EL110: Beyond the Click: Adventures in +R2.0

Intro to Errors AS Learning: What it is, why we should be doing it, and what to watch out for

Let’s begin with a discussion of the term "errorless learning”, which is important to us because is offers the mindset that underpins everything we will be doing in this course.

However, the term "errorless learning" is also somewhat problematic, for reasons I will present below. 

I will then introduce what I teach instead, which is an "Errors AS Learning" approach to training.

As such, I’m going to start by discussing the mindset, which is what I want us to embrace. And then we'll go over the possible pitfalls this term sets us up to encounter, and what to do instead.

 

"Errorless Learning": The upside

First off, what does "errorless learning" even mean?

For some, the term evokes images of dogs running willy-nilly, never being told they're wrong, and being clicked whenever they accidentally happen upon the desired behavior.

For others, the term means setting up training, and being so perfect in our delivery, that our dogs never make a mistake in the learning process.

Neither is what we will be doing here!

The idea of never punishing our dogs, and never letting them know they’re wrong, is on target, but that's where the similarity between what I just described and the actual practice we'll be doing.

Technically speaking, the term “errorless learning” is used to describe an approach to teaching that seeks to minimize errors through carefully planned teaching conditions. It's about meticulous planning, strategic arrangement of the training environment, and us (the teacher) having an extremely clear understanding of the target behavior and of learning more generally.

Be forewarned, putting errorless learning into practice is not easy. And it may require a complete overhaul in your thinking about, and approach to training. More on this in a moment.

While new to many of us, the term has in fact been around for a long time. Specifically, it originates from work done in the 1930s by psychologist B.F. Skinner, who argued that "errors are not necessary for learning to occur.”

Skinner was responding to an 1898 research paper titled "Trial and Error Learning", which described learning as a slow and arduous process. Skinner disagreed with this conclusion, countering that, in fact, “Learning success resides in the environment, not in the learner”.

More precisely, Skinner proposed that “Errors are not a function of learning or vice-versa, nor are they blamed on the learner. Errors are a function of poor analysis of behavior, a poorly designed shaping program, moving too fast from step to step in the program, and the lack of the prerequisite behavior necessary for success.” 

In other words, Skinner believed that learning is the responsibility of the teacher; that is, if the learner is not learning, if the learner is making mistakes, the problem is not to be found in the learner.

Rather, the "problem" is located in the learning environment. I don't like the word problem in this context, so let's call what we need to focus on the "component that needs tweaking". So! When our learning struggles to understand what we're asking, we want to examine and tweak one or serveral of the following: 

• The “topography” of the behaviour (briefly: the picture of its components—don’t worry, we have a whole lecture on this topic later in the course); or
• the shaping criteria; or
• adding more splits (to avoid lumping); or
• the antecident arrangement (props and other information)
• teaching a behaviour they need to have in place first.

In an Errors AS Learning approach, (just as in 'errorless learning') the dog is indeed never wrong. But this doesn’t mean mistakes don’t occur! They do. They will! They are part of the process.

Repeat after me: It is impossible to be perfect!

Just a little aside: If we can do something perfectly, that implies there is nothing to be learned. As we are teaching NEW things, learning is inherent. Which means errors are too.

So! in an Errors AS Learning approach to training, we're not trying to avoid errors! Yes, we do want to minimize them where possible.

However, our actual goal is to use errors as data to understand what our next step forward is.

What we need to understand is that the "error" in the dog's behaviour is the result of the shaping plan, not of the dog (or of the teacher! It's critical to be as compassionate with ourselves as with our dogs).

 

Behavior doesn’t occur in a vacuum – it occurs in response to a set of conditions. Our dog is simply responding to the learning conditions, and, if those conditions do not provide sufficient information, the dog will respond accordingly, by making a mistake.

Which might look like: Repeating a previously reinforced or rehearsed behavior, freezing, improvising by throwing out random behaviors, shutting down, barking, being silly, or even leaving.

If you find your dog engaging in any of the above behaviours during training, then adopting an Errors AS Learning mindset and approach to teaching will make a big difference!

Practically speaking, throughout this course we will become very aware of how to break down the behaviours we wish to teach, set up our learning environment strategically, maximize our dog’s success, and minimize their chance of making errors and performing “junk behaviors”.

And, most importantly (at least from my perspective), minimize frustration to maximize a positive emotional state throughout learning.

If you think about it, training is pretty much a game of charades we play with our dogs. We act out what we want, and they guess. When they guess correctly, we give them a treat.

Have you ever played charades? It can be a lot of fun! But it can also be confusing and frustrating. I like it, but it’s not something I want to play all the time.

Now imagine if it is the only way you can communicate with someone, or get that thing you want most dearly. For our dogs, charades is our main method of teaching and learning. I believe, therefore, that it is our responsibility as teachers to make this guessing game as easy as possible.

The better we get at it as a team, the more complex the behaviours we can work out. But we need to start simple, and get some foundations in place. We need to think things through carefully, and communicate clearly. This is what Errors AS learning is all about, and what we will be doing in this course.

Example

Here’s an example of me teaching a new behaviour to one of my canine partners, Desirée. The behaviour is a new “base position” (a default start position – we will be working on base positions later in the course), which is an automatic down on a mat.

The “antecedent arrangement” or set-up is me sitting on the floor, cross-legged, in front of the mat. I want Des to see this set-up and know her job is to come and lie down in front of me.

In this clip, I use an "Errors AS Learning" approach by tweaking what I'm doing / the information I provide, based on the behaviour she does! Note how she get a cookie every time, even when the behaviour she does is not the one I want.

Of note, I also use the following techniques, which we will be working on in this course: Transport, a marker system, luring, shaping, stationing, base position, micro-movements, strategic placement of reinforcement, loopy training, antecedent arrangements, establishing a cue hierarchy, behavioural momentum (my favourite), and an end-of-training routine!

Yes, all of that in just three minutes! Can you spot them as I train? (we will break down this video in a future lecture)

The clip comprises the first and second sessions of training this behaviour; by the end of the clip, Des has the behaviour down pat.

You will note that she does make errors (ex: sitting instead of lying down), each time because of unclear information on my part. I quickly adjust my behaviour (Errors AS learning), and she corrects hers:

 

"Errorless Learning" – The downside

Before wrapping up this discussion, let's take a look again at the original term of "Errorless Learning", and why I've made the shifted to "Errors AS learning" in how I talk about this philosophy.

1) First, the term "errorless learning" puts all the responsibility of both learning and training on the shaper/teacher. In other words, on us.

This suggests that our learners are merely passive recipients of our wisdom, which is certainly not the case! Our learners come to us with plenty of their own ideas, undersanding, and cognitive abilities!

 

2) The very term "errorless" suggests that there should be no errors, and so if there are any, we are to blame. And that's a lot of pressure!

I don't know about you, but I don't need any more pressure in my dog training. I'm already my own worst critic and the first to blame myself when things go wrong. I don't need entire dog training theories to give me more reasons to beat myself up!

We're all about compassion here: For our dogs, and FOR OURSELVES.

And so, I changed the term to emphasize the importance of not taking this concept  literally.

We need to embrace the spirit of trying to plan and organize our training in a way that maximizes learning and minimizes frustration. But perfection is simply not possible. And we need to be OK with that.

 

Eileen Anderson wrote a great blog post outlining both the history of errorless learning and her critique of the term (not the concept), which you can read here. She suggests that we rename the approach “Enhanced Chances of Success Teaching”. Hmm…. I like that. I wonder if it will catch on.

Here, inside BtC (and with all my training), I will call it Errors AS Learning (or perhaps we can abbreviate it to EAS... hmmm, that's almost EASY! I need to think on that: Errors AS learning YAY? But I digress...)

 

3) Our dogs are not "black boxes or response".

Another reservation I have is that errorless learning, as presented by Skinner and company, treats the learner like a black box of response, instead of as a thinking, breathing, feeling, and creating being.

And as such, the learning process that emerges from this thinking is both limited and left too rigid for my liking. There is room for the dog to be creative in learning. In fact, I often depend on it.

Supporting our dogs in building their creative thinking skills is particularly important for more complicated tasks such as herding, search and rescue, agility, scent work, and gundog work and so on.

In these complex contexts, we depend on our dogs to think outside the box, to be creative, and often to improvise in order to get the job done. And we want to make sure that our learners maintain and develop this ability!

As such, I'm going to strongly encourage you to embrace the concepts presented throughout this course, and work them into your current training practice in ways that make you more successful and training less stressful, but don't try to do this all at once. And don't try to train everything in such a precise, careful manner. 

There is room in dog training for messiness; in fact I believe it's critical. Sometimes we need things to get messy to get new ideas, for example. (Deb Jones wrote a great blog post recently on the importance of messiness in training, which you can read here)

I use the techniques and concepts taught in this class to train foundation behaviors; to build my relationship with my dogs around training; and to give them the training behaviours that they need in order to do the more complex work.

As such, when things get complicated, my dogs have strong core behaviours that anchor them in the messiness, and the confidence derived from a well-established, positive conditioned emotional response (+CER) to training developed through their initial learning in a low frustration environment.

I advise that you work through the exercises in this class a few minutes at a time. Work on each piece as an exercise in and of itself, sort of like sit-ups and pushups to help build your overall strength as a trainer.

If you’re not sure what to do, experiment! Try short sessions, and, if things don’t work out, step back and use the "data" you've collected as a learning process to hone your skills and decide on your next step.

Hence: Errors AS learning!