GO!

Clicker training question

This is a place to gain some understanding of dog behavior and to assist people in training their dogs and dealing with common behavior problems, regardless of the method(s) used. This can cover the spectrum from non-aversive to traditional methods of dog training. There are many ways to train a dog. Please avoid aggressive responses, and counter ideas and opinions with which you don't agree with friendly and helpful advice. Please refrain from submitting posts that promote off-topic discussions. Keep in mind that you may be receiving advice from other dog owners and lovers... not professionals. If you have a major problem, always seek the advice of a trainer or behaviorist!

  
(Page 2 of 3: Viewing entries 11 to 20)  
1  2  3  
Squ'mey

too old to eat- any more KD
 
 
Barked: Tue Jan 22, '13 2:47pm PST 
I don't even call myself a trainer..I just work with my own dogslaugh out loud I do use a clicker for new behaviors. But I guess getting my dog to turn on the light, open the fridge, get me a pop, go back & close the fridge, then turn off the light...in one session of about 20 minutes...means I'm not doing something right? shrug He gets an "atta boy," or a "keep going" to encourage the completion of the chain. And a happy-dance-of-lovins" at the end.

Edited by author Tue Jan 22, '13 8:01pm PST

Approved by forums moderator
ARCHMX Asher RL1X RL2X RL3X RL

we will dance in- the ring without- words
 
 
Barked: Tue Jan 22, '13 4:45pm PST 
The question was about clicker training. I think Gail Fisher addressed the issue very nicely:

While I am clearly an avowed clicker trainer (a true believer – one might say a missionary), I have no ill feelings for trainers who lure continuously to get a behavior. Or who choose to mark the behavior simultaneously with presenting the treat. Or who use the click as a “keep going” signal or “praise marker.” Or who reward the dog only in position. Or who lure to help the dog when he’s confused rather than using the clicker to allow the dog to figure it out himself. I have nothing against people who use clickers with other methods of training. Just please, please, please don’t call what they do “clicker training.” It robs the rest of us of being able to clearly communicate who we are and what we do. We are clicker trainers.

And Kathy Sdao's's article on the differences between clicker training and training with a clicker from Karen Pryor's site is pretty clear:

Are You Clicker Training, or Training with a Clicker?
By Kathy Sdao on 10/01/2006
Filed in - Fundamentals - Training Theory

I began teaching people how to clicker train their dogs in 1996. At that time, most pet owners had never heard of clicker training and few class instructors took it seriously. Mine was the only advertisement in the local Yellow Pages that mentioned the word "clicker." I had to persuade students to even try this novel gadget.

A decade later, clickers are now common in dog training classes. But, I suggest, clicker training still is not common enough.
Are You Clicker Training, or Training with a Clicker?

I do believe "clicker training" is an unfortunate term for what we do. It's misleading in two ways:

You can "clicker train" without ever touching a clicker. I did this when I trained marine mammals. During those 11 years, I used various behavioral markers, including an adjustable-pitch Acme Silent Dog Whistle (with beluga whales), an underwater acoustic ping (with US Navy dolphins in the open ocean), the word "good" said with specific pitch and inflection (with a walrus named E.T.), and a single silent clap—a visual marker (with the dolphins at the University of Hawaii's Kewalo Basin Marine Mammal Laboratory).
You can use a clicker for training, yet be doing something quite different than "clicker training." I've met trainers who see nothing odd about holding a clicker in one hand and the transmitter of a remote shock-collar in the other. Their carrots are backed up by big sticks. Clearly, this is not clicker training.

So we're left trying to define genuine "clicker training" in order to distinguish it from "training with a clicker." Before I attempt this, let me acknowledge a few things:

Clear definitions matter. They allow us to talk about abstract ideas with as little ambiguity in our conversations as possible.
Definitions are social constructs. They aren't handed down from on high, etched into granite tablets. We—everyone who uses the language—create them, through discussion and consistent usage, though the opinion of experts does tend to carry more weight.
Clicker training is a powerful behavior-modification process. It warrants delineation from other training approaches.
It is needlessly divisive to define "clicker trainers" (versus "non-clicker trainers"). We are talking about a method and a philosophy, not a classificatory label. Though I passionately promote clicker training in seminars around the country, I sometimes use other training techniques (e.g., classical counter-conditioning). When working with animals, sometimes I'm clicker training and sometimes I'm not. Whether I (or anyone else, for that matter) should be labeled a "clicker trainer" seems beside the point.

And so, when you use a clicker to train your animal, are you clicker training (CT) or training with a clicker (TWC)? To answer this, consider the following questions:
Clicker training is a powerful behavior-modification process. It warrants delineation from other training approaches.

Is the click an event marker?

CT: The click pinpoints a behavioral instant, a moment of muscle movement. (With clicker-savvy animals, the click may sometimes be used to mark instants of non-behavior.)

TWC: The click is used in a less precise way, as a general signal that the animal has earned a reward.

Is the click a release?

CT: The click informs the animal that his movement met the trainer's current criterion; that is, his behavior was "enough" to earn reinforcement.

TWC: After a click, the trainer may require further behavior from the animal before paying up (e.g., after clicking, the trainer withholds reinforcement because the animal didn't stay in place, or didn't finish the weave poles).

Is the click meaningful?

CT: It is essential that the animal recognize the click as an independently meaningful signal. Therefore, great care is taken to ensure that the sound of the click occurs in a sort of "stimulus void."

TWC: The click is often overshadowed or blocked by other stimuli salient to the animal, occurring simultaneously with the click (e.g., the presence of a food lure on the dog's nose, the trainer's intention movements toward food). As a result, the animal becomes desensitized to the sound of the click; he will not react to it in ways that indicate anticipation of food or play (e.g., flicking the ears, looking toward the source of the click, wagging the tail).


Does the click predict a strong positive reinforcement?

CT: The click is paired with the animal's deepest desires: food, toys, interactive games, social companionship, etc.

TWC: The click is often paired with weaker reinforcements such as praise and petting.

Is the treat delivered "in position"?

CT: The emphasis is on delivering the treat as soon as possible after the click (though never simultaneous with the click). The treat is delivered regardless of the animal's position subsequent to the click. The trainer knows that the animal's position at the instant of treat delivery is reinforced, and so, when planning a training session, considers various ways to provide the reinforcement.

TWC: The emphasis is on delivering the treat while the dog is still in the correct position. Treats may be withheld if the dog moves out of position when hearing the click (e.g., the dog forges ahead of heel position or gets up from a sit).

Who is doing more work: the trainer or the trainee?

CT: The trainee is the more active participant, moving more than the trainer who remains relatively passive. The animal's job is to behave, that is, to move; the trainer's job is to observe the animal and to deliver timely, consistent, frequent reinforcements.

TWC: The trainer is the more active participant, moving more than the animal who remains relatively passive. The trainer is focused on making behavior happen, and uses food lures, body language, and physical prompts to "help" the animal.

Is speed of acquisition of a few key behaviors the most important goal of the training process?

CT: Each training session is an investment in the animal's future ability to learn. Knowing this, the trainer sacrifices instant compliance to gain momentum toward the goal of accelerated learning, when the animal has "learned to learn" and training becomes virtually effortless. This is often accomplished by allowing animals to get "unstuck" on their own, without lures and prompts from the trainer.

TWC: The priority is getting the animal to perform a particular behavior (e.g., getting a dog to lie down quickly and completely). Lures, prompts, and physical molding—all behavioral antecedents—may be used to speed this process. The animal may learn this initial behavior quite quickly, but also may be hindered in future learning situations by a tendency to remain passive, waiting for "hints" from the trainer.

Are all four quadrants of the operant conditioning grid used equally?

CT: Clicker training is an intentionally "unbalanced" form of operant conditioning. It has a preferential option for positive reinforcement (i.e., the trainer adds stimuli the animal desires), and, to a lesser extent, negative punishment (i.e., the trainer removes stimuli the animal desires). In most cases, clicker training avoids using positive punishment (i.e., the trainer adds stimuli the animal dislikes) and negative reinforcement (i.e., the trainer removes stimuli the animal dislikes), knowing the fallout that can result. Clicker training gets rid of unwanted behaviors using extinction, the training of replacement behaviors, management, and negative punishment.

TWC: The four possible consequences are used in proportions that are more equal. Positive punishments such as collar-pops, physical manipulation, and verbal reprimands are used to get rid of problem behaviors and to deal with the animal's non-compliance. These aversives are interspersed with clicks and treats.
I've met trainers who see nothing odd about holding a clicker in one hand and the transmitter of a remote shock-collar in the other. Their carrots are backed up by big sticks. Clearly, this is not clicker training.

Is the main emphasis control or communication?

CT: Clicker training is an elegant and effective method for communicating with animals in a coherent way. It challenges humans to strip away the constraints of verbal language and to tap into a more universal way of conveying information. Control of the animal's behaviors then flows as a by-product of consistent, clear communication and effective motivation.

TWC: Behavioral control is the principal goal of training. Communicating with the animal is the means to this end.

Is it important to realize the animal's full behavioral and cognitive potential?

CT: At its best, clicker training maximizes each animal's potential. It strives to make the animal a fully active, thinking participant in the training process. It encourages the presence of "the other" by constantly expanding animals' behavioral repertoires and by providing ever-greater cognitive challenges.

TWC: Training with a clicker may also aim high, attempting to tap into the animal's maximum potential. Often, though, the ultimate goal is a specific repertoire of discrete "obedience" behaviors, performed reliably on command.

Of course, when you come right down to it, a clicker has no inherent meaning. It can be used in all sorts of ways, both within animal training and outside that realm (e.g., US Airborne troops in World War II used clickers to identify friendly forces; Catholic nuns before Vatican II used them to cue the movements of students in church). My hope, though, is that the term "clicker training" will come to have a standardized meaning and that my colleagues who "train with clickers" will call their method something else.


One of the things I have noticed is the desire of traditional trainers to co-opt the language that force free trainers use to distinguish themselves.
[notify]
Zephyr

1213425
 
 
Barked: Tue Jan 22, '13 5:06pm PST 
Until 'clicker training' is actually trademarked or the original poster puts 'R+ Responses only' I'm pretty sure anyone is free to respond. And there are reasons for doing what is being described even if you disagree.

As for the co-opting of language, it's a little reminiscent of Donald Trump trying to trademark 'You're Fired', more than slightly ridiculous.

Edited by author Wed Jan 23, '13 10:45am PST

Approved by forums moderator

ARCHMX Asher RL1X RL2X RL3X RL

we will dance in- the ring without- words
 
 
Barked: Tue Jan 22, '13 5:12pm PST 
I am pretty much in line with Bob Bailey, who, when asked :

How do you define clicker training?

Responded:

I don't. Karen Pryor does, in my book.
[notify]
Zephyr

1213425
 
 
Barked: Tue Jan 22, '13 5:17pm PST 
Dandy. I'm glad she has a definition that is widely agreed upon.

That doesn't however render someone's explanation of what's happening and a possible reason why, irrelevant.

Edited by author Tue Jan 22, '13 8:01pm PST

Approved by forums moderator


Member Since
12/31/1969
 
 
Barked: Tue Jan 22, '13 8:55pm PST 
..and bob bailey is a marker trainer proponent, explicitly not a 'clicker trainer'.. according to seminars attended. laugh out loud

Edited by author Tue Jan 22, '13 8:56pm PST

[notify]
Lobo

"Stubborn" dogs- don't need- corrections
 
 
Barked: Tue Jan 22, '13 11:27pm PST 
Let's try to remain on track and keep the snobby remarks to a minimum. (:

Baby,
In short, there aren't any "rules" to using a clicker, however clicking without rewarding isn't clicker training - it's training with a clicker. And that is not always bad. (: It simply isn't "clicker training."

Has your question been answered? (:
[notify]
Smokey

Let's play tug!!
 
 
Barked: Tue Jan 22, '13 11:28pm PST 
I was trained with the click=treat method, and given Smokey's temperament, I doubt I'd try anything else. Freeshaping is fairly nervewracking for him. I have to work hard to make it as low stress as possible, and the food reward seems to serve as the best way to dissipate the tension. I have read some articles by people who fade the treats, sometimes altogether, but keep the clicks. Even if you believe this doesn't constitute real clicker training and, for example, prefer to use a marker word instead if you're not going to give an immediate reward, I think it's an interesting discussion about how dogs learn. I generally think that being younger, more confident, and more food motivated predicts a dog being naturally better at understanding longer strings of conditioned associations. This is not to say other dogs can't be taught, but I think you generally have to work harder at it, and fade the rewards much more slowly and steadily.

For example, my friend's dog is a puppy, very food motivated, and generally very excitable but not very anxious. He learns long strings of cause and effect relationships very easily. For example, he learned that barking and then stopping earns a treat, that the sun going down predicts dad coming home, and that a certain car sound means Smokey and I are coming in the door. Smokey, on the other hand, is an adult, less food motivated, and has a lot of anxiety that probably distracts him from forming associations. For about 6 weeks, several times a day, I got out of the car, closed the door, walked around to the other side of the car, waited for him to stop crying and sit, and then got him out. His crying behavior didn't decrease a speck in those 6 weeks. And although he's been to the same drive through with me dozens of times, and always gotten a hamburger patty, he shows no excitement until I actually start to feed it to him. So, when I'm working with Smokey, I tend to reward a lot. He has to get the reward over and over before he starts to feel confident that he's doing the right thing. And I tend to fade the rewards very slowly. Like I might reward one approximation of a behavior 100 times before I start only rewarding a better one. Sometimes I will even give a small reward for the previous iteration, and a big one or a handful for the next step. If I were trying to teach him a string of behaviors, I would go at a snail's pace to keep him from getting overwhelmed and walking away, crying, seeking attention, trying to diffuse the situation with play, etc. If I'm working with my friend's dog, it takes him about half a second to decide, ok, this isn't the behavior she's looking for, I'll try a new one. He seems to suffer no mental discomfort about it whatsoever. So I think it would be entirely possible to train him to great heights only rewarding now and then.
[notify]
Nare

Woo-woo- whineybutt
 
 
Barked: Wed Jan 23, '13 12:07am PST 
I dunno. If I click (even on accident) I always follow up with a treat.

I usually make Nare do a string of behaviors-- Sit, platz, head down, Fire!, stand, bow, spin, reverse, platz

And after the veeerrrry last one, he gets a click/treat. sometimes I click after only 2 commands, sometimes I do 20 of them before a click.

It all depends on the dog, I think we tend to tailor this stuff to what works for each individual.
[notify]
Jackson Tan

Lad about town
 
 
Barked: Wed Jan 23, '13 2:15am PST 
I will also reward at the end of a string. It works. For the reason Zephyr put forward. Simply, it does build up drive. Some dogs get more and more amped with every click or yes or whatever knowing there's gonna be a party at the end, and work harder.

I frankly find these distinctions tiring. This 'you're not a clicker trainer, you're a trainer who uses a clicker.' business drives me absolutely nuts. I hate quibbling.

If it works, do it. It's simple.

Edited by author Wed Jan 23, '13 2:21am PST

[notify]
  (Page 2 of 3: Viewing entries 11 to 20)  
1  2  3