Earlier today I was working on Rugby’s hold portion of the retrieve behavior, specifically looking to build a strong hold on the glove. He’s got very terrier tendencies (as expected), and has a high frequency of thrashing and shredding soft materials. So I anticipated that I’d need to build this behavior with careful increments. Even more careful than usual.
As part of my shaping plan, I wanted to make glove thrashing/shredding behavior less likely, so I altered the glove to make it stiffer… by rubber banding it to a stick. Once I have the hold the way I want it, I can incrementally increase the floppiness of the glove to get the finished product.
Because the hardest part about getting a reliable hold in the behavior chain is that the dog must come flying in (based on my ideal performance), which is very exciting and arousing, and then instant hold a control behavior with duration… physically and mentally challenging. It has a lot in common with stopped contact behavior in agility. The dog has to switch states very quickly from movement to stillness.
So I wanted to set him up for success by giving him a progression that built on the most critical skill… controlled stillness. I started by reinforcing eye contact a few times because that’s part of our structured control game that we play… it’s familiar to him. Then I added a chin rest, again a familiar structured control game. I added movement by tossing the treat out so that he returned to my hand with some movement, and reinforced him for chinning with control. Basically the pattern I want for the retrieve.
When that was going well, I swapped the glove/stick for the chin. Now, he’s done this game many times with sticks and dowels and dumbbells. So I wasn’t starting from scratch at all. The only new part for him was the fabric wrapped around the stick. A pretty decent increase of only one variable. And it went well! But that’s not what this post is about!
What this post is about is THIS:
After a short break, I brought him back out and this time went straight for the glove/stick. Maybe not my best choice, but not awful. But then I clicked a bit late a few times and got a little mouthy/regripping behavior I didn’t like. So I removed and re-presented the object to get a cleaner behavior. And then I did it again. And again. He did give me a better grip, which I reinforced. But I think this is where the two-way street of reinforcement ran me over. The next time he regripped in a way I didn’t like, I did the remove/re-present thing again… because I had been reinforced just before. And then I did it again. And again. The errors started mounting up, and my rate of reinforcement plummeted.
And guess what happened? He started to sniff a bit after he ate the treat on the ground. Then he started to sniff a bit more. He’d look like he was hunting a bit for another treat or a crumb, rather than scarfing the treat and snapping back to repeat the behavior. Fortunately I noticed fairly quickly for me and stopped digging the hole.
Was he really hunting for treats all of a sudden? Distractions? Coincidence? Same treats, same floor as the previous session and even earlier in the same session… what was different? The training.
I changed the criteria and was off with my timing, making his path to reinforcement unclear, and the result was frustration and thus displacement behavior.
Oops… my bad.
Take a look at the two training sessions and compare. The first is the good one, with a clear progression and criteria. The second is the not ideal one.