Last week I wrote a blog about a well-researched concept; errorless learning. I am not going to reinvent the wheel, so if you haven’t read it, please do so, here.
It sparked both intense criticism and high praise. It also ignited some nastiness and straight-up bullying, but I don’t think any of that warrants a response here.
What I want to address instead is some of the criticism because much of it illuminated my miscommunication on a few pieces of the information I tried to convey in that piece of writing. So, here goes…
The Role of Frustration
Perhaps the biggest criticism of errorless learning procedures is that if our dogs don’t make mistakes in learning they will never learn to tolerate frustration. The argument is that frustration exists in life (and in sports), therefore it must exist in learning.
This is a belief that is unique to dog training and dog trainers, and whenever something is species-specific we would do best to examine it further.
So, do our sport dogs need a high tolerance for frustration? And should they be gaining this tolerance through our training of them? The answer is it depends. I think if you are going to be using primarily trial-and-error training procedures your dog does need a high frustration tolerance in learning scenarios because he will, quite simply, be frustrated often. If I choose to use primarily errorless training procedures I’d argue that no, my dogs do not require a high tolerance for frustration in learning. Why should they, if I am setting them up for success, not failure? And why wouldn’t I do so, when I am the one with the huge frontal lobe, after all?
The argument that dogs need frustration tolerance for life is valid, but has nothing to do with our training or their learning environment. Yes, we all need to tolerate frustration in our lives. This starts at puppyhood when pups fight multiple other squirming bodies to get to mama’s milk, and progresses later on through the endless frustrating events that occur in daily life. The cat swats at you when you try to play, the big dog runs faster than you in the field, and sometimes you will lose your toy to the abyss that is the space between the couch and the wall. Life is one endless cycle of frustration tolerance, learning need not be.
Some of the arguments for allowing frustration to play a large role in learning speak of “resilience” as a virtue that sort of training endows our dogs with. Yes, resilience and persistence are important for sport dogs. But again, I’d argue the learning environment is not where this needs to be acquired. Excellent breeders can follow the Puppy Culture model for puppy-raising to create resilient puppies, and if you are interested in dog sports there is no reason not to seek a breeder following this program. Later in life and throughout, we can provide our sport dogs with persistence-building puzzles that have nothing to do with training and everything to do with empowering our dogs with the knowledge that they can and do solve any problem they are faced with.
In this video the dog learns that she can conquer something that makes her a little unsure in order to access food she wants. I’d call that resilience.
Below, the same dog is able to push a noisy pan away and figure out how to overturn the plastic lids to get to the treats. She has to problem solve this by using her body (you can see her use her paw to find the answer), and she wins! Persistence paid off, and she won’t forget it.
And finally, a whippet conquering a more advanced puzzle that helped prepare her for many aspects of her performance career, all in a few seconds:
Errorless vs Less Errors
There is a difference between a teaching procedure that is errorless by design and a teaching procedure that involves a very high rate of reinforcement with a few errors thrown in.
All good training has a high rate of reinforcement with few errors, even trial-and-error procedures.
An errorless procedure is one that is designed to introduce difficulty or challenges at such incremental levels that the learner never gets the answer wrong. It is a procedure in which “pre-errors” are caught, and the trainer makes adjustments accordingly, and actual errors are met with an end of session and change of plans. (The end of session should always mean the dog gets to do something else fun, like a tug game, by the way). An error is not an expected part of the process; an error means the plan went awry.
In this example I am teaching Idgie scent articles with an errorless procedure I created. This is by no means a complete process; this is her third session. You can see that when she makes “pre-errors” or indicates that she even notices the cold articles I adjust accordingly. In three sessions she has made one error (retrieving the incorrect article). I stopped, let her play a different training game for a while, and reviewed my video to find out what went wrong. I identified the problem, fixed it, and moved forward.
The Culture of Dog Training
I mentioned in my last blog that the trial-and-error formula for dog training is widely accepted and that suggesting an errorless format will often be met with resistance. The criticism I met when I posted last week is proof of that statement. Yet, interestingly enough, there are quite a few often-used dog training procedures that are errorless by design! Here now, are two lists:
Generally Accepted Errorless Dog Training Procedures
- House training
- Any quality aggression or reactivity treatment protocol
- Recall training
- Any quality resource guarding treatment protocol
Training Procedures that CAN and SHOULD be Errorless (but are typically designed to include errors)
- Jumping
- Scent articles
- Weave poles
- Stays
- Loose leash walking
- Heeling
- (Fill in the blank…)
My point is that there are a few high-stakes situations (the four examples I used above are all high-stakes) in which errorless learning is considered paramount. Not only is it unpleasant when our puppies have house training errors, it slows our progress in that area, and we all know it! If my dog fails to recall when cued the error is likely reinforced (squirrel, anyone?) which definitely hurts my chances of having the flawless recall we all want. Of course anyone who works in the field of behavior modification for pet dogs knows that if we ever witness aggression (or, whatever behavior we are trying to change) we have screwed up–we pushed the dog to far, the trigger got too close, etc.
The question is, if errorless learning is the best choice for these high-stakes situations (lack of house training and aggression are two common reasons dogs are relinquished to shelters or euthanized, while a lack of recall can get your dog killed), is it not the best choice for our sport and pet training?
It depends on what kind of dog trainer you want to be. And that’s for us all to decide.
