Eric Brad posted a really great question this last summer on his FaceBook page, Canine Nation. Yes, it’s December now. It takes me this long to mull things over sometimes.
Have any of you ever used a technique for teaching/encouraging recalls called “Runaways”? It involves running away from your dog, hiding from your dog, or even getting in the car without them when they choose not to come when called.
Can you explain the premise on which this technique is based and why it can be effective in getting the dog to come back more reliably?
I looked around and there are several variants of this. Most people recommend the human running away and hiding as a consequence for a dog not coming when called. Ian Dunbar recommends it (as a one-time exercise) without the recall. Just leave the puppy when he is preoccupied and hide from him. Yet other people use running away from a puppy and even playing hide and seek as a game and a motivator. And Trish King of Canine Behavior Associates has a protocol called Abandonment Training that has some similarities but also important differences.
Let’s look in depth to see how this “runaway” thing can work.
First, here is how people seem to think the first version is supposed to work.
Version 1: Negative Punishment
Just to be clear: the following is not a good behavioral analysis.
- Antecedent: Human calls dog
- Behavior: Dog fails to come (Note: NOT a good behavioral description)
- Consequence: Human runs away and hides
- Prediction: Failing to come when called decreases (probably untrue)
I believe that people think they are using negative punishment in this scenario. Negative punishment is the process where something is removed after a behavior, which results in the behavior happening less often. The human removes him or herself from the dog’s environment. But Houston, we’ve got a problem.
In behavior analysis, we need to specify a behavior, in this case, whatever is thought to be punished. And we didn’t. “Failing to come” when called is not a behavioral description. People try to make it into one when they say the dog is “blowing them off” or “giving them the paw.” The fact those also are not realistic descriptions either should give us pause.
So let’s think about what the dog might be doing instead of coming when called.
- Chasing a squirrel
- Digging a hole
- Staring at something in the distance
- Barking at a stag beetle
- Et cetera! The possibilities are infinite.
Problem #1. If we think we are punishing a behavior, what is it? We can’t be punishing an individual behavior if the dog is doing something different every time he doesn’t come when called and we left, and he probably is. However, it gets worse.
Problem #2. Timing. Let’s say Fido is sniffing the ground. You call him. He keeps sniffing. You disappear. Sometime later, he notices you are gone. The problem is that your leaving and his sniffing are not connected. We can’t really expect Fido to review the last two minutes and think, “Oh let’s see, she called me. What was I doing then? Oh yeah, I was sniffing. Now she has disappeared. I’d better not sniff anymore.”
So this may feel like punishment from the human’s standpoint. “I’ll show him! I’ll go get in the car if he doesn’t come!” Just like some beleaguered parents end up threatening to do with their kids. But the analysis doesn’t work out that way. (See below under Abandonment Training for a protocol that removes these problems.)
In truth, we are not looking to punish something as much as we are trying to get behavior. We are trying to increase the behaviors of the dog staying connected and coming when called. So let’s re-analyze it as a reinforcement scenario. We will have to move our behavioral lens over one notch and start with the human leaving. So now we get the Dunbar version.
Version 2: Negative Reinforcement
- Antecedent: Human hides, leaving puppy alone
- Behavior: Pup looks for and finds human
- Consequence: State of being alone ends
- Prediction: Pup looking for human increases
The behavioral mathematics work out better here, although it’s a tricky scenario. I’ll be discussing it in more depth in a later post**. The aversive is being alone, which is very scary for many animals, especially young ones. The puppy can learn to escape being ditched or shorten the time by looking for the human more often, and noticing sooner when the human disappears. The pup can learn to avoid it entirely by looking for (or paying attention to the whereabouts of) the human all the time.
Let’s think about what the dog learns and how, though. Here is a true story from a friend, in her words:
Years ago I had my little sensitive Sheltie down at the oval. She was off leash and running around sniffing the ground, hot on the trail of something or other. Not having much experience at the time I called her to come however, she was that intent on sniffing that she was oblivious to my recall. A Trainer told me to go and hide berhind a car so that she could not see me. I did as I was asked and after hiding for a few minutes Elsa finally realised that I was gone. I will never forget the terrified look on her little face as she ran up to everyone checking them out to see if any of them were me. Broke my heart. I would never put my dog through that again. She obviously was not well trained at the time and should not even have been off leash in that environment. I had set my dear little friend up to fail.
This method to me seems quite extreme, especially since teaching a dog to check in frequently and come when called is not rocket science. If you have built a great foundation and relationship with the dog, you would have a very hard time even doing this test. I personally don’t want to teach, “Stay with me or else.” It’s using coercion as an unpleasant topping on a really nice cake.
Version #3: Running Away
Now the cool thing, and I think the point Eric Brad was getting to, is that you can also use running away as part of a positive reinforcement protocol. My original title for this piece was about the continuum, and how the same actions, running away and hiding, can range from extremely rewarding to completely aversive. Here is the other end of the spectrum from what I described above. Take a look at these puppies practicing running after and with their people.
This is a much happier scenario:
- Antecedent: Human runs ahead of puppy (optionally calling to them)
- Behavior: Puppy chases and catches up
- Consequence: Treat and/or play is added
- Prediction: Pup running after human increases
And you can see it increase in the movie, can’t you? By the end of their turn, many of the puppies can hardly be left behind anymore and are happily running with their humans.
So what’s different? First, the humans didn’t hide. They just ran a few feet away from the puppy in wide open spaces. They didn’t lump straight to disappearing. Second, the puppies got a fun treat when they caught up. Third, they got to chase something, a favored activity for many dogs!
Now let’s compare the three different behavioral analyses. In Version #1, the human running away and hiding is the consequence, not the antecedent. It is performed as a result of something the dog did (or didn’t do). It is not an antecedent, or cue for a behavior.
In Version #2 it is an aversive antecedent. The human hides without warning from a young, dependent animal. The puppy is prompted to relieve the stress of being left alone by finding the human.
But in Version #3, the running away scenario, it is a non-stressful or very low stress antecedent. There might be a couple of moments early on where the puppy is going “Huh? where did he go?” But the humans stay close and out in plain sight and the pup learns the game. The difference was that they started out with baby steps. And at that point in the game they called the pups after they were coming to help teach them the cue.
And that leads us to…
Version #4 Hide and Seek
So if hiding is aversive, as described in Version #2, how come some people play it as a game?
The hiding as described in Versions 1 and 2, if used as a training technique, is lumping. In some cases deliberate lumping. The pup has not learned a fun game based on trying to find mom or dad. Suddenly disappearing on a puppy is quite different from doing a careful buildup. The people who do play hide and seek as part of a recall game take it in steps, just like with any other good training. They start off like the people in the puppy video I linked to above. They take care not to push too far and lose the pup’s trust.
Here’s a a short post and video by Mary Hunter, playing hide and seek with her parents’ dog Ginger. Ginger clearly thinks this is fun. Notice that Mary barely even goes out of sight before she calls Ginger. She is demonstrating a great way to start. Building very carefully on this foundation will probably result in a dog who associates looking for and finding her human with some of the best fun ever.
I will mention that I failed to make hide and seek fun for two of my three dogs when I tried it. It’s easy to go too fast, or fail to notice signs of stress. Clara, the formerly feral dog, and Zani, the sensitive one, both got a little stressed out when I tried this game. Summer was fine and had quite the drive for it. My failure with the others may be solely a result of lumping on my part, or could be connected with their temperaments. But I shudder to think of the effect on them if I suddenly disappeared on them in real life as some kind of test.
Version #5: Abandonment Training
I want to mention that there is a formal training protocol for reactive/aggressive dogs that fits into the “runaway” category. It is Trish King of Canine Behavior Associates’ Abandonment Training for Aggression. And although it is not something I would choose to do because of where it lands on the Humane Hierarchy (negative punishment and arguably positive punishment), it addresses the problems I noted in Version #1 above.
It is a protocol for when the dog performs aggressive behavior. It consists of the handler throwing the leash at the dog as a tactile cue and leaving. (There is also a long line on the dog.) But there are some preparations set for the exercise.
Trish King knows her learning theory, so this method is superior to the “just up and leave” method in terms of coherence to the dog. First, the dog is taught a cue that means the handler is leaving, and is positively reinforced for coming along. (That cue includes the tactile experience of having a leash dropped on its back.) So the dog develops some fluency, in neutral situations, of turning and leaving with its handler on cue. Second, a specific behavior (or set of behaviors) is the target for the punishment. In this case, it is barking and other aggressive behaviors, not a “non-behavior” such as not coming when called). Third, the cue for leaving is delivered exactly when the dog is performing the undesired behavior, so the relationship between the two is clear.
Good Methods for Teaching Attention and Recall
Except for the puppy recall video, it appears that most of the above is about what not to do. So here are some more resources for kindly and fun methods for training recalls, and a couple of inspirational vids as well.
- Teach Any Puppy to Come When Called
- Donna Hill’s series on Shaping a Recall
- Dog Training Bacon Recall
- Romeo Day 4 Recall Practice (includes collar grab)
- Fun Recall Games to Play with Your Dog (more advanced)
And for some inspiration:
Note: In case you regularly check out what is coming up, this post is the one that was formerly referred to as, “OMG Could she really be talking about the Continuum again?”
**In Version #2, there could be some positive reinforcement as well: adding the human back into the environment could cause joy as well as relief. Whether the aversive is present depends on the internal responses of the animal, but that’s not uncommon. We can’t “see” if a shock collar causes pain, but we can tell whether it can be used to drive behavior or not. And in both cases we can study the visible behavioral responses and body language of the dog. A third party, present when the puppy was left, could tell pretty easily in most cases whether the puppy was panicked or having a great time searching.