101 Things: Thinking Outside of the Box

So You Want To Learn About Shaping

If you ask a dog training group how to get started with shaping as a training method, you can nearly guarantee that one of the first comments will be a suggestion to play 101 Things To Do With A Box. There’s a pretty good chance that a similar conversation is how you found this article.

The training community presents 101 Things as the traditional beginner’s shaping exercise. At this point, it’s almost synonymous with shaping altogether in some training communities. In particular, it gets recommended as a panacea for building confidence and creativity. Deciding to experiment with 101 Things is almost a dog trainer coming-of-age tradition.

And I have issues with that.

Apocryphal Versus Original

Let me begin by saying that what I am describing here is the way that 101 Things is presented in the amateur dog training community at large.

This is distinctly different from the exercise Karen Pryor originally described. (We’ll talk about the real deal later.)

That difference matters.

What I am describing is how I see the game commonly presented in amateur training groups.  For this article, we’ll call this the apocryphal version to distinguish it from the original version described by Karen Pyror.

Apocryphal: (adj.) A story or statement of doubtful authenticity, although widely circulated as being true.

What Is 101 Things To Do With A Box?

As it’s usually described, the apocryphal version of 101 Things is:

1. Present your dog with a novel object with no other information.
2. Reward them for any novel interactions with the object whatsoever.
3. Do not reward for any repeated behaviors.
4. Hooray, your dog is now creative and confident, probably.

The emphasis on novel interactions here is key.

Shaping For Creativity

This exercise was widely popularized in Don’t Shoot The Dog (Karen Pryor, 1984), based on a similar exercise done with dolphins.  The goal was to increase the learner’s repertoire of offered behaviors. Or in other words, to build creativity.

You may be surprised to learn that in the original passage, Pryor does not say that each interaction has to be novel.  She only specifies that you should begin by rewarding any behavior offered in the direction of the object, traditionally a box.

Despite this, training community gospel says that from step one, you should exclusively reward for novel variations on behavior and that repeat behaviors do not get clicked at all. If your dog nose targets the box (click/treat), touches it with its paw (click/treat) and then nose targets the box again, the second nose target does not get rewarded because that behavior has already been used. It fails to meet the novelty criteria.

In theory, the intention is to teach the dog to offer a wider variety of behaviors and increase their behavioral repertoire. It’s meant to install the concept of offering and creativity.

Beyond the fact that it’s not how Pryor originally described the game, let’s talk about the issues with this version on its own merit.

Issue #1: It’s Not Shaping

Or at least, it’s substantially different from anything else you’re likely to be categorizing as shaping, and it’s a particularly abstract form if it is shaping at all.

It’s certainly not Shaping 101, the foundation exercise.

In traditional shaping, you reward successive approximations toward a finished goal. Traditional shaping includes a lot of mostly-repetitive behavior, because shaping is all about sculpting a moving bell curve.

To over-simplify, most of your reps are going to look pretty much the same. You’ll have some outliers toward the top and bottom of the bell curve. You reward the upper 80-90% of the bell curve and don’t reward the lowest 10-20%. The lowest 10-20% fades away without reinforcement.  The upper 80-90% gets paid, so it maintains or increases. The average gradually trends upward to follow the reinforcement. Your bar moves incrementally higher. The dog is still right most of the time.

The old adage says, “Good shapers are good splitters.”

By that, we mean that people who are good at shaping are very good at finding tiny ways to tease apart specific aspects of a behavior, teach and refine them individually, then combine them together to build a cohesive whole.  We want carefully cultivated variations on a very narrow theme.

This is the utter and complete opposite of what you are doing in the apocryphal version of 101 Things!

In the apocryphal version of 101 Things, your goal is wide variability and any repetition is immediately put on extinction. Extreme variability isn’t just tolerated — it’s the only thing that gets paid. That means you effectively have almost no control over steering the final behavior.

Shaping is a scalpel.  The apocryphal version of 101 Things is a chainsaw.

Both of them technically cut, but I know which one I want in my surgeon’s hands.  Know what I mean?

If you are trying to shape for tighter sits in the obedience ring, and your dog is throwing random behaviors indiscriminately, you are going to have a dog standing, laying down, backing up, spinning, barking, raising paw, and getting frustrated.

And your sits will not be any cleaner at the end. They will not be cleaner because you will not be shaping.

This is not what shaping looks like.

In shaping, your goal is moderate, structured changes.  You want the dog to repeat the same general behavior so that you can sculpt it in your desired direction. You prune out the parts you don’t like and nurture the parts you want to keep.

You can think of it as a kind of behavioral topiary, if you want.

Photo credit: Michel Rathwell via Flickr

Issue #2: How Do You Get It On Cue?

In the apocryphal version of 101 Things, what happens if your dog offers something really great?

I’m talking like the Mona Lisa of dog tricks, perfect execution, flawless form, 10/10, the Golden Buzzer, the whole nine yards.

After you marvel at your dog in wonder, how do you make it happen again?

You get to click it… once.

And if your dog has learned that nothing ever gets clicked again when you’re “shaping,” pray tell, what are you going to do with that creative breakthrough?  I’d like to hear your plan to put it on cue if your dog knows not to offer things twice.

Yeah, I don’t know either.  Sure would be a handy place for actual shaping, though, wouldn’t it?

Repetition with minor variations is the engine of shaping.  In ideal shaping, you want tiny steps in the right direction, sculpted carefully by the handler, not wild random guessing on the part of the dog.

You want finesse, not flailing.

Be a surgeon (or a topiary-ist?), not a lumberjack.

A dog offering a huge variety of completely unrelated behaviors during a shaping exercise is a bug, not a feature.

Issue #3: It’s Not A Beginner Exercise

The apocryphal version of 101 Things is often frustrates both the dog and the learner, and no wonder: it’s pretty much straight-up extinction! Friends, that’s an incredibly thin schedule of reinforcement, especially for an inexperienced dog.

Frustration is the kiss of death in a shaping session.  If your dog is frustrated, you need to stop and re-evaluate your criteria.

Even if you were using this exercise specifically to increase spontaneous behavioral variability (which, again, is not traditional shaping), this would be a brutally steep learning curve.

Lag Schedules

In Applied Behavior Analysis, what you’re doing is called a “lag schedule.”

In a lag schedule, the learner is shaped for incrementally increasing levels of behavioral variability within a specific theme. It’s used to encourage new spontaneous responses instead of repeated, stereotyped behavior.

Well okay, that sounds like 101 Things, right? So what’s the problem?

The difference is that a lag schedule specifically marks how many behaviors have to be novel in a row. For example, in a Lag 1 schedule, a dog could nose target, paw target, nose target and all three would be reinforced. In a Lag 2 schedule, the dog would need to do two novel behaviors before a repeat could be clickable again, and so on.

The number of novel behaviors before an offered behavior is re-clickable only increases with the learner’s ability level.  So while it teaches variability and creativity as a skill, it also explicitly builds that skill by starting with very small amounts of variability and moving upward from there as the learner succeeds.

And critically, in the articles I was able to find, I saw nothing larger than a Lag 3 schedule, and that was with humans.  A Lag 3 schedule would be A, B, C, D, and then A would become reinforceable again.

That is wildly different from putting an object in front of a learner, rewarding them for any novel interactions and putting literally all repeated behaviors on extinction forever.

The apocryphal version of 101 Things is the equivalent of a Lag Infinity schedule. It’s almost entirely extinction.

No wonder people think their dogs “don’t get shaping!”

Issue #4: It Doesn’t Build A Trainer’s Shaping Skills

The apocryphal version of 101 Things To Do With A Box does not teach new trainers how to shape.  It also convinces trainers who are otherwise capable of learning about shaping that shaping is too hard.

Shaping isn’t too hard. The apocryphal version of 101 Things is too hard, and that’s because it isn’t shaping.

By mis-teaching 101 Things, we set our newer trainers up for failure. We ask them to present their dog with a frustrating, difficult and abstract exercise as their first attempt at a new type of dog training. And then we tell them that this is supposed to be easy. After all, it’s the beginner game!

When they struggle with the apocryphal version of 101 Things, many of those trainers decide that shaping must be above their skill level.

Or worse, they decide that their dog’s frustrated flailing is the same thing as offering behavior. They get very enthusiastic about teaching this game to their novice dogs, and they frustrate their dog to the point that the dog frantically offers behaviors rapidfire without thinking in hopes that something, anything, will be clickable.

Good clicker training should not be frantic, even with fast, high-drive dogs.

BUT MOM, frantic flailing is my FAVORITE!

So, What Do We Do Instead?

Let’s narrow this down and create more targeted, appropriate goals for our learners, both human and canine.

We can start by looking at the original version of 101 Things To Do With A Box.  Unlike the apocryphal version, this is a good place for new shaping teams to start.

The Original Version of 101 Things

Many clicker trainers play a game with their dogs that I have nicknamed “101 Things to Do with a Box” (or a chair, or a ball, or a toy).

Using essentially the same procedure we used at Sea Life Park to develop “creativity” in a dolphin, in each session* the dog is clicked for some new way of manipulating the object. For example, you might put a cardboard box on the floor and click the dog for sniffing it and then for bumping it with his nose, until he’s pushing it around the room. The next time*, you might let the dog discover that pushing the box no longer gets clicked but that pawing it or stepping over the side and eventually getting into the box is what works. The dog might also come up with dragging the box, or lifting and carrying the box.

One dog, faced anew with the challenge of the box game, got all his toys and put them into the box. Click! My Border terrier once tipped the box over onto herself and then scooted around under it, creating the spectacle of a mysterious traveling box. Everyone in the room laughed hysterically, which seemed to please her.

Karen Pryor (Don’t Shoot The Dog, 1984)* Emphasis mine. I’ve also added paragraph breaks for ease of reading on mobile devices

The Critical Difference Between Original And Apocryphal

Did you catch that?

In the original (real) version of the exercise, the criteria isn’t “all responses within a session must be novel.” Quite the opposite!

In the original version, each session has a different target behavior with an object. Those behaviors are shaped from the ground up from small offered behaviors which are refined toward a goal as the behavior shifts. The only extinction is between sessions, not within them.

That’s a HUGE difference!

In this version, the trainer keeps extinction to a minimum. The dog gets it right much more often than wrong. Unlike the apocryphal version, this original version is actually shaping.

You can read more about the original version here on the Karen Pryor Clicker Training website.

What Does A Real 101 Things Session Look Like?

Pick an object.

Place it conspicuously in the training area very close to where you will be standing or sitting. You want it to be close enough that you could reach out and touch it. Large objects are usually easier for beginning shapers.  I usually start with a chair, box or platform.

Be prepared to click as soon as you bring your dog into the room.

As soon as your dog acknowledges the existence of the object in any way, click and reward in the direction of the object. ANY acknowledgement is “legal” at this point. The bar literally could not get lower. Looking, pawing, nose touching, sniffing, glancing, walking around, etc.

As soon as your dog finishes swallowing the first treat, look for the next behavior oriented toward the object. It will usually be almost immediate, because right now, our criteria is any interaction with the object. It can be the same thing they just did or something different, as long as it’s directed at the object.

When they acknowledge the object again, click and pay toward the object.  Repeat this step about twenty times.  You’re looking for a “holy CRAP, this thing is AMAZING! Look what it can make my human do!” response.  This often manifests as eye contact with you, then abruptly looking toward the object and turning back to look at you.  “Did that work too?” as if they can’t quite believe their luck. Do a couple more repetitions once you get this response.

This is the step where we start to have standards.

Choose a body part. If you’re just starting out with shaping, pick between nose and front paw — ideally whichever they seem to be offering more of.  Now you’re shaping for interactions between this body part and the object.  For the sake of simplicity, we’ll call it their paw and a box in the example, because that’s pretty typical.

Now you’re clicking for any interaction between the dog’s front paws and the box. Touching it, stepping toward it, raising a paw in its direction — anything paw-related.  Accidental motions totally count. It doesn’t have to be intentional at this stage. Both front paws count. If you’re questioning whether it counts or not, err on the side of yes.

If you go more than ten seconds without clicking at this stage, throw a freebie treat for your dog to chase and then lower your criteria when you get back.  Click the re-approach while your dog is still in motion toward the object. Click a couple of orientation-toward-object behaviors, and then start looking for paw-specific behaviors again.  “Remember the game we were playing?”

Beginner-Friendly Shaping Goals

When someone asks about shaping, they almost certainly want to learn about reinforcing successive approximations, not shaping for spontaneous novelty.  And despite the reputation, the apocryphal version of 101 Things is more likely to frustrate a new dog than build confidence, creativity or a relationship.  If you want to help your dog become more confident, aim for clarity, not creativity.  Confidence leads to creativity, not the other way around.

As a community, we’ve got to stop throwing newbies in the deep end and wishing them good luck.

What was your first shaping exercise?  Let me know in the comments below!