed the hoof with my hand. I chose to use brushing the hoof with my hand because it’s easy to do, regardless of why I have the foot up, and it’s the first thing I usually do with babies, but any simple behavior that includes holding the foot quietly will work as well.

Now one obvious question would be “Isn’t putting the foot back down a good terminal behavior?” and in a more advanced horse, it could function as one. But what I have found is that when using a terminal behavior as part of the teaching process, it is important to choose a terminal behavior that I can control. By control, I just mean that I can decide when to ask for the behavior or set up the environment so that the behavior happens when the horse has met the desired criteria for duration. Yes, the horse can stop or cease doing the behavior at any time, so this is not about making the behavior happen, but if he wants to get clicked and reinforced, he has to complete the chain with no shortcuts. It’s worth remembering that the downside of chains is that animals will rush through them to get to the end, so if I want to build duration using a chain, I need to set it up so that it’s clear when the criteria for duration has been met and that there is no reinforcement for jumping to the end of the chain.

This might be clearer with another example of how to use a short chain to build duration in movement. Here’s one way to increase duration at the walk by adding the terminal behavior of going to a mat, target, or other destination.

Let’s say I have trained my horse to walk forward on cue and he will walk for a few steps (3) before I click. Now I would like him to walk forward for more steps, building up to the ability to send him forward and having him keep walking until he gets another cue. I can do this by setting it up so that he is doing a short chain: walk forward -> stop at destination (mat, target, specific location). The destination should be indicated by a behavior the horse already knows so if my horse doesn’t know matwork or targeting, I need to teach that first.

I start by placing the destination object, or positioning the horse, so he has to walk the number of steps he already knows (3, in this example) to arrive there. Once he has learned to do this, I can start to build more duration for walking forward by changing the distance to the destination. I still want to follow the general guidelines for building duration by increasing the distance slowly and maintaining some variability so the horse has some easy reps mixed in with some harder reps, but it’s easy to do this because I get to decide how to set up each repetition. Here’s an example using walking forward to a mat. It might look like this:

  • walk 3 steps to mat – click/treat
  • walk 4 steps to mat – click/treat
  • Walk 6 steps to mat – click/treat
  • walk 4 steps to mat – click/treat
  • walk 5 steps to mat – click/treat
  • walk 7 steps to mat – click/treat

If my goal is to fade out the mat, at some point I can just start clicking before the horse gets to the mat and he will learn that he is getting clicked for walking forward and not necessarily for going to the mat. I can do the same thing with targets or specific locations. I’ve done a lot of work where Rosie gets clicked for arriving at a corner.

Once you get the idea of using short chains, it becomes easy to think of ways to use terminal behaviors to improve preceding behaviors. Some examples of other ways to use them to build duration are:

    • Backing – In the same way that I can use a destination to build duration for walking forward, I can use a destination for walking backwards. I’ve played with using targets and mats to help a horse learn to back for more steps or with better mechanics (diagonal pairs, relaxed topline).
    • Any other movement behaviors – Giving a horse a destination can be incorporated into a lot of ground and ridden work to make it easier to build duration.
    • Husbandry behaviors – A lot of husbandry behaviors require the horse to stand or hold still while I do something. If the horse starts to anticipate the end of the behavior and end it too early, it can be helpful to add a specific terminal behavior that teaches the horse to maintain the behavior for slightly longer.
    • Leg lifts – I’ve used a touch as a terminal behavior for leg lifts. The horse learns to hold his leg up until I touch it. This has the nice advantage that the touch can also be used to reinforce the leg being in the desired position because touching the leg when it is in a certain position also reinforces that variation on the behavior.

General tips:

    • Choose a simple, easy, and short behavior that you can use to mark the end of the behavior. It can be a change in your behavior (something you do to the horse), or the environment, or something you set up as a goal behavior for the horse. I like to think of the terminal behavior as the period at the end of the sentence.
    • Consider whether or not you will want to continue to use the terminal behavior as part of the final behavior. If you just want to use it as part of the teaching process, then you need to plan how you will fade it out.
    • Using a terminal behavior does not mean you never click for the duration behavior.  You can (and probably should) sometimes click and reinforce the duration behavior at times. This is because some chains can start to break down if individual elements are never clicked and reinforced. I usually start by clicking the duration behavior to get it started, then I add a terminal behavior, and start working toward my goal duration. At various points along the way, I may choose to click for the duration behavior instead of cueing the terminal behavior. I tend to do this either to confirm that the duration behavior is the right answer or if I want to introduce the idea that I might not always ask for the terminal behavior.
    • Remember to choose a terminal behavior that cannot be offered prematurely by the horse.  One of the downsides of chaining is that animals tend to rush to the end of the chain if they think they can. I try to think carefully about how to set up the training so the horse can’t take a shortcut and expect to get reinforced.

I’ve put together a little video that shows the use of short chains to build duration. You can watch it at: https://www.youtube.com/watch?v=3mjLuUpWijw&feature=youtu.be. Please ask questions if you have them. Here is a brief description of what we were doing:

Aurora:

Foot pick up and hold. Aurora was learning to hold her feet up for longer. I am showing how I use the behavior of brushing the bottom of her hoof before I put it back down.

Red:

Backing to a mat. Red has several cues to walk backwards, but over time his default response to his back cue had become to back 2 steps and then stop. This was often all I needed around the barn, so he lost the idea of backing for more steps. Since I didn’t want to build duration by withholding the click if he only backed two steps, I set up a new exercise using a mat so that he learned to back for more steps as I increased the distance to the mat.

I have to say that when I started this project, I underestimated how hard it was for him to back 5 steps in a straight line. So, in addition to getting 5 steps of backing, we spent a lot of time working on symmetrical and even steps. This was great for him as it helped him work through some crookedness in his body, but it did make the behavior more complicated to teach. You’ll see me checking the position of, and resetting the mat because I wanted to make sure it was directly behind him so he would land on it only if he backed in a straight line. If I hadn’t wanted to work on backing in a straight line, I could have used a wider mat.

I also want to note that I wanted a square halt on the mat since this is a behavior he knows and this exercise was part of set of related exercises that were focused on straightness and stopping in balance. If you are just using the mat to mark the end of backing, there’s no need to be that particular about how the horse lands on the mat.

Red:

Backing while I’m sitting in a chair. This was just a fun project that came out of the backing to the mat. I wanted to see if he really understood that the back cue meant to keep going and if he would do it without me accompanying him (I usually walk toward him as he backs).   I taught this by slowly shifting the position of the chair (relative to him) so he learned to back up until his nose was even with the chair. I did add a pole to help him stay straight since my position makes him inclined to want to orient toward me. When I have time to play with this again, I could fade out the pole if that was important to me.

8 thoughts on “A Training Strategy for Building Duration: Backchain it by using a terminal behavior.

  1. Hi Katie,
    Thank you for a great post! I was wondering, would the cue “whoa” serve as a terminal behavior when teaching duration for a gait such as walk, etc. The cue “whoa” would be well understood through in-hand practice. Would this compare favorably with using a destination such as a mat or corner? Or, am I misunderstanding what a terminal cue is and it’s more about being a cue which predicts a change in behavior?
    Thanks for helping me understand this better!

    Like

    • Hi,

      I don’t usually use “whoa” as a terminal behavior unless it is associated with a destination, as in “go to this place and then stop.” The reason is that I don’t want the horse to think it’s worth offering the terminal behavior before he has completed the desired amount of duration. If I use “whoa” on its own, then a clicker-wise horse may start offering to stop before I am ready and this can frustrate both of us.

      The idea behind the terminal behavior is that I can set up the environment or create a predictable pattern so that I can control when the horse does the terminal behavior, while at the same time showing the horse what is expected. I hate to use the word “control” but in this case, it just means that I want to be able to set the amount of duration ahead of time and not have the horse guess when I am going to click or ask for the next behavior. let me know if this is still not clear.

      Liked by 1 person

  2. Thank you for all this information. Ive been doing CT with my mare for 3 mths now, unfortunately I think I rushed a little and nothing was concrete. We where mounted and doing targets, a broken clavicle later and 2 mths off work I’ve got a lot of reading and groundwork to do. Thank you again for your wealth of information.

    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