Last weekend, I got to attend a brilliant seminar with Susan Friedman, Ken Ramirez, and Helen Philipps. Phoebe and Fanta got to join me for a four-day training vacation in Puchberg am Schneeberg.
For the first time, I got to attend a lecture and learning lab with Ken Ramirez of Shedd Aquarium and Karen Pryor Academy fame. He talked about a number of fascinating training topics, four of which I want to share with you:
- Concept Training
- Training Multiple Animals
- Building non-food Reinforcers
- Shedd’s Dog Project
- Concept Training
Concept training is a specific kind of generalization or abstraction. For example, the concept “bird” does not only refer to the prototypical sparrow, but also to flamingos, ostriches, ravens, chickens and macaws. Understanding a concept means being able to apply a mental category to a novel situation, e.g. regognize a winged animal you’ve never seen before as a “bird”.
Examples for concepts we can teach to our dogs include modifiers (right vs. left, round vs. square, colors, numbers), space conceptualization (e.g. a guide dog knowing how tall his owner is and being able to guide her around low-hanging obstacles like tree branches or ladders sticking out of the back of a truck), matching to sample, adduction (the combination of well-known cues to form a new behavior), and mimicry.
What distinguishes the understanding of concepts from other kinds of cues is that in traditional training, dogs are taught to associate a praticular cue with a particular object (“ball” = this tennis ball) or a particular behavior (“sit” = lower your rear end to the ground). A dog who understands a concept, on the other hand, is able to apply a set of rules to a context he has never encountered before. He would, for example, be able to pick the ball out of a number of new toys – or he would be able to show you the bigger (or smaller) of two objects he has never seen before. He hasn’t simply learned the name of one particular object or behavior, but formed a mental model that allows him to draw inferences about novel situations.
It blows my mind that we can teach our dogs not only the names of things, but also abstract concepts. I was particularly happy about the opportunity to learn from Ken since I had just missed getting a spot in Claudia Fugazza’s seminar a week ago!
Prerequisites
Concept training is a fun way to hone our training skills and our dog’s thinking skills. It’s not a necessity for most pet and sports dog owners, but a grand prix level behavior, as Alexandra Kurland would say.
Ken points out that before attempting concept training, our dogs should have a background in creative learning games and free shaping. They should also have good generalization skills; i.e. they should have experience not just with the fact that “down” means “down,” no matter whether you’re inside our out, at the dog park, in your living room or at a restaurant, but that “down” still means “down” when you’re out of sight, when you turn your back to your dog, when you’re lying on the floor, or when the cue is given by a different person. Furthermore, it helps to have a well-established default behavior that allows the dog to patiently wait and observe as you set up a training scenario, show your dog something, or give a set of cues.
Modifier Cues
Ken distinguishes 3 types of modifiers. Internal modifiers are objective and remain constant. An example of internal modifiers is left vs. right: they are always viewed from the dog’s perspective, and there is no confusion from the dog’s point of view – left always means her left, never someone else’s. External modifiers are variable and subjective, e.g. “big” vs. “small”: when compared to a golf ball, a tennis ball is big. However, when compared to a soccer ball, a tennis ball is small. Abstract modifiers are the most challenging ones. An example for abstract modifiers would be teaching numbers (counting).
Before you begin
Ken advises us to think about a number of points to consider before we start our concept training sessions:
Do you have your target behavior solidly on cue? That is to say, let’s assume you want to teach left vs. right. It’s best to not start with a new behavior, but use a well-established one. So before working on the differentiation of left and right, make sure the behavior you are planning to use – go into your crate, paw shake, retrieve, or target – is well established.
Ken further advised us not to dwell on one step too long: the longer you stick with one step, the more difficult generalization becomes for the animal. In concept training, we want the animal to recognize that she’s never done learning – the situation is always going to be different. Furthermore, Ken advises training in pairs from the beginning (i.e. using two rather than one objects from the first session onwards), and making sure to train in a relaxed atmosphere. If we’re going to adduct cues, Ken suggests deciding on the syntax first – it should always be consistant. An example would be Noun – Modifier – Verb: Ball – Large – Retrieve. In his own training, Ken uses the action cue (e.g. Retrieve) last. This eliminates the need for a a release signal, since as soon as the animal hears the action cue, he is off to do the job he has been told to perform.
Ken walked us through an example of internal modifiers and showed us a video of Coral, one of the rescue dogs participating in Shedd’s dog project, learning to discriminate left from right:
Coral learns to touch either the right or left target stick on cue.
Steps for teaching internal modifiers:
1 Teach right and left with known behavior […]
2 Fade prompts
3 Increase distance
4 Change targets
5 Change behavior
6 Add obstacles
7 Increase complexity
Coral had “Target” solidly on cue, and had a default sit + eye contact. This allowed Ken to introduce the cues “left” an “right” from the very beginning. He sat Coral in front of the two targets, cued “right” and pointed to the target on the right target while saying “target”. Once this worked, he started fading the hand signals. Next, he increased the distance, and finally, he moved to a new set of objects (two kennels: “right kennel” and “left kennel”).
Of course, I had to try this with Phoebe! The behavior I chose is one she knows well: targeting a paw light. Here’s our first session:
Steps for teaching external modifiers:
1 Determine objects to be used [Ken’s example was 4 cylindrical blocks as shown in the image above)]
2 Start with one of the extremes, either the smallest or largest size available.
3 With subjective, variable modifiers it is usually helpful to make sure one extreme is well understood before teaching the opposite modifier.
4 Pick a pair to begin with focusing on one extreme […]
a Vary location of objects
b Switch the non-target object often (in other words, if focusing on large, keep changing the small object so that it’s not the same – but always smaller. And the large object remains the same and constant)
c Once animal reliably understand [sic!] the extreme (largest of the objects), change it so that a smaller version of the object is now the large.
d Switch objects and vary which is large as soon as animal demonstartes that he is ready.
e Use errorless learning (error reduced learning) if needed.
5 Once first half of the modifier pair is well learned, move to the opposite extreme.
6 Switch regularly between the two – while also changing focus and non-focus objects and position.
7 When your animal is recognizing those two modifiers reliably well with one type of object, transfer to a second type of object.
[…]
Of course, I had to give this a try as well. I don’t know whether I’ll have the patience to truly teach the concept of size, but at least I started, and we had fun. Here’s a video of our 5th session:
Abstract modifiers: counting.
Ken told us about a project he’s been working on with rescue dog Coral: counting! Fascinatingly enough, Ken has taught her to count up to 14 (!) objects – and to do so faster than we could count them ourselves! Here’s an older video that shows the beginnings of the counting project with Coral:
Of course, I’m going to try this with Phoebe, too! I’m not ambitious enough to teach her to count up to 14, but counting up to 3 would be quite nice! In theory, it should be easy enough – probably easier than size discrimination: 1 object in a tray becomes the cue to touch target 1, two objects in a tray become the cue to touch target 2, 3 objects in a tray become the cue to touch target 3, and so on. I’ve already found myself a tray, a couple objects, and a number of targets. I’ll keep you updated as we get started.
Matching to sample
Matching to sample is one way of using adduction. It is a concept that has always fascinated me: animals learn to point out an object identical to the one you indicate, and can even do so with novel objects.
Let me show you my favorite matching-to-sample video. It gives you a good idea of what matching to sample looks like. Plus, there’s a poodle in it!
Mimicry
Mimicry means observing another animal (of the same or a different species) and copying its behavior. Research on mimicry has been done with dolphins by the US Navy (1991), and with dogs by Claudia Fugazza (Do as I Do, 2014) and Ken Ramirez (2013). Interestingly, Ken and Claudia Fugazza didn’t know that they were both researching and working on mimicry, but independently, both came up with very similar protocols for teaching it. Ken taught dogs to copy each other’s behavior while Claudia taught them to copy a human’s behavior.
Ken showed us videos of dogs and marine mammals learning to mimic a known behavior they saw another animal of the same species perform, and then even mimic a novel behavior – one they had never performed in their lives. Wow – this is extremely impressive. Unfortunately, none of the videos Ken showed us are on youtube, so you’ll just have to go and visit his next lecture yourselves if you want to learn more about this fascinating topic. However, to get a little glimpse at what mimicry looks like, here is a clip from Claudia Fugazza’s latest Austrian seminar – the one I unfortunately didn’t get a spot at:
I’m already looking forward to Ken’s upcoming online course on concept training (probably at Karen Pryor Academy in 2017).
That’s it for now – I’m going to hit publish and save the remaining reviews of Ken’s topics for my next post. Stay tuned!