I just read this article. To sum up, it explains why it's important to reward both effort-success and results in narcotics dogs (and by extension human children). I say effort-success because they're rewarding the dog doing the search properly, even though it didn't find narcotics (result). The dog's not doing a sloppy search (just effort).
But anyway, my whole point here was to do a service dog take on the article without all the jumbly language. It made my brain hurt a little.
My example of a task where it's important to reward both effort-success and results is pushing the handicap button in order to make a door open. I actually know of a handler whose dog got (momentarily) bummed out because he pressed the button the way he was supposed to, and the door didn't open. The button was broken—the dog had done his job right, but since the result (door opening) hadn't happened, he thought he'd failed.
The way I teach my dogs to hit the button is, first, by seeking out a button that will not open the door. A closed shop works well because I can tell if they've hit the button squarely by the little clicking the door makes as it tries to open against the lock. (As opposed to a malfunctioning button.) Using the clicker, I go from rewarding just a touch of the button all the way up to hitting the button firmly and squarely. I don't really care if they use their paw or nose, although I am glad that Kaline has switched to using his nose since he got tall.
So the dogs start out learning that they're being rewarded for hitting the button accurately. At this point, the door actually opening hasn't entered into it.
When we move on to open stores/working buttons, the reward comes for both hitting the button accurately and for the door opening. This means if they hit the button right and the door doesn't open, they get rewarded. If they hit the button a bit sloppily and the door opens, they get rewarded. Hitting the button accurately is the “effort-success,” the dog doing the task correctly. The door opening is the "result," the thing I want to happen.
I think I've probably oversimplified a little, but the basic gist of this is that with service dogs or narcotic detection dogs or kids doing math problems, it's important to reward both the effort-success and the result. If you only reward effort, you're eventually going to get sloppy work because the result doesn't matter. If you only reward the result, you'll kill your student's motivation, because even if they do everything you taught them correctly, they may still "fail" if that doesn't produce the result.
When you reward both, you get a dog who is motivated to hit that button squarely, or do a thorough search. You basically vaccinate the dog against depression over “failure.” And I think that's a pretty cool thing!