As many of you know, I did a seminar on new research on canine behavior in Madison last October. (Oh, and by the way, have I mentioned it is now available as a DVD — hot off the press? Tee Hee, we’re pretty excited about it!) It was a great experience for me, because like all teachers do, I ended up learning even more than I had hoped to learn about the topic myself. One of the findings most interesting to me relates to how we respond to a correct response from our dogs. Two studies suggested we would be wise to be thoughtful about how we do so:
First, Lindsay Wood did research at Hunter College comparing 20 dogs who were trained to nose a target using either a clicker as a marker, or the word “good.” Not surprisingly, the clicker trained dogs were significantly faster to reach criterion than the dogs who heard “good” rather than “click” before they received the food reinforcement. I say not surprisingly, because first, acoustically the sound of a click is better at getting an animal’s attention because of it’s abrupt onset and offset, and also because it is a “broad band” sound, meaning it stimulates a large range of acoustic sensors in the brain. In addition, clicks are unique in the “sound scape,” unlike words like “good,” and can be produced consistently, whereas speech tends to vary depending on a variety of factors.
But get this: Another study by Smith and Davis, published in Applied Animal Behavior Science (2008) taught Basenji’s to target an object using either a 1) Clicker + Food or 2) Silence + Food. In this case, the click of the clicker was NOT more effective. The dogs who received no acoustic marker after the behavior learned at the same rate [note I’ve edited this from my first post. See subsequent posts later in the month for more of this. The plot thickens!] the group in the standard “click/treat” category.
I’d love to hear your interpretation of this finding. Here’s mine, feel free too add (or subtract): In the case above, the dog was asked to do a relatively simple behavior, touching nose to cone. In that case, any sound was simply a distraction and the dog did better if there was nothing between the action and the primary reinforcement, the food. However, I would argue that if the behavior being trained is more complicated, and requires shaping and precise timing of the marker, then a clicker would be more effective. As I thought about it, I realized that’s exactly how I use clickers myself: not for simple actions like sit and lie down, but for more complicated behaviors that have to be divided into steps. Make sense?
[Again, more to come on this! I’m re-reading the study and going over the methods, and as I said earlier, the plot thickens!]
In this same study, note that the dogs who were clicker trained were more resistant to extinction, but note that “extinction trials” took away the primary reinforcer (food), but not the secondary one (the click). Compared with silence, it makes all the sense in the world that dogs would continue nosing the object if they heard a click afterward than silence, yes?
I’d love to hear your thoughts… or any more research you know about that might be relevant. Research like this is what made the Madison Seminar such a hoot to do. We reviewed the DVD and Tawzer video did a super job recording the seminar, hope you enjoy if you are so inclined.
MEANWHILE, back on the farm: I’m back in pre-seminar mode, about to leave the office to get ready to do a day-long seminar on canine behavior at UW-Madison. I’m speaking to vet students, vets and vet techs, and very much looking forward to it. (I’m giving them a case study to work on: It’s Willie, so that’ll be interesting, hey?) It’s ridiculously warm here still, in the 30’s and 40’s during the day, but it freezes at night and produces lots of problematic ice. Most of the ice has melted though in front of the farm house, so I don’t worry as much about Willie slipping on it as I did. Whew.
Here’s what we’ve been doing when we weren’t working at our jobs, exercising us or the dogs or doing chores: working on a 1,000 piece jigsaw puzzle for the last 2 weeks. (Talk about the addiction of intermittant reinforcement! I hereby admit to becoming addicted to this puzzle; every night I’d say: “I’ll just find one more piece to fit in, then I’ll stop.” Fifteen minutes later…..). The night we finished it I ordered 2 more 🙂 (love the company, Pomegranate, which makes high quality puzzles and cards) because, for me, jigsaw puzzles are a “flow” exercise, in which I think about nothing but what I’m doing, rather than “monkey-minding” and thinking about work, chores, etc etc…..). Sort of like meditating for lazy people.
Jen says
I know that just this morning, teaching a behavior on a lark (taking Elka’s Gumby toy out of a box), she did far better after we brought a clicker and treats into it than with just the simple (so we thought) cue to “pick it up”. She will pick Gumby up if he is simply on the floor or on a piece of furniture; she will “find Gumby” if he is hidden while she is out of sight and then cued. But when we placed a box on the floor, and placed Gumby in the box? Oh, that was new and weird!
Granted, my fiance had not yet witnessed our clicker work, and our previous clicker work had been shaping “go to your bed”, so Elka would circle around, look at Gumby, lay on her bed, come back, etc. while I waited for the basics (touching Gumby was not, I thought, an unreasonable first step) to click and treat. He also urged her a bit more than I would have alone, which I think was VERY distracting.
teresavet says
I’ve read the bansenji’s article, and I think the problem is the dogs have been just conditioned to the cliker. So we have dogs that have just learnt something (to pair the click with food) and have to learn the second behavior just after that. For me that’s more difficult that just one behavior. And second, in my experience, dogs respond best to the clicker when they have learnt a couple of different behaviors with it, they have generalised the clicker use.
I didn’t like the second part either, same reason as you. They didn’t withdraw the clicker, just the food, I would have withdrown both.
I’ll look for the other article, it’s new for me.
We in Spain are in the low twenties this week. Siberian wind is running wild. No snow, though in Madrid.
I share your love of puzzles. In the days when I had time (and no insomniac two-year-old daughter) I used to addict myself to them. So relaxing!
Jeff says
I know I very much prefer no clicker. I personally find it irritating. Just the clicking sound. That’s just me though. Like someone clicking a pen, cutting their nail or what ever I just find irritating so I hate the clicker training. I think honestly though if I had to guess the dog can sense my discomfort with the clicker and probably interpreted it as a discomfort or displeasure with their performance. Where if I am not using a clicker and not near anyone that is then I am happy go lucky normal and more relaxed.
So I have always got better and faster results with no clicker. However, that could just be my irritation in the clicking sound going across to the dog.
Margaret McLaughlin says
In my experience dogs learn much more quickly with the clicker than with a verbal marker. However, this is not scientific–I mostly use the verbal marker “Yes!” with a skritch when I don’t have the clicker/bait bag/tennis ball with me. I have made some effort to “load” the “Yes!” as I would a clicker, but the reinforcers available when I mostly use it–the skritch, or a high hand touch–are probably not as reinforcing to my dog as food or a ball.
A second, but related issue, is rate of reinforcement. The more reps you can get into a minute, the faster your dog will learn, & once you’ve overcome the initial awkwardness with the tools, & developed some dexterity, you can get the behavior/marker/reward sequence more quickly with a clicker & small soft treats than with anything else I’ve seen or heard about.
Beth with the Corgis says
I find it hard to believe there would be any consistency in results across different dog/handler combos, or across different desired activities. Targeting a cone is very basic. I have to wonder, if one of the dogs offered several behaviors in quick succession (nosed the cone, pawed the cone, barked and spun), how would the handler let the dog know which of those behaviors was being rewarded?
Both of mine get too wound up by the clicker. Maddie just stares at it intently and refuses to offer any behaviors at all. It’s the same way she behaves if I hold treats in my hand. It may be her innate personality, or it may be that she was once a show dog and may have been trained (as many are) to focus on the “bait” no matter what went on around her, but she works best if there are no visible treats at all.
Jack shapes beautifully with a marker word, but with the clicker he will start to offer random behaviors (known and new) in rapid succession and give frustration-whines. I find he is too sharp with it. I only spent a few minutes “loading” the clicker when I first used it, so I don’t think I overdid that. But he finds it too stimulating and tries too hard and ceases to enjoy training. I save the clicker for times when he has an opposing stimulus in the environment that is equally (or more) rewarding than what I have to offer, since the clicker makes the treats more rewarding to him (it seems) than treats alone. So for instance, I will click the instant his head turns to me if he is flying off to eat some yummy cat poop that he smells in a field and I give a recall command. Or I’ll click in agility class if he looks at me instead of trying to find the treat someone else dropped in the grass. But for normal shaping, no it is counterproductive.
I HAVE trained both dogs using just treats and had it work well. I used just treats (no words) to get Maddie to keep her whole body in the crate when told to go into it; she would hang her head out, then progressed to having both front feet out. So I just stood in front of her with her treat (which she stares at, of course) and looked in another direction. As soon as she slowly, slowly offered to back into the crate, I just gave her the treat. It worked well. But again, there were no competing behaviors and so I did not have to differentiate WHICH of her actions was being rewarded.
Similarly, I taught emergency recall with no marker words. Just started by repeating the recall word over and over whilst feeding them hot-dogs, tiny pieces so they got like 20 pieces each. Two days of that and all I had to do was say the recall word and they’d come running. They of course think it means “I have lots of hot dogs right here for you!” and since they never otherwise get hot dogs, they think it’s wonderful. But again, that was a behavior with no competing behaviors between which to differentiate.
I can’t imagine how any of us could be quick enough with the treats to train without marker words when a dog is offering several new behaviors at once. And most dogs who have done any shaping at all WILL routinely offer multiple behaviors at once when asked to do something new. Paw waves, barks, sits, downs, spins are all offered in quick succession and the handler needs to “capture” for the dog which behavior was being requested.
Julie says
Well, I don’t have access to the full article but from the abstract I’m wondering — did they really mean they clicked *after* the correct response, ’cause if so, they did it wrong, and no surprise there was no difference from using no reward mark. Also, no surprise that it was longer to extinction for the clicked group (continuing to use a secondary reinforcer) as opposed to the group with no reinforcer at all! And they are also deducing that “basenji dogs may become progressively impaired with age in the acquisition of stimulus-reward contingencies” from this study? Couldn’t possibly have anything to do with how they have been trained and handled, rather than just their age? Hmmm. . .
Finally, I would love to see a study comparing lured vs. free-shaped behavior (was targeting lured in this study? it often is). I tend to find a difference in response and retention of behaviors taught using those two methods even when both are clicked.
Beth with the Corgis says
I realize that I just contradicted myself when I said Maddie works best with no food visible, and then gave an example where I modified behavior using food. In this case, though, she always gets a treat when I ask her to go into her crate and she does. I was actually withholding a treat she was already expecting, and the first time I did it she stood and started at me for easily three minutes before doing anything else. When she did back one paw slowly into the crate, she did so while staring at the treat (not me) the whole time and actually stumbled over the crate lip as a result. The move was in slow-motion. If she were out of the crate and I held a treat, she would offer a sit (she used to stand and it took awhile to condition her away from that) and then just sort of shuffle her front feet to demonstrate that she was, in fact, sitting. And that would be all I would get out of her with the treat visible. Hence, no treats most of the time til after we are done.
Roberta says
I’ve never used clicker training but here is something new we are doing: yep, silence. When I set up the food bowls (currently 19), I only crate two dogs – my wee Sheltie and Seymour – Sheila for her comfort and Seymour cause he proudly goes into his crate. The others are loose while I do supplements, meds, water, kibble and stir. Occasionally, if someone is too close or jumping on my legs, I’ll turn around making sweeping motions with my hands “back up” and click my fingers (hmmm…) w/o saying anything. I feed dogs one or two at a time (you should see Betty Boop, a Bagle – she is quite vocal but boy, she knows if she is not in her crate, no food for her – hilarious). Even when I go into the front room to feed the two there, no one, not even the Coonhounds, counter-surfs the remaining bowls. I then make a stack of bowls for the breezeway, gesturing those dogs to go down and crate up – and they do with some bays and jumping (not on me but out of enthusiasm) but no vocals from me. I find this quite fascinating; it helps me to not yell, I think it lessens the frustration of waiting out of sight for meals which results in a lot of baying and complaining. We all seem to be more relaxed before, during and after the meals.
Susan Mann says
I know Bob Bailey has said multiple times that for simple behaviors, “chucking food” is as, or more effective than, using a clicker. Save the clicker for complex behaviors. While he doesn’t have a study to back that up, ABE kept extensive records (which were unfortunately lost in a fire) since they were a business and efficiency was important.
Ann-Marie with Corgis, too says
Beth, first of all, remember, corgis are really, really food motivated. I only used the clicker with mine when I started training them in obedience class. After that, just the idea that they would get a treat for something was good enough for them, especially Skippy, who is clicker shy.
Fortunately, he is one of the wonderful few who takes great joy in problem solving for its own sake. He will happily play “Get the kibble out of the plastic milk jug” all day, while Faye, the other one will look at me and say, “Oh, Mommy, you have GOT to be kidding. You’re going to make me work for DOG FOOD???”
And the poor brain dead sheltie, Dodie, just looks from the jug to me, and says, “I’m just a poor simple sheltie. I smell something, but I don’t know where it is.”
I would like to try silence. Sometimes it’s hard for me to get a bark in edgewise around here. Two corgis and a sheltie make for a noisy household. I have been working on not letting them push me around when it is mealtime, refusing to move toward the food bins until they are quiet. It is slow going, but we might get there, eventually.
Larry C. says
I think vocal commands are a hindrance to training. Understanding a voice is artificial and difficult for a dog. Body language is more natural. It is faster and easier to train a dog to hand signals, with a simple audible command like a whistle to get them to look at me for the command. I learned this years ago by watching a deaf girl train a pup.
My dogs are often much too far away to make food rewards practical, so I don’t often use them. They are helpful starting a pup.
Donna and the Dogs says
I’ve always been too uncoordinated to be proficient with the clicker, so I’m a verbal marker user, I use several, including “Good,” “Yes,” and, “Ooops.” – Obviously I use them for different purposes.
Although I had great luck with luring, marking, and rewarding while teaching new behaviors for Rally class, it became a real issue when I started taking Nose Work classes, where I am NOT supposed to verbally mark the behavior, because you want the dog to “Stay” on the scent, and not turn to you for a reward.
What happens often when you use a marker in Nose Work is that your dog will locate the scent, and you’ll automatically mark the behavior, and next thing you know, your dog is orienting towards you, instead of the source of the scent.
The problem I have, is I can’t seem to keep my mouth shut. Which is why I’ve also never cared for Competition Obedience either. Give me Rally where I can “chat my dog up” – and I’m thrilled.
So, although I love Nose Work, and all three of my dogs appear to enjoy some form of it, I need some self control training myself. Either that, or a gag or some super glue…
Beth with the Corgis says
Larry, I agree about the hand signals. Verbal commands are helpful in that we can use them when the dog isn’t looking at us, but mine were all taught concurrently with word + hand signal. They respond much more reliably to signal alone than to word alone.
I inadvertently taught them that a smile was a release on a “stay.” Whoops, stupid human thinking about words and not paying attention to body language….
On the other hand, I also inadvertently taught my male that “excuse me” means “move before you get walked into/plopped on the floor.” I had no idea I was doing it, and yet he knows it. In the same way, they both learned “time for bed” means, “go in our crates for a treat.” No hand signals for either, and both work whether I’m sitting, standing, walking, facing the dogs, in another room, etc.
Contrary to conventional wisdom, I have found my dogs naturally pick up short phrases (with their inherent rhythms, and rising and falling inflection) than single-word commands. I have no idea which trainer began the instruction that single-word commands are best and should be used almost exclusively, but my own personal experience is that the opposite is true. After all, “time for bed” and “excuse me” and “where’s ultra ball? Get ultra ball” sound much different from each other than “sit” sounds different from “stay” spoken in the same monotone. My dogs have both learned phrases with no conscious effort on my part, and don’t get confused they way they do between words (say the word “tree” to a dog who knows “treat” and see the reaction.) With phrases, I also don’t need to get my dogs’ attention first by saying a name or making eye contact, or any of the other preliminaries I need to use to focus them before issuing a single-word command.
kecks says
Thanks for sharing the puzzle and card shop. 🙂 Wonderful stuff!
Barb says
That looks like a Charlie Harper picture. Love his work.
Beth with the Corgis says
Since I posted earlier, I’ve been pondering the “phrase” vs “single word command” issue and it occurs to me that sheepdog handlers also use short phrases: “Come by,” “Away to me,” and “That’ll do.” Hmmm. I believe (but could be wrong) that single words are used for the equivalent of “halt.” Which makes sense, since single words tend to have a braking effect. On people, not just dogs. Imagine someone yelling at you “Hey!” or “Wait!” or “Look!” or “You!” Any number of sharply intoned single words will make us stop what we are doing and attend to the speaker.
Annie R says
@Ann-Marie, thanks for a good laugh; the poor Sheltie is probably the most manipulative, getting you to do something before she will — reminded me of my very sweet but not-that-bright Cody, a Husky-Shepherd mix, who used to get so excited by the presence of food that he couldn’t even see that there was something to do besides stare at me until he got me to give up and hand it over. Fortunately he was totally cooperative by nature and just loved me so much that he didn’t need to do anything fancy to fill his role in life.
I too am way too uncoordinated to handle clicker, leash, treats and oh yeah, stay upright and don’t trip over the dog. It’s empowering to think I can actually trust my instincts which traditionally have been to use simple hand signals and a 2-3 word cue, followed by “YES!” with my mostly herding-breed-type mixed breed dogs. The hand signals really paid off in Cody’s older years when he was deaf as a stone but would still stare at me.
And I agree that it’s effective to use the sharp one-word cues for correction/attention-getting. “Wait” is a powerful word in our group as it means stop moving and let Mom catch up, whether at a street corner or fork in a trail. They don’t have to sit but they know to stop and look back for me. It’s not too formal but it does work.
I don’t do any showy stuff currently, but did do agility a few years back, and hated that so many things started with a “T” — table, teeter, tunnel, etc. Fortunately agility lets you use whatever you’ll think of in the split second you have to cue the dog; I liked a couple of action words, like Hop up, Walk up, Get in, etc. Worked well for my very motivated, visually-oriented dog who was looking where I was going or pointing; I think the words were actually a secondary cue for her. Some of this seems to depend on the dog’s habitual MO, right?
Pike says
I couldn’t find the study – so am obviously not sure of the exact set-up.
What came to my mind immediately was a workshop from last week’s Clicker Expo in Portland by Kay Laurence (about intelligent luring). The sound of the clicker is, of course, just one specific way of marking a desired behavior. She often uses another one: Opening the hand (that happens to hold the reward treat). Much more subtle (to us) than the auditory marker – but I was wondering if there might have been a visual marker in that study as well that preceded the food reward.
And we all know that dogs respond rather well to visual markers. Would it be possible to get a direct link to that interesting study?
Beautiful puzzle – definitely Harper! I want it!!
Michelle says
I think some simple behaviors could be taught just as quickly without the clicker (like the cone touch) and even though I am technically a clicker trainer, sometimes I don’t bother to use it. But when it comes down to refining the details of a behavior, the clicker is really powerful.
I’m thinking about my dog Buck who I am teaching to touch a target stick. Getting him to make the connection that food comes when he touches the target was very easy and probably didn’t require a clicker. But then he began to bite the stick and I had to start making sure I was very, very precise with my clicking so I captured the nose touching it before he opened his mouth to bite. Without the clicker, it would have been a real challenge to do so.
But in a more general sense, I have to agree that silence is golden when it comes to dog training. I used to babble to my dogs during training..”oh good dog, yes that’s it, no try again!” until another trainer pointed it out to me. Then I started making a very real effort to shut myself up (haha) and I saw huge improvements as my dogs were less distracted by what mom was saying and they were able to completely focus on the task at hand.
Kat says
As I’m now training my second dog as an adult I’m very struck by how differently the two respond to training. I use a clicker, marker words, and a quick nod. Well socialized Ranger watches for the nod–really more of a quick jerk of the chin down and back up a tiny amount. I didn’t choose this consciously but Ranger pointed out that that’s his most reliable marker that he did it right and I realized that he’s correct. I only nod when his effort was exactly what I wanted. Makes shaping an interesting exercise as you can see him listening for the click so he knows he’s on the right track and watching for the nod so he knows when he’s got it. I’m not much of a talker so Ranger is trained on a tongue click that I laughingly describe as a command to look at me and read my mind. I use it a lot for all sorts of things, heel, leave it, watch, etc. and since he can read me like a billboard he looks at me and does whatever it is I want. I find I don’t even articulate it clearly in my mind I just notice he’s getting a bit far ahead make the noise and he returns to my side in a lovely heel what I’m doing is more like an instinct or an impulse. Ranger responds to words, phrases, and gestures for cues as well as the mouth noise. I consciously train with gestures including a thumbs up for good job as I’ve known so many dogs that lost their hearing as they aged.
The very unsocialized Finna works much better for gestures than verbal cues and she relies on the clicker to tell her she’s done what I wanted. Verbal cues need to be very consistent and I find that all my verbal cues to her are taking the same form da da da da (Addams Family theme song before the snaps) “Finna About” “Finna Let’s Run” “Finna Slow Down” “Finna Come Here” etc. Since they are more sung than said the cues are more consistent and easier for her to recognize. I’m even beginning to use this form for things that are typically single word cues–“sit” is becoming “Finna Please Sit” She’s eager to please but never having spent much time with humans during her formative period she needs different cues and markers than the well socialized Ranger.
JJ says
I too just came from ClickerExpo in Portland, and this idea makes my head explode.
I was pretty sold on the concept of a click marker before the conference and am even more so now. So, the study seems a bit suspicious to me.
I also have another thought on the idea of simple vs complex behaviors. Trisha says that she might not use the clicker for an easy behavior, but might use it for a complex behavior – where one has to break a behavior into little parts. Which interpret to mean that you are breaking a complex behavior into simple parts. Which brings us back to not using the clicker because it is a simple bit???? So, no not making sense to me…
@Pike – Hey! Did we meet???? I was at the luring lab. I was owner with the Great Dane. You can’t have missed us. 🙂
trisha says
JJ: Love that Clicker Expo was so inspiring, I felt the same way after I attended. So many great speakers and great idea! Re your comment about ‘simple’ versus ‘complex’: By ‘simple I meant actions that are natural to dogs and only take one step to train. I don’t have to divide a sit into 5 or 10 or 20 steps to be able to get a dog to do it on cue.. once their butt starts to go down, the rest of the body follows. That’s compared to an action that needs to be taught by being dividing into steps: like teaching a dog “Are You Ashamed of Yourself” by burying his head under a blanket, or doing a High 5 when he’s not the type of dog to raise a paw by himself. Hope that makes sense now.
LisaH says
This is a great topic, so many possible variations depending on the dog, the handler and the given situation. I’ve been doing agility (weekly classes & monthly trialing) for several years now w/my 2 BCs and the teacher gives us various exercises to help understand communication. At times students are told to not speak during their run – usually they are the nonstop talkers at their dog and tend to confuse, bore, or distract their dog. For example, a poorly timed verbalization during a jump can lead to a dropped bar. Also, there are times the handler is flustered, calling out the wrong obstacle names so therefore becomes more flustered … and its often more likely the dog is attending to the handler’s body motion, feet, shoulders, hand, & eyes, & not to their words. Of course there are exceptions & exceptional handlers who successfully use verbal cues in order to do more distance work on the course.
As for using silence, I think its very powerful as it allows the dog to make a decision, to be a part of the process. I use silence in that I will ask for something just once, then wait for the behavior, or sometimes I say absolutely nothing and just wait for the behavior which I realize as I type is for an already trained behavior, such as sitting at a closed door & waiting for it to be opened & then to be released. Or in greeting me my dogs will run to the carpeted area next to the kitchen as I ignore them in the kitchen. (Actually when my older dog was a pup I would ignore him if he jumped on me in the kitchen- which is the 1st room you enter in the house -in greeting so I ignored this behavior & he just started to go the carpeted area & would sit & wiggle butt which got LOTS of pos attention & it became the house habit.)
We have also run the agility course w/o using our arms to lessen the habit (for some) of flailing arms that are also distracting. My teacher likes to emphasize giving the dog the clearest info in the most efficient manner.
I have experimented w/the clicker but tend to find I am not as smooth as needed. I like teaching both verbal & hand gestures (in case of distance/noisy area or deafness in old age) and my verbal requests may be single words -wait, stop come, sit – or phrases such as “stay by me”,
“go find …..”, or “go to crate”.
Lastly, I have used treats, toys & praise for reinforcers – agility seems to be incredibly self-reinforcing for my 5 year old & toys for any & everything, as well as praise, while the 20 month old will do anything for food, then tennis balls, & praise is always required.
Beth with the Corgis says
Ann-Marie, you are right about Corgis being ridiculously food-motivated. I think mine would work for pieces of tree bark. Although one of mine (the male) is unfortunately capable of weighing what treats I have against what he wants to do, and decides accordingly. So he’ll recall off of something buried in the woods if he saw me put cheese in my pocket before we left the house, but not if he saw me put Charlie Bears in my pocket. Once we had friends over and were putting out veggies and cheese and crackers. I gave him a baby carrot, which he quite likes. Then someone dropped cheese. He literally spit out the carrot, ate the cheese, sniffed around to make sure he didn’t miss any cheese, then went back and ate the carrot. The dog is as smart as your average kindergarten student. Which is sometimes, but not always, a good thing.
Back to silence, markers, and complex behaviors: I used Susan Garrett’s 2×2 method to train weave poles. You are meant to be quiet while training weaves. You don’t want your dog looking back at you for affirmation or to get a reward. Anyone familiar with the method knows you use a building-block method (a bit different from shaping) to gradually make the dog understand he is always meant to go between two posts. So, the handler should be quiet after giving the initial cue to weave. However, there is a “but” and it’s an important one: using this method, it’s crucial to give a “Whoopsie!” cue when the dog enters the weaves from the wrong side. Once the dog has associated weaving with getting his favorite toy, weaving itself becomes intrinsically rewarding and you must NOT allow the dog to self-reward by continuing to weave after a wrong entrance.
So it’s one complex behavior I can think of where no verbal cue is given to let the dog know he is performing correctly or not. If he misses a set or blows out of the weaves early, you simply don’t reward that effort. But if he goes in backwards, that is stopped instantly. So, it still requires giving the dog some feedback.
It was the only example I could think of where complex behavior is trained using only rewards and not marker words or a clicker.
Pike says
@ JJ: Loved your Great Dane! What a wonderful pup. I sat with a friend pretty close to you – in the first 2- person row next to the dog area where you guys were. Got some nice pictures of you and your large friend. Head exploding is about right as a Clicker Expo summary 🙂 It was awesome!!
Beth with the Corgis says
Ann-Marie, I wanted to add that I’ve had success with my pushy talker Corgi by doing the opposite of what he wants when he barks. Specifically, if he is barking at mealtime, rather than just stop when he barks I turn around and face the other direction. If he barks when I pick up his empty dish, I just put it down again. If he barks when I fill the dish, I set it on the counter and take a few steps away from it. Snubbing him seems to help more than just not moving. I know what you mean about a noisy house; mine both bark for different reasons, so chances are one is barking at something.
JJ says
Trisha: *Thanks* for the additional explanation. I’m going to think some more about what you are saying. I do understand the distinction you are making.
jackied says
Not done a jigsaw in ages, but I think I’m going to add your linky to my birthday list!
Interesting reading about the clicker experiment. I’m struggling a bit with my new dog (a Springer), who barks in frustration when I try to free-shape her using a clicker. Even when she does approximate the thing I want, she thinks I am clicking her for the barking part. Yet my other dog (SpringerxBC) does brilliantly with free-shaping. Maybe I should try silence with her.
Jo says
just a quick thought: maybe there is a difference between dog breeds (or better dog types – of course not every dog reacts the same). Basenjis are very “primal” dogs compared to many other breeds. Maybe there is a difference in the way primal or wild animals react then pets that are more domesticated.
I also use the clicker just for more complex behaviors, where it is really important to click/mark quickly at a very specific moment. Mainly because I don’t think it is necessary to use a clicker for things like teaching a dog to sit and I am very happy when I don’t have to have one more thing in my hand. Often am used to capacity with concentrating on the dog and holding food – so I am happy not to have to hold a clicker and use it in the right moment 😉
JJ says
@Pike – how fun to make a connection outside of the conference!
I learned so much at the Expo, especially from Kay Laurence. I think Pike is right in bringing up the idea of a visual/physical marker – such as a hand movement when training using luring and speculating how such a marker might compare to a click. (Kay really changed my understanding of how to effectively use luring by introducing the idea of a visual marker.) I would be very interested to see a study which is more involved than the one we are discussing now. I’d like to see a study that addresses several forms of marking the behavior–vs silence and a late reward.
I also want to make sure that the study addresses some of the excellent points raised by others here. For example, in the study how skilled were the teachers in using the marker? How long have they practiced using a clicker before doing the study? Do the teachers understand that the marker has to come exactly on the behavior, not after? And that if you are going to error, better to error too early than too late?
Also, does the study make the distinction of working with dogs who are already clicker savvy vs not? Because, as mentioned above, if you are working with dogs who are not clicker savvy, then you are teaching two things at once while using the clicker. That may be fair depending on what you are trying to show, but I would want to know all the parameters in figuring out what is effective. I would want to find out if the average clicker savvy dog is faster than a non-clicker savvy dog when learning a new behavior with (maker + reward) vs (silence + reward). I also think that the individual dog may make a difference. Some dogs initially start out afraid of the clicker, etc. Some dogs are more visual than auditory. Etc. All of that makes a difference I think.
trisha says
JJ and all: And I’m going to think about this more too (and read the paper again….). Always good to keep pondering, hey? Thanks for your comments, all of you, as a communication groupie this is fascinating stuff to me. I’ll keep you posted after I re-read the paper….
JJ says
Trisha and Pike: I would *totally* understand if either of you don’t have time or interest, but I figure it wouldn’t hurt to ask: I would be ever so grateful if Trisha would give my e-mail to Pike, and Pike would e-mail me the pictures you took of Duke. I brought a camera and meant to give it to someone sitting in the audience. Then I got totally stressed out about making sure Duke lived up to the very high (for us) standards of the conference (not even allowed to sniff another dog-yikes). I forgot about the camera until it was over. Not the end of the world, but if I could get some pictures, that would be a nice souvenir of the lab. Thanks!!! (even for just listening)
JJ says
On the topic of puzzles. Check out these shaped animal puzzles. I have these two (see links below) up in my house. I too get addicted when I do a puzzle. I have to limit how many I do.
I have up right now:
http://www.seriouspuzzles.com/i6731.asp
and one like this:
http://www.seriouspuzzles.com/i6749.asp
If you scroll down on the pages, you can see that there are a bunch of puzzles in this them = puzzles that are shaped like animals and have a bunch of animals inside. I think the finished product is wonderful to look at as well as put together.
jackied says
… just to add that last night I tried the ‘silence’ free-shaping method with my confused Springer, and it seemed to work better than with the clicker. Maybe some of it will come down to which suits which dog best.
em says
As a person who does very, very little tricks or formal training-(none of the things my dogs can do required any shaping and I’ve never even thought about a clicker), I don’t have much to add to the conversation. After some reflection,I will say this, though- that silence has been the GOAL of my dog training.
I use vocal commands and sounds really only to get my dogs’ attention- vocalization, even in a positive tone, is likely being interpreted as a soft reprimand- my dogs don’t flinch or jump or anything (unless they are doing something that they KNOW they should not be doing), but they generally treat my voice as a warning that they are not doing exactly what I want. What I do want they mostly pick up from body language, gestures, facial expressions, situational context etc. Sandy responds a bit better than Otis to vocal instruction (he’s exceptionally resistant, though he picks up “accident” words like ‘food’, ‘eat’ and ‘grandma’ out of conversation readily enough). They both can respond to called commands like ‘stop'(until I come to you), ‘wait'(until I say you can go), ‘go slow’, or ‘keep up’ without looking at me, but I think that they are mostly reacting to tone and context-neither dog needed hardly any training, but both will stop in their tracks 100+ yards away when they hear that sharp command. Most weeks, however, they won’t hear it even once between the two of them.
On an average day, I walk the dogs off-leash for three miles and speak to them maybe a dozen times. A good day would be half that. It sounds sort of cold, put that way, but my experience is quite the opposite- I’ve never felt more close and connected to a dog than I do on our quiet walks. They don’t hold a heel, (they will, but that requires talking), they range around me, but there is something about being able to just walk, and allow the dogs to just be with me without either egging them on or holding them back that is at once deeply peaceful and deeply moving. When we are with other people and dogs, obviously there is much more chatter, and more fun, too, but that sense of serenity gets lost.
So I wonder if dogs might be disturbed by vocalization on some level-either because they have picked up a negative association from other interactions with people or because dogs themselves tend to vocalize when disturbed, upset, or excited, not as much when calm and content. Both dogs seem to enjoy being cooed over as someone pets or cuddles them, but Otis, who had a rough start in life, used to be quite suspicious of anyone approaching him while speaking in an incongruently upbeat tone. Sandy will start wagging right on cue if someone tells her that she is good, but I’m not sure, on reflection, whether this means that she is pleased by it or simply excited…possibly excited with an overtone of nervousness and a desire to please and appease the speaker.
In any case, my markers for good behavior tend to be physical/gestural rather than vocal. I can’t remark on whether a clicker or consistent vocal marker would work better, but I can say that silent approval is more obviously apparent to a dog than many people might think, or at least, that dogs can and will work beautifully without audible markers.
Larry C. says
@em, I walk my dogs the same way, and often don’t say a word. I whistle when I change direction to give them a heads up. They will look to see what I’m doing and adjust. Because they are responsible for their own safety, they make consistently good choices. They follow the tractor when I am farming, keep away from equipment and give moving vehicles a wide berth.
Since they are only on a leash when we go somewhere, a leash is an occasion for joy. They all lead very well, which is odd, since they rarely get the opportunity to practice. I taught them all to heel without a leash and they are never tied. Three of my dogs are bird dogs, a large Munsterlander, a blue Picardy spaniel and a Labrador. For safety, their collars are so loose they could easily slip them, but they never do.
When my wife and I are holding a conversation, the dogs all congregate where we are. Our voices are music to them. They understand a lot, and sometimes they will do tasks that they have never been trained to do if I just ask them. Sometimes. 🙂 Body language will only get them so far.
Jo says
@ Em: interesting thoughts! I absolutely second you here
“On an average day, I walk the dogs off-leash for three miles and speak to them maybe a dozen times. A good day would be half that. It sounds sort of cold, put that way, but my experience is quite the opposite- I
Mary Huntsberry says
Great discussion! I’m new to commenting so please be kind and forgive me if my etiquette is not perfect. 🙂
Looking over the study’s procedure, I’d love to see video of how they did this. It turns out that the same person trained all dogs. In both conditions this person sat on the floor within arm’s reach of a food bowl where treats were dropped after retrieval from a bait bag around his or her waist. To me, the entire sequence of hand movements to get food from the bag to the bowl would be the best predictor that food is coming, regardless of click or no-click. This is especially true since the dogs were naive to clicker training. However, regardless of training history, dogs learn what stimulus best predicts food. Maybe, in that training context (without prior history of clicker training or, as others have commented, an adept trainer) the dogs in both groups learned the trainer’s hand movements were the most salient and predictive stimuli that food was coming.
So the questions I have are as follows: what happened to the click? Did it provide any salient feedback facilitating training? Was it paired with the hand movements? If so was the
Rebecca Fouts says
Not sure the second study would work out quite so well if you added any sort of distance into the mix. After all, it’s easy to treat the dog timely if you’re standing right beside it. But try to communicate what they did right with silence when they’re away from you, and thus impossible to give them the treat in a timely manner, would be next to impossible — and is one of the reasons the clicker was INVENTED. They couldn’t give the marine animal a fish while he was doing a flip – so how did they communicate to the dolphin, whale, etc that it was the FLIP they wanted, not the swimming back to them after the flip.
I don’t use the clicker for simple behaviors like down/sit, either. In fact, I only tend to pull it out for complex behaviors or those with great distance.
I also have struggled with the issue of my dogs simply becoming overly stimulated when I get the clicker out. He starts throwing behaviors at me, which while part of the game (they need to be inventive) — and yet, because he’s so aroused, he’s not always in an emotional state where he can think clearly WHY he’s getting that click. He’s so excited he’s incapable of concentrating on the point of the game.
I’ve also learned how important it is to choose very careful what you teach first which clicker, as it tends to become a default behavior — which can then become a problem later when teaching new behaviors. I taught Ammon a nose touch first — and that now becomes a problem sometimes. I try to work on a new behavior of collection over the jumps, and even without getting the clicker out, he first gives me his default behavior – touch the bar (knock it down). It then takes a long time to extinguish that so that we can actually get to work. We had the same trouble when we tried to free-shape the weave poles. His default is to nose touch them, slap them.
I’ve seen a verbal marker work almost as well as long as a trainer is VERY consistent with the intonation of the marker. They sing it almost, so it’s more consistent. I wonder, did they tape record the verbal marker for the study? So was it a matter of verbal marker or just the consistency of that verbal marker?
I guess I like clicker for some things, and find I do pull it out more frequently over the years – but I’m not a compete convert. I don’t use it for everything.
47 says
I think it depends on the dog. I have 3 chihuahuas now, 2 of them clicker trained. My first one learned formal training with no clicker. He works best with silence. My second chihuahua on the other hand, is more of a “What’s in it for me?” type. Silence results in watching her little hind end disappear off into the distance. Praise is not half as interesting as food, and the clicker wins out over all.
My third one is still with mom, and eyes have only just opened, so obviously no training yet!
Tass says
I’ve trained 3 corgis for agility and I get consistently better/quicker responses to a ‘yes’ verbal marker than I have with a clicker. I’ve never had any luck with the clicker…not sure why.