That is, how much of the training the dog gets completely right versus makes an error.
Scientists call this perfect learning balance the “sweet spot” and find that it applies to all mammals learning things like a new language, instrument or, you guessed it, Schutzhund! Whether it’s a new helper learning to teach guarding or your dog learning dumbbells, the ratio of success to failure should roughly be the same for optimal learning.
That ratio is 85% easy reps to 15% difficult reps and is referred to as
"The Eighty Five Percent Rule".
What defines easy or difficult?
It depends on where the individual is in their learning, but easy reps would be characterized by getting the “right” answer and getting rewarded as a result. Difficult reps would be characterized by the dog making an error in the process. We can use shaping a dumbbell hold as an example. In the early stages of teaching it, (by the 2nd or 3rd session) you're rewarding as soon as the dog touches the dowel, which happens constantly. For only around 15 percent of the time, you're withholding the reward in an attempt to get mouthing, the harder modification. If your ratio for not rewarding the touches dropped, the dog may stop trying altogether, slowing your progress significantly. So we only ask for a little more difficulty (mouthing) a small percentage of the time, making the dog successful the vast majority of the time.
One of the biggest failures we see this in is heeling, both in obedience and especially protection. 85 percent of the time, the reward should be given before the dog makes any mistakes, such as looking away. A small portion can be set up for slightly more difficulty repetitions (of course rewarding if the dog nails it), advancing the dog’s level and pushing him forward in the training. The mistake we often see, however is a much lower success rate; handlers pushing the dog to do difficult reps on half or even the majority of them. This leads to the dog constantly failing, either being corrected, not being rewarded, or both. Science tells us that learning occurs much slower when this ration is so far out of the “sweet spot”, leading to a drop in overall enthusiasm for the training and slower understanding of what's being taught.
And really, do you enjoy learning something if you’re wrong half the time, particularly if there are consequences? Or if you’re right or win 100% of the time (which is not ideal either)?
The reality is, we want to feel like we’re winning, yet still challenged. The challenge or difficulty keeps us trying, the winning or success keeps us enjoying the process.
And when’s the last time you met a kid who hates school yet excels scholastically? A good attitude towards learning allows a larger amount of dogs to shine. So while the best dogs can hold up to training that is made more about failure than success, getting the ratio right with a dog who is not a genetic super star can mean either proudly enjoying your nationals or never going further than an IGP 1.
So how do you keep your training at the optimal ratio for difficult versus easy?
You set up your repetitions purposefully, not haphazardly. You have a plan (albeit a fluid one that can change immediately when needed) and that plan is designed to make the dog successful most of the time and POSSIBLY fail every once in awhile.
Here’s a scenario: I’m just starting to heel for the first times with my 1.5 year old dog in protection (secondary). Without the helper, we are comfortable doing long stretches of heeling here and there. We’ve also done short stretches with a big wedge on the ground.
For my first attempt at getting heeling in protection, how much am I going to ask for before rewarding my dog?
If you said 1 second of eye contact in basic position with no movement into the heeling, you are correct. I need to reach my 85 percent success rate so I will start with something I’m fairly certain I can get quickly and reliably. Now I need to set up one or maybe two reps (out of ten or twelve) that MIGHT lead to failure.
What might that look like?
It needs to be something where, if the dog fails, I can immediately do the next rep at the easier modification. My difficult rep would be asking for a few more seconds of simple eye contact as I give the bridging marker “good” to help. Depending on how the dog does at the difficult reps, I can alter my next ones. With each success, you can often continue to push forward towards more difficulty. However, once that line is crossed, go back to the area of near 100 percent certainty the dog will succeed. This not only keeps the dog in good spirits, renewing their confidence, BUT also reminds them of a version of the behavior that gets rewarded. Too many failures not only lowers the dog emotionally, but also increases the chances the dog has forgotten what the correct solution is.
This is particularly important to remember if the dog is just learning or has taken a long break.
Evaluate your difficulty reps with more scutiny.
When I asked for those couple more seconds, did he stare demandingly without even thinking of looking away? Was he looking at me, but I felt it was weak and not committed? Or did he completely bomb and drop his head immediately? What the dog does dictates your next move in finding the sweet spot. The more practice you have as a dog trainer asking for complex behaviors, the more this constant set up will come instinctively, and you’ll won't have to think about it too hard.
It often comes intuitively for pet dog trainers who while maybe not aware of it, have learned this while working with pet dogs, dogs not as forgiving of training errors as a high quality working dogs.
Working with dogs who make you hyper aware of the mistakes you make (like a pet dog or lower quality working dog) can elevate your training skills in a way not possible when only working with very high quality and forgiving dogs. When experimenting with the sweet spot, try working with your friend’s dog, shelter dogs or dogs unfit for work. These dogs will make you perfect your skills much faster than a “good” dog”. The skills you learn there will allow you to train the good dogs better, faster and with more clear communication.
Don’t take our word for it, check out the study posted below :)
https://www.nature.com/articles/s41467-019-12552-4
Comments