The continuous nature of targeting

I worked some with Beau over the weekend. Beau is a big black and white horse who was very badly abused at some point. He has made a ton of progress in the past year and is now a pretty friendly guy. Beau enjoys interacting with people and will approach you in the pasture, follow you around, and sniff all over you. This summer, he’s clearly at the point where he’s ready to make some big strides in his training.

I have done only a tiny bit of clicker training with Beau because we have been giving him plenty of time to get comfortable approaching people and hanging out with people. Clicker training can be quite stressful to some horses with backgrounds like Beau’s. If the horse is terrified of people, clicker training can create a conflict. The horse wants to stay around and get the treat, but the horse also still really wants to get away from the person. With Beau, we’ve given him plenty of time to figure out that people are interesting to be around and are not going to harm him in any way. We’re at that point, so it’s time for the clicker training to begin!

Beau already understands the meaning of the click and will wait patiently for a treat without mugging the trainer. On Sunday, I worked some with him with a tennis ball on a stick. I wanted him to target the tennis ball by touching it with his nose. Clearly, this was way too much and he didn’t want anything to do with that tennis ball and stick.

So, I made the task a whole lot easier. I took the tennis ball off the stick and placed it in the palm of my hand. This wasn’t as scary and Beau was willing to sniff and investigate the ball. However, if I clicked while he was touching the ball, he would jump back, surprised. I didn’t want that kind of reaction, so I started changing things up and doing a bit of experimenting.

Trainers often think of targeting (and other behaviors) as single, discrete events. The horse halts, the dog sits, the horse touches the target, and so on. However, all behavior is actually a continuous stream of smaller behaviors. With touching a target, the horse has to turn his head to look at the target, extend his nose toward the target, touch the target, hold his nose on the target until the click, remove his nose from the target, look for the trainer’s treat hand, turn his head toward the treat hand, take the treat off the trainer’s hand, turn his head away, chew the treat, and then begin again with step one, looking for the target. Even each one of these individual steps could be broken down into smaller behaviors and muscle movements.

So, when clicker training, we don’t have to click the moment the dog’s butt hits the ground during sitting or the moment the horse touches the target. We can place our click earlier (or even later) in the continuous stream of behavior to achieve slightly different effects.

For example, when Beau started jumping at the sound of the click, I experimented with changing the placement of the click. At first, I worked some on clicking as his nose was approaching the tennis ball, but before he actually touched the ball. He stopped jumping at the sound of the click since I was clicking before the point where he was previously jumping. This worked well because approaching the ball has to happen first before he can touch the ball. So, even though I wasn’t reinforcing touching the ball, I was still reinforcing a good approximation to touching the ball.

Once Beau had this pretty good, I wanted to find a way to work on targeting the ball, but still without clicking while his nose was on the ball. Time to get creative! Beau and I experimented with a new chain of behaviors. I would hold the ball out. He would touch the ball with his nose for a second or two. Then, I would pull the ball several inches away from his nose and click the clicker. A slightly odd way to do targeting, but this actually worked pretty well! I could tell that he did understand that the game was about touching the ball because by the end he would orient toward the ball and then touch it with his nose if the ball was in front of him, below his nose, or out to the side of his head.

Now, this is not the same as late clicking, when a trainer has poor timing and clicks significantly after the behavior has happened. This is because I created a predictable chain between the behavior I wanted to reinforce (touching the ball) and the click. So, when I removed the ball several inches, this marked his successful touch and also signaled to him that the click would be coming very soon.

Beau and I will keep playing with some of these different variations on targeting. Once he gains confidence with touching the ball and other objects, I’d like to see if we can go back to “normal” targeting, where the click comes while his nose is in contact with the ball. This is a work in progress, so I’ll report back as we keep playing with it. I’d also love to hear what you think about this, since it is a rather unusual way to work on targeting.