Are you putting the cart before the horse?
Positive reinforcement training is not a synonym for “using treats”.
I do use food liberally in my training and encourage my clients to do the same: food can be a primary reinforcer, meaning it has inherent value because dogs need it to survive. And there is tons of value in non-contingent feeding (where food is given without being a consequence of a particular behavior). Food is a love language. This post is not to suggest that there is anything wrong with using food in your training.
But just because something is an appetitive stimulus (good thing) doesn’t mean it is a positive reinforcer. Whether we class a stimulus as a reinforcer or a punisher is determined by its effect on future behavior. And what happens after a behavior is what determines if that behavior is going to happen again. The order of operations is important. Behavior is shaped by consequences, just as learning is defined by consequences.
When we use food in our training, it can only be labelled positive reinforcement if:
The delivery of the food is a consequence of the target behavior
The target behavior frequency increases
So if you’ve ever heard someone say that they “tried positive reinforcement but it didn’t work”, a more accurate translation of that would be “I tried giving my dog treats but it didn’t change their behavior”. Because positive reinforcement works, by definition! You just have to do it right.
Another lament I hear often is that their dog “only listens if I have treats”. This problem actually gets me kind of excited because it’s easy to fix and opens up a dialogue on the importance of mechanics. (Tell me you are a KPA graduate without telling me you are a KPA graduate…). It means we have to go back to the ABCs.
A is for antecedent: the conditions that exist before the behavior happens
B is for behavior: the action, described in clear observable terms
C is for consequence: the conditions elicited after the behavior happens
Antecedents set the stage and signal which contingencies are at play. These are your cues. Consequences shape and maintain behavior. These are your reinforcers. Careful not to mix those up.
In pretty much every “only listens if I have treats” case, the food is appearing before the behavior. These folks have been taught that in order to ask your dog to sit, you hold a treat over their head until their butt hits the ground. That’s all well and good if you’re skilled enough to fade a lure, but most people aren’t taught how to do this, so now you have the presence of food acting as a cue/antecedent instead of a reinforcer/consequence. Then they try asking the dog to sit without food in their hands, and the dog no longer understands what is being asked of them because the cue is totally different.
Many people will interpret this as the dog being stubborn, manipulative, spoiled. But what has this process actually taught the dog? I see food, then I sit, then they give me the food. The sitting behavior is contingent upon the presence of food.
I had a client whose dog loved to nap. If her humans asked her to come outside with them for a potty break, she would decline in favor of continuing to nap. They could sometimes get her to the top of the staircase with some treats. The dog would stall at the top of the steps until they tossed a big biscuit down the staircase. Then she would trot down, eat the biscuit, and have a quick pee. This went on for some time (without me knowing!). They were confused when the behavior of “going down the stairs to go potty” did not increase in frequency. In fact, the stair-descending behavior stopped altogether unless it was preceded by a flying biscuit.
In this case, the conditions for “going down the stairs” only existed in the context of “biscuit tossed down the stairs first”. If anything, the behavior of “stalling at the top of the stairs” may have been reinforced by the eventual treat toss. But food was not a consequence of the stair descending behavior. So we can say confidently that, in this case, the use of food was not reinforcing the stair descending behavior because it did not come as a consequence and that behavior did not increase in frequency.
I had another client who wanted to teach her dog to put his paws up onto trees and fire hydrants. I love teaching “paws up” behaviors but they beat me to it before we could discuss mechanics. They placed a treat on the surfaces they wanted their dog to perch on. One goal of this behavior is to be able to move your dog out of the sightline and get distance from triggers. Out on a walk, the human spotted a trigger, tried to get the dog to “paws up” on a ledge, but didn’t have the reflexes to grab a treat first. The dog did not understand the cue without the presence of food. He didn’t jump up, and then he was too close to the trigger and began to bark.
So what do we do instead? If you can’t get your dog to perform a behavior without luring them with food, it’s time to slice that behavior into more manageable pieces. In the stair example, we could click and treat any movement towards the stairs, any weight shift forward, one paw inching its way towards the next platform. If we mark and reinforce these approximations, we can begin to slowly raise our criteria, asking for a little more behavior, a few more steps, in order to elicit a consequence. Another option is to fade out your lure within ~3 repetitions, but I see this as a last resort. You can also hire a trainer to make a plan for you :-)
There are plenty more examples where these came from-- people holding out handfuls of treats in hopes of enticing their dog to return to them, undereducated trainers dropping treats onto a bed as they try to teach a “place” behavior, on and on and on. Let us remember that while everyone is capable of using positive reinforcement, it is still a technical skill that requires precision. Yes, do use food in your training. But remember to mind your ABCs!
