In “Slipper retrieve,” a post from four years ago, I introduced the concept of “discovery training.”

The idea is that some of the behaviors we want our dogs to learn are self-reinforcing, to the extent that an extrinsic reinforcer such as treats is not only unnecessary but may even detract from the natural discovery process of the dog learning the pleasure of the behavior itself. Examples of behaviors well suited to discover training are house-training, wherein the pleasure of the dog relieving himself is an intrinsic reinforcer for the act being performed outside, and wherein, for a dog with an innate instinct to retrieve, the act of retrieving is a powerful reinforcer for the act itself.

Because I dislike being on the receiving side or the sending side of coercion and manipulation in human interaction, and perhaps for other philosophical and psychological reasons, such as an attraction to simplicity and efficiency, the notion of discovery training appealed and I guess still appeals to me, and I believe I’ve often Incorporated it into my efforts at dog training.

But as I was rethinking the concept of discovery training recently, I suddenly had one of those Eureka! moments. Wait! Perhaps my use of discovery training with Lightning has caused or exacerbated his keep-away habits!

Well, it’s probably more complicated than that. For example, in “The Prisoner’s Dilemma, keep-away, and patience,” I described another mechanism that is also likely in play. And growing up with Ryley, DW Renée’s Golden Retriever, who is almost the same age as Lightning, may have also been an important factor. In TRT, Mike Lardy says something along the lines that he has never seen a successful field retriever who grew up with a sibling as a companion.

Nonetheless, discovery training means that the dog looks to himself for the rewards of his behavior. I would think that’s desirable in human development, but I have now begun to think it’s counterproductive in dog training.

My reasoning is this. Dog training is about engaging in the bargain that the dog gives you what you want, and in exchange you give him what he wants. For the dog, that takes place unconsciously, as the behavioral mechanism called operant conditioning. But that mechanism nonetheless loses its power to the extent that the dog doesn’t depend on you for reinforcement.

Instead, I now think that to maximize the effectiveness of operant conditioning, we want to maximize the give and take in our relationship with the dog. Thus treats and other extrinsic rewards that are available only from the trainer are preferable to intrinsic rewards for the vital goal of engendering cooperative interaction.

For Lightning, I think the most prominent example of this principle is his tendency to go into keep-away mode rather than complete a retrieve, a tendency I’ve been battling, and finally, gradually defeating, literally for years now. Keep-away is a game he can play without my cooperation. In fact, it seems likely to be more pleasurable, as a game for the dog, the more the behavior annoys and animates the other party. It may be a healthy game for dogs to play with one another, and in any case probably is inevitable in doggy interaction, but it is destructive when you’re training a dog for a team sports activity.

So this is the new plan: I won’t be reluctant to incorporate intrinsic reinforcement into my interaction with my dogs, but in the first place I won’t be shy to also use extrinsic reinforcers, and in the second place, I’ll make an effort to pair intrinsic reinforcement, when it occurs, with rewards that the dog associates with our relationship rather than just pleasure the dog obtains without my participation.