One of the projects I've been working on with my own dogs is about what trainers call "cue transfer" and what behavior analysts call a "delayed-cue" procedure – getting them to perform a known behavior with a brand new cue. The lore goes: if you present a new cue, followed by the old/existing cue for the behavior, your dog will start to anticipate the old cue and will perform the behavior when you give the new cue.
But ... it doesn't always work. Or sometimes, it works, but it takes a really long time to get the new cue reliable.
So, I've been trying to tease it apart, and see what's actually going on there. Here's a little clip of Cheaters and Rosie working on it with me.
Rosie's new cue is a hand signal for down ("stop" hand, raised above my shoulder). This was her second session with the new cue.
And, Cheaters' new cue is the word "go," which means he should run straight ahead of me. Again, this was his second session with the new cue.
I think there's probably a lot going on here – and not just one "rule" about how to transfer a cue. I'm excited to keep playing with it and see where this goes!
One of the things we're talking about in the graduate PORTL class I TA for at UNT is back chaining. Specifically, we're talking about how we can use back chaining to build a reinforcement system. The reinforcement system teaches our learners:
* where and how we will deliver the reinforcer
* when we will deliver it
* which behaviors they need to do to get and consume the reinforcer
Once the learner understands this chain, we can extend the duration of behaviors within the chain, without frustrating our learner. They know what is coming next! In other words, there is no guessing or worrying about whether the behavior they are performing will lead to reinforcement. Our learner knows the next step in the chain, and the step after that, and the step after that ... all the way up to delivery of the reinforcer.
Here's an example of Rosie and me practicing a reinforcement system with her ball. After back chaining the reinforcement system, I add duration to Rosie's down. Then, I add another task to our chain: touching the target stick.
Notice how Rosie doesn't ever "throw" behaviors at me or guess about what she should do, even when there is a pretty lengthy pause between my reinforcement signal ("catch!") and delivery of the reinforcer. She knows exactly what to do and when to do it, and she is confident reinforcement is on its way!
What makes for a great social interaction?
Take a look at this play session between my teenage border collie, Rosie, and her new friend, Cappy.
In this video, Rosie is pretty pushy! She keeps pestering Cappy to do what she wants to do — play! Little baby puppy Cappy responds by asking for lots of breaks. He turns his head away from Rosie, flicks his tongue at her, and shakes off. But Rosie was still able to talk him into a pretty long play session!
How?
By acknowledging and reciprocating Cappy's signals. Every time Cappy asks for a little break, Rosie mirrors him.
Social interactions – whether they are puppy play sessions or training sessions with our dogs – are a back-and-forth conversation, an interconnected chain of behaviors between two individuals. They can be a series of breaks away or a series of follows. The more one team member tries to push their agenda, the more the other team member is going to withdraw or avoid the other team member – they will break from the session.
Conversely, the more a team member acknowledges, reciprocates, or "follows" the other team member, the more likely the other team member is to rejoin them — and the more likely everyone is to have a fun, mutually enjoyable interaction!
There are some really interesting conversations happening in the dog world right now about what "consent" means (or "assent," as we would refer to it in behavior analysis).
First, what qualifies as assent/consent? Does the dog need to be happy about it? Or does the dog just need to tolerate it? Aren't there some things a dog needs to have done to it that it will never be happy about?
And, second, is it even OK for us to ask dogs to do things that are 1) not critical to the dog's well-being and 2) not ethologically "natural" for the dog? In other words, even if we *could* train a dog to consent/assent to these things, should we?
These are really big questions, and there's a whole lot to unpack. I don't have answers, but I do have some thoughts, which could probably best be summed up by this video of my dog running and jumping into my arms.
Some background on the video: When I got Cheaters as a puppy, being picked up and held was very, very aversive for him. He growled, snapped and bit when he was picked up, and he avoided or moved away from anyone who reached toward him.
We could say, it's not natural for dogs to be picked up or carried, and it's not usually necessary for their well-being. For times when it is necessary, we could say that tolerating being picked up and held is "good enough" – that the dog doesn't need to be happy about it.
But. That's not the approach I took.
And, Cheaters now actively seeks out being picked up and acts downright joyous when I hold him. I believe his quality of life is greatly improved from when he was a puppy.
Again – a lot to unpack.
One of the ways my friend and mentor, Mary Hunter, taught me to think about building duration is with a chain, rather than with extinction (just waiting longer and longer or for more and more, before reinforcing).
The other thing I think about is that our learners are never doing *nothing,* which is sometimes the way we tend to think about duration behaviors – "just stay there and do nothing while I count to five."
I've been playing around with extending the distance and duration of a Frisbee game with Cheaters and Rosie. I'd like to be able to send them over a jump, then through a tunnel, and then over another jump. But if I tried to extend the duration/distance between the obstacles by moving them incrementally farther and farther apart, I'd probably end up with a whole lot of frustration and confusion and errors.
Instead, I'm extending duration by giving them something *to do* at each step, and chaining those small steps together.
So I can start out with two jumps, close together, on each side of the tunnel. Send them over one jump and there's the next jump and there's the tunnel and there's the next jump and there's the next jump and there's your Frisbee.
Then, I can either backchain further by adding another jump before the tunnel, or I can take out some of those in-between steps. In other words, I can remove the jumps closest to the tunnel. Either way, I'm increasing the distance/duration between the obstacles, and my dog is not engaging in any "junk" behaviors between them.
I end up with nice clean behaviors and happy dogs.
Rosie's reaction to waves
Good training sessions often are about as exciting as watching paint dry; they look like a whole lot of nothing at all.
Sometimes this can feel like we aren't *working on the problem.*
But the best way to work on the problem is almost never by continuing to expose the dog its triggers. Instead, we can identify how a *perfect* dog would behave in any given situation, and train those skills independently of the triggers.
This is true for all types of problem behaviors, including "obsessive" or "neurotic" behaviors.
https://sarahowendogtraining.com/blog/ocd-and-neurotic-behaviors/
Yesterday, Rosie and I started to work on a little baby right finish, where she approaches me from the front, wraps around my right side, and ends by my left side. Because cues get attached to the behavior from the start (whether we intentionally give cues or not – they are looking for predictive signals), we might as well train the cue at the same time as the behavior. As long as we're using an errorless approach and minimizing the chances that the cue will get associated with the wrong behavior, it will save us the step of transferring the cue later on. I'd like her right finish cue to be my sweeping right hand.
Tonight, we worked on teaching her that all the other signals in the environment are irrelevant. So: standing with my right foot forward instead of my left foot (feet are not a good predictor of a right finish), using a different toy (the activity we're engaged in is not a good predictor of a right finish), playing at a different time of day ... even my other dogs are helping her pick out the relevant signals from background noise.
Putting it all together in a Frisbee sandwich. This is Rosie's first time working on a dumbbell retrieve. She has many of the component pieces for a formal retrieve at this point – run toward the thing I threw, pick up the thing I threw, return straight back to me with the thing in your mouth, and deliver it to my hand – so this went quickly and easily. Except for the look of utter disdain on her face when she realized it was a dumbbell I threw and not her Frisbee 😂 Don't worry, Rosie, we'll keep doing Frisbee sandwiches for a long, long time.
Rosie's loose-leash walking around distractions – auditory and visual stimuli, including a friend – is looking pretty good! She looks about the same with our helper in the ring as she looks without her there.
The goal is that all the distractions become nothing more than low-key background noise, and that I don't have to feed her treats constantly or fight to be more interesting than all the exciting things in the world.
That means we had to start in an easy environment and gradually introduce distractions as she was ready, without ever pushing her over her threshold. We're making good progress!
