top of page

On Rewards

“when should I fade out treats (or other rewards)?”

What if the government cut you a check each month you don’t speed? Would it reduce the likelihood of you speeding? What if the check was $1 for every day you don’t speed? 30-31 days in a month. So long as you don’t speed at least one day a month, you’ll get a check.

I’d probably say “meh.”

What if it was $5?

I’d probably be quite a bit more vigilant about the speedometer. That’s a potential for $150 per month, after all. Depends on how big of a hurry I’m in.

What if you never get a check except on randomly selected months each year, but that check could range from $1,000 to $1M? Would you speed then?

In the $1 scenario, speeding every day I can beat traffic and save myself frustration, so it’s not worth it to me. In the $5 scenario, I’d be doing really well 29 days out of 30, but along comes a day wherein if I get somewhere ASAP I will be given a BRAND NEW CAAAARRRRR!!! (Hypothetically.) The convenience in the first case, and the free car in the second, are competing motivators.

Would you go for the free car if you might get $1M that month? Course not. ...You’d think really hard about it, at least.


Obedience should be reliable. And there should quickly come a day when you’re not feeding treats every single time your dog sits. Getting rid of rewards seems like a convenient and laudable goal! You can, after all, stay consistent with corrections and giving clear feedback for wrong choices (and you should, that’s a given). But you’ll likely end up with a dog who’s motivated by avoidance, and who is more susceptible to taking chances in the face of competing motivators.

However, if your dog’s obedience is in maintenance mode (meaning, the learning is over and you’re maintaining the obedience patterns and behaviors), and you employ an intermittent reward system that gets your dog gambling on the right choices, you’ll have a happy, enthusiastic dog with reliable obedience.

Find those things that range from $1k to $1M for your dog. Throw them randomly for right choices (frequency should be proportionate to value). See what happens.

55 views0 comments


bottom of page