It’s easy for any training method to be misused. In the case of positive reinforcement training, we start to lose good results when the training becomes more about the treat than the behavior.
I’m working with an Alaskan Malamute at the moment named “Klondike”. He’s a juvenile (20 month) intact male with a behavior complaint of excessive jumping, on people, which quickly escalates to ever more aggressive posturing. He has scared a handful of animal people (neighbor, farrier, vet) to the point that some have mentioned he might need to be put down.
Klondike has been put on prozac by another DVM and the owners were taught to train with treats–ad nauseum–for good behaviors. After I started working with the dog, the owners said that I’m the first person from whom they’ve heard anything about phasing out the treats, or about rewarding the dog with attention/petting instead of (or in addition to) the food rewards.
The wife mentioned that she didn’t like all the emphasis that had been placed on giving treats, and I asked why. “It’s like giving a sticker to every kid in kindergarten. Everyone just gets told how good they are all the time and they don’t even have to do anything!” she quipped. I would paraphrase her objection as being a stand against grade inflation: use rewards injudiciously, and pretty soon we have an animal (child) who expects to be rewarded for little effort. The animal (child) doesn’t even really know why they’re getting rewarded, they just learn to expect a reward whenever the trainer/owner (teacher/parent) is around.
This is all very well when we are working with a fearful subject, animal or child, and need some way to establish a pleasant association with our presence.
But if any real training is to be done, rewarding with food treats needs to be done precisely.
Let’s look at the flip-side of this: If an animal (child) doesn’t understand why the rewards show up, it follows that they’ll be equally unaware of why the rewards stop. The animal’s (child’s) behavior begins to be shaped by an addiction to the possibility of a reward, instead of being shaped by understanding that certain behaviors make life more rewarding.
That is positive reinforcement gone bad. Instead of a cooperative, pleasant, or useful partner, we have a spoiled maniac obsessed with getting their next fix, and frustrated when that fix doesn’t show up. I see this every time I see an owner dangle a treat in front of their pet as they ask for a behavior; to be honest, it kind of makes me nauseous. Watching an addict in action, animal or human, is no fun.
We’re operating at a neuro-chemical level when we use food treats–it’s fundamental evolutionary biology which dictates that animals are hard-wired to want food. Once animals associate a certain behavior with increasing the likelihood of food, they will reliably perform that behavior. But it’s not easy for a trainer to deliver food at the right moment in time. We have to dig for food in our pockets, pick up a small piece, and hand it over. During all that time (we’re talking seconds) the animal has probably changed its behavior. Maybe it was doing what you wanted when you started to reach for the treat–in Klondike’s case, let’s say he was walking near your left side, at “heel” position. By the time you get the treat from your pocket, Klondike has wandered a little ahead or behind, or off to the side, or found a squirrel to chase. The training moment is lost, but you now have a treat ready, which Klondike smells, and starts begging for. Klondike will try a host of behaviors in order to get the treat, but Klondike has no association between what he did a few seconds ago that you actually liked (walk at your left side) and having the treat show up.
Unless we can precisely pair a desired behavior (heel) with a reward (food) then we’re not training. We’re just casually carrying around one of the most powerful motivators known to animal-kind. The bottom line is that food is a very, very powerful motivator. As with any power tool, food tends to take over the project unless it is used with great precision.
Great precision is really hard to master, and I think this is why training with food can get a bad wrap. The professionals working with Klondike before me were not wrong to introduce food to Klondike’s behavior modification. But they were misunderstood. They were explaining how to use food to diffuse a tense situation (Klondike around new people). That is very different from using food to reinforce specific behaviors (like teaching “heel”.)
Stepping in to say that they could use food less often, or use attention instead of food as the reward, was my way of saying, “Let’s not have food around regardless, while you’re teaching a specific behavior. Let’s teach Klondike that what he does can make the food show up.”