Extinguishing intrinsically rewarding behaviors?
What do you guys do to extinguish intrinsically rewarding behaviors that occur at a distance from you? For example, Lucy gets great pleasure out of licking the floor. So much so that typical P- techniques (e.g. timeouts) or R+ distractions (e.g. watch me, targeting, etc.) don't phase her at all. The reward she gets from licking the floor is enough to keep her coming back regardless of how I train against it.
Here's another example. Tyson has grown very attached to his training bumpers. He will happily run around the yard carrying one in his mouth. The good news is he doesn't chomp on it. The bad news is he has now gotten very sloppy on his retrieves, often detouring around to inspect stuff or taking a "victory lap" around the yard with his prized bumper (or frozen bird). The pleasure he gets out of that is enough that he doesn't care to make a straight retrieve and earn my praise/food reward. In this case, because Tyson's "infraction" occurs at a distance from me, I can't really use P- except to withold a treat when he finally delivers the bumper. But then I'm further disincentivising him from delivering the bumper (delivering gets no reward, but running around with the bumper is it's own reward). I've tried differential rewards with increasingly poor success (give one piece of kibble with no praise for a sloppy retrieve, multiple pieces with a lot of praise for a prompt and direct retrieve).
So, my questions is, staying in the R+/P- realm, how do you extinguish intrinsically rewarding behaviors that occur at a distance from you?
Here's another example. Tyson has grown very attached to his training bumpers. He will happily run around the yard carrying one in his mouth. The good news is he doesn't chomp on it. The bad news is he has now gotten very sloppy on his retrieves, often detouring around to inspect stuff or taking a "victory lap" around the yard with his prized bumper (or frozen bird). The pleasure he gets out of that is enough that he doesn't care to make a straight retrieve and earn my praise/food reward. In this case, because Tyson's "infraction" occurs at a distance from me, I can't really use P- except to withold a treat when he finally delivers the bumper. But then I'm further disincentivising him from delivering the bumper (delivering gets no reward, but running around with the bumper is it's own reward). I've tried differential rewards with increasingly poor success (give one piece of kibble with no praise for a sloppy retrieve, multiple pieces with a lot of praise for a prompt and direct retrieve).
So, my questions is, staying in the R+/P- realm, how do you extinguish intrinsically rewarding behaviors that occur at a distance from you?
Comments
For Lucy, there has got to be SOMETHING that is more rewarding than the floor - raw steak, raw fish, something... but Lucy has that allergy issue. So you really don't have all the tools at your disposal that you might with another dog. that, IMHO, is what really complicates the Lucy thing - I mean it really handicaps your training. For her, and I'm sure this would suck, but I'd just gate off an area she doesn't lick and keep her in there. Its for her own health, she needs to be restricted, that's the only way you can make that situation work.
For Tyson, I would not leave the bumpers out for him to play with. I would only pull them out when its time to work and if he is not working for you take them up and try again the next day.
For the times when he does do his work properly I would "jackpot" him - give him a really good treat, or lots of good treats for bringing the bumper right back you to... When he brings it back but with less enthusiasm or with a detour I would give him less treats, like just one bit of kibble. This tells him "thanks for playing, but that was not exactly what I wanted" .. and when he does it well you are telling him "oh yea, that was perfect, you rock!"
That's my $0.02.
----
Oh and agrees .....never leave working items out for free play. Only use the bumpers with your training and play work.
Snf
Jackpotting can work really well and really teach him what you want. Give him a really good treat and lots of it for bringing the bumper right back you. Remember, lots and lots of small pieces of the treat. Same principle that a kid would rather have 25 pennies vs 1 quarter.
Initially, having the bumper on a rope may help too.
And I also use a "uh uh" or "aht" to redirect them when they are doing something incorrect but it is a distance from me. For instance, when I wanted to teach both Sachi and Kuma how to down on recall, I wanted the down as soon as I gave the command, not 10 strides later and directly in front of me. So if they "ran thru" the command, I would go "aht" and stomp my foot to try and stop them, then if they downed, they got a click. If they didn't down and kept coming, I repeated the exercise. I found that they learned best with a combination of positive and negative (in the form an an "uh uh") reinforcement when they are doing long distance (from you) exercises.
Proud to report that both will down on command from a full blown run anywhere in the yard as soon as I give the command :-). And 95% of the time, I can then give a sit command and both will pop up from the down and sit, without moving toward me.
I brought this up partly because I had the opportunity to hunt over some full-time bird dogs (2 Brittneys, a GSP, and an English Pointer) and seeing the difference in their behaviors was eye opening. For those dogs, finding then pointing birds and retrieving are better than any food reward you could give them. The Brittneys didn't even care to chase bumpers into the pond to cool off, they went off in search of birds. It just highlighted the "uphill" battle that I'm facing taking a big game dog and training him to hunt birds. A battle I relish, but a battle nonetheless.
ETA: Thanks for the ideas Kris! I've sort of been coming to the conclusion that you need some very minor P+ at times to get the point across when you're working at a distance. Because the reinforcement rate is slower by necessity, you need some additional tools. I'm considering training a "go on" signal (technically a "conditioned tertiary reinforcer"), but I even see that as having some limitations.
Unfortunately, with carpet throughout the house, I would cave into the licking and settle on finding a special piece of safe carpet or fabric that she would prefer to lick over the regular carpet. (Maybe bait "her" carpet/fabric with some rubbed liver and introduce it in a closed crate?)