Proofing Behaviors in Dog Training

BONUS: I’ve uploaded a PDF printable for proofing behavior in the AGILITY MASTERCLASS. 

The golden rule of dog training: one repetition does not make a behavior. Just because the dog does it once, doesn’t mean they will do it again. And even if they do it again, under what circumstances? When I am training my dog to do a behavior, in a perfect world, I want my dog to be able to do the behavior the first time I ask, every time I ask, with any distraction, in any environment. Creating a high level of fluency in a behavior – so that it is consistently successful no matter what – takes a lot of time and effort on the part of the trainer. After training the initial training the behavior, or even during the initial training, proofing plays a large role in creating this fluency. When I think about proofing a behavior, the phrase I think about is, “Can they do it if _______?”

Initial training & Success Percentages

When I am initially training the behavior, I am thinking about each training session in terms of the success percentage. The success percentage is defined by the number of reinforced responses over the total responses x100. So if they do 10 behaviors and 5 of them meet the criteria and are rewarded, they have a success percentage of 50%. My goal is to get the behavior to 80% success in two training sessions before making it more difficult. When we talk about proofing, we are making the circumstances harder for the dog to do the behavior. If my success percentage is low, and I then make it more difficult, it’s going to be much harder to meet my training goals. 

Proofing

Once I have the behavior I want with a 80% or better success percentage, I’m going to start proofing or adding challenges to test my dog's understanding of a behavior. These challenges usually come in the form of a distraction. That could be a small distraction, like if I ask my dog to sit and take a step away or turn in a circle. More advanced level proofing would be if I asked my dog to recall, but while the dog is running at full speed with other dogs. 

When I am proofing a behavior, I am intentionally adding/changing a circumstance that has a small likelihood of creating an error. For example, if I asked my dog to sit (behavior) and took a small, slow step back (small distraction/change), at the start there may be a chance the dog gets up from the sit (an error) or they hold their sit. In this training session I want to think about these conditional statements:

When the dog does (insert the behavior), if I add/change (distraction/circumstance) and they meet the criteria, then they get a reward. 

When the dog does (insert the behavior), if I add/change (distraction/circumstance) and they do NOT meet the criteria, then they get nothing and we try again.

Or more simplified: 

Behavior + Distraction → Correct Performance → Reward

Behavior + Distraction → Incorrect Performance →  No Reward

So the dog learns that they only get the reward if they are able to do the behavior in the presence of the distraction. 

Troubleshooting

Most of the challenges in proofing involve choosing the level of distraction or what to change. Increasing the difficulty of the challenge too quickly can lead to a rapid decline in the dog’s success, and from experience, increase frustration of the handler. For troubleshooting, I use a couple different approaches. The first is preventative: my number one intervention is making sure we start with lots of small, low level distractions that we can add to the behavior. The more we build the easy stuff, the easier it will be to add a more moderate level of distraction. Next, if I find difficulty in advancing, I think about “splitting the difference” between the last successful criteria and what we are trying to do. So if in that sit taking one step back has 85% success, but taking 5 steps back has 35% success, this is a big jump. There’s a chance that through enough repetition and rewards, we could get back up to 80+%, or alternatively, we could split the difference, and try 3 steps instead. When advancing the criteria, we are intentionally adding the opportunity for them to make a mistake. Remember that mistakes are okay – they are part of learning and so long as we are in a controlled environment, which we are creating, we are doing so with a reasonable amount of safety.** When we add the difficulty, as with any criteria, we may see a small drop in success and this is normal – we just want to maintain a high enough rate of reinforcement to grow it back into the 80 – 100% range. So if I’m concerned that my dogs aren’t successful enough for the distraction, I want to take a half to a whole step back in the level of criteria. 

BONUS: I’ve uploaded a PDF printable for proofing behavior in the AGILITY MASTERCLASS. 

in my dogs so I can keep track of some of the steps I’m doing to proof my behaviors. This way I’m being intentional and not just relying on my memory. 

**Safety note:
Control of your environment and circumstances is pivotal in proofing more advanced behaviors. I am NEVER going to go from low-level distraction recalls to trying to recall my dog off of a rabbit racing through the yard. I don’t want to leave room for error for my dog in a situation where their safety or the safety of others is at risk. In proofing my recalls, I might use strategies such as advancing to recalling off of the scent of wildlife (rabbit poop!), can my dog do an off leash recall around other crated animals in a contained field, eventually can they restrained recall or short off leash or long line recalls when I would be willing to bet $1,000 that they would be successful every time. Eventually, one of my most relatable (and advanced) levels of distraction is being able to recall them when they are running hard with other dogs at a distance. In these layers, I’m only moving up with extreme confidence in my dog’s abilities and still having controls in check. So for example in the last example, if I recall off the other dogs and they do not – can I ask those experienced dogs to lay down (and they do) which will make it less reinforcing to keep playing with them and easier to be successful on the second attempts.