The traditional solution to shopping during pile work is using a long line, as mentioned in Mike Lardy’s video and in an earlier post in this journal. But Lightning has been developing a different, though possibly related, behavior: overrunning the pile, then picking up a bumper on the way back toward the handler, sometimes with a little shopping.

Besides being a bit annoying, I could see this hurting in competition if two dogs ran straight to the mark during their turns, and one picked it up cleanly while the other first overran it. Unfortunately, it seemed to be self-reinforcing: Lightning was doing more and more.

I have various theories about why he was doing it, but that’s not the behaviorist way. I just wanted to change the behavior.

A long line might have worked, but stopping the dog on a dime felt a little risky to the dog’s well-being. Besides, it wasn’t just that I wanted Lightning not to overrun the pile; I wanted him to mentally select an article as he approached and then grab it the instant he arrived.

The solution I decided on was pure positive reinforcement and closely akin to clicker training: I would reinforce the desired response with a marker for which Lightning had a positive association, then follow-up with a primary reinforcer. But I would withhold reinforcement for an incorrect response, so that those responses would eventually extinguish. Clicker trainers use a clicker as the marker, also known as a bridge or secondary reinforcer. I would use a gunshot from a blank pistol.

Here was the actual training plan:

1. With Lightning at heel, toss a bumper a few feet in front of him and cue Fetch. He already has a correct trained response to Fetch, so he picks the bumper up as soon as he can. But rather than me just taking it from him, I fire the pistol, then take the bumper and chuck it away for him to excitedly chase.
2. Repeat, but cue Back instead.
3. Continue to repeat, but begin adding more and more distance. Every time Lightning picks the bumper as he reaches it, I fire the pistol simultaneously with the pickup, then call Here and grab the bumper to throw it for him as quickly as I can. But whenever Lightning overruns the bumper, I instantly call out No, Here (our no-reward marker), calling him back to me without a bumper. If he picks up a bumper anyway, I meet him half-way, gently take the bumper, and toss it back to the pile.
4. Once Lightning is solid with a single bumper, I start back at a short distance and repeat the same sequence, but this time with a pile of three bumpers (near one another but not touching).

The idea that the moment Lightning grabs a bumper, a gun goes off and an instant later he gets to chase a thrown bumper, but none of that happens if he runs past the bumper, provides the kind of clarity that makes positive reinforcement so powerful. Lightning was retrieving cleanly from a pile of three bumpers at 20y in a single session. In the next session or two, we’ll extend the work out to 50y, and then resume our pile work training sequence, hopefully with this problem put to bed.