Errorless Learning: How Could that be a Bad Thing?
By Deborah Jones, Ph.D.
In positive reinforcement training circles the concept of errorless learning is a very popular one. The basic idea is to set up your training sessions so that your dog has little to no chance to be wrong. This means that your rate of reinforcement is likely to be high, which is a good thing. It also means that your dog’s frustration level is likely to be low, which is also a good thing. So, where’s the downside?
For the longest time I couldn’t quite isolate why I wasn’t 100% on board with this idea. Everyone around me seemed to agree that it was a great approach. And on its face, it is. What could possibly be wrong with setting up your learner to be right so that you get lots of behavior to reinforce? It’s a great way to preserve your dog’s attitude and enthusiasm while developing new behaviors.
In order to accomplish errorless learning it is necessary to plan your training sessions carefully. You need to think about exactly how you will set up the environment so that success is highly likely at each stage of training. This means that you cannot go into a training session and just ‘wing it’. When you have to make a plan you are much more likely to be thoughtful about the entire process, which can only be helpful for your learner.
So far, I’m not making much of a case for the downside of errorless learning am I? But I do believe there are some serious issues with being too fixated on errorless learning. Do we learn from our mistakes? Of course we do! Is it always a soul crushing experience? Absolutely not! What good can come from discovering that errors don’t pay off? In a best case scenario you stop doing what doesn’t work pretty quickly, and switch to something that earns reinforcement.
A few errors here and there in training are NOT disastrous. Even with the best plans and execution they are likely to happen. In most cases they motivate the learner to try something else that works better instead. Working through errors can build up some resilience and resistance to extinction. They encourage the learner to be more thoughtful and persistent and creative.
I think there’s a real danger to never letting a dog make mistakes and learn from them. It can breed an attitude of learned helplessness; just giving up when efforts don’t work out immediately. That is NOT a quality I want to develop in my dogs. I want them to have the ability to withstand a lack of reinforcement without folding like a house of cards.
This doesn’t mean that I throw my dogs into the deep end during a training session and let them sink or swim by allowing error after error after error. But I also don’t think it’s a mortal sin if my dogs make a mistake and don’t get reinforced now and again. They will survive and learn from it. If they are making repeated errors then I realize that my criteria are too high and I need to make changes to increase my dog’s success rate in the session.
The other big issue I see with the concept of errorless learning is that it encourages trainers to have analysis paralysis. They are so worried about executing a less than perfect session that they don’t even try. There is a real fear of ‘messing up their dogs’ and ruining not just their training, but psychologically damaging their dogs, if they don’t have the absolute ideal session every single time. Luckily for all of us our dogs are typically much more resilient than that. In fact, much of the time they manage to learn in spite of us rather than because of us! I think we’d all be happier if we just loosened up and had fun with our training rather than worrying so much about absolute perfection.