A primer on corrections

[Donald McCaig]

 

 

Dear Doggers,

 

 

Corrections are controversial, subject to abuse but necessary if you want a dog who can surpass himself; a dog who is trustworthy on or off lead in the dangerous, unpredictable, fascinating big world; or should you want a dog that will answer a mile way on rank sheep.

 

While praise is a necessary training tool it is less useful than corrections.

 

Corrections can be puzzling for a trainer befuddled by Ms. Pryor’s notions and genuinely difficult for the gentle and/or sentimental owner who confuses corrections with abuse or cruelty.

Yes, brutes have corrected some dogs to death but yes, some dogs have been ruined, and/or driven mad by ineffectual, confusing training whose alleged virtue is kindliness.

 

1. The best correction is the least effective correction. In practice this means offering slightly too harsh corrections rather than too mild ineffectual corrections. Bill Koehler was right: ineffectual corrections are “nagging” and teach the dog to ignore you and despise whatever task you are teaching it.

 

2. Corrections may teach as well as reinforce. It is perfectly okay to cuff the little puppy the first time it puts its teeth on your ankle. “Gee, that didn’t work. Won’t do that again.”

 

3. Corrections must be appropriate to the dog. I have had dogs so soft a slight shift in posture or fluttering finger was a ferocious correction. I have had others - just as lovely - who simply couldn’t hear unless you broke their focus by a shout or a threatening lunge or - in extremis- a hurled whippy stick at their feet.

 

4. Hurting a dog teaches it nothing. Startling it can produce a mental reset.

 

5. If out of temper, do not correct. Recall and put the dog up.

 

6. Mean what you say. Don’t correct unless you will follow through if necessary

 

7. You must be able to enforce your corrections. If a dog is too focused or too far way or too frightened or confused, let it go, don’t correct.

 

Since useful corrections depend on reading the dog and pretty good timing, it takes experience to master them. Beginners will not correct perfectly. Given an affectionate owner and sane dog, this is less problem than one might think. Training occurs within the context of the total human/dog relationship. I have made most mistakes people make with dogs they love but do not understand and found in practice, that dogs forgive and try again. They don’t expect a perfect leader. (I have had dogs who sighed “Oh, Dopey Donald is at it again.”)

 

Here’s a specific example. I bought Jake, kennel raised in Scotland when he was 2 1/2 years old. He’s the most physical dog I’ve ever owned: glorying in his body. Until I stopped it, he’d burst outdoors and run around the farmstead - 1/4 mile - so hard and fast he made six inch bare grass/bare snow ruts. He ran because running is carefree and adrenaline kept him stoned. Jake doesn’t like to think, he likes to run.

 

A year later, he’s a better sheepdog than he was and is starting to feel that joy.

 

I don’t let my dogs out of the house ahead of me. I don’t know what/who is waiting outside, dogs can hurt themselves bursting through a narrow doorway and I will not be knocked down. Rule of the dog pack: one dog bursts, all dogs burst.

 

I teach three commands: “Get away” (from the door so I can open it), “Wait” and later “Stay” while I open the door and stand there. “Okay” is the release.

 

Jake liked “Okay” but “Get away” and “Wait” were nonsense syllables.

 

“Get away”: Swat the dogs with my cap, repeating the command until all the dogs are off the door. “Wait”: Cap, rough voice tempt them with a part open door but shut it when they make a try. After they can “wait” properly, put them on a down stay and let them out by name. If a dog (aka Jake) bursts out of turn, quick-shut the door so he runs into it. Growl return him to his place, call another dog by name.

 

Jake still doesn’t like to think, he prefers to run but these days I murmur “get away” and he does, “Wait” or “Stay” and he does. I open the door wide, step outside, yawn without looking say “Fly” “Peg” or “Jake” and that dog and none other bursts past me.

 

Affection, consistency, corrections, patience.

 

Donald McCaig

 

 

 

[ourwully]

Dear Donald

I always enjoy your comments, especially when they have just a hint of sarcasm. I always feel like you are somewhere in the middle battling those pendulum swings. Anyway, this primer is excellent. If there was a section on this forum for beginners, and this would probably would be a good thing, this would be a sticky for sure.

 

As you noted all dogs are different. Now that Juno is almost two I have come to the conclusion that I have not been strict enough with her, especially of late. About 6 months ago I was counting my recalls and making sure I was doing about 20 a day. When I mentioned this on the forum many readers suggested that this was too much. At the time her recall was improving steadily so I cut back a bit and stopped counting. Despite constant training since then, I think her recall has deteriorated. I was reading a bit from Koehler yesterday and I think Juno is thinking but not the way I would like her to. With no negative consequences to her non responses to recalls she is making the choice to chase squirrels rather than to come to me. In her case, I just don't think the positive reward can be high enough.

 

What is really interesting is that we often walk with Callie, a 3 year old black lab so she has been present for a lot of Juno's training. Now Callie is highly motivated by food so when I call her or Juno she comes to me like I am the most interesting thing on the planet. One excellent suggestion I got on the forum was to use a special recall (whistle) infrequently with a super high value treat only when I knew Juno would come. I have been building this recall faithfully for over a year but even it has not been perfect. However, if I blow the whistle when Callie is around Calle stops immediately and bolts to me every time. Her recall is 100%. She is a much different dog!

 

Tomorrow I am going to start with a set of private lessons to try get things back on track. I am also going to keep your primer in mind as I certainly don't want to be nagging for the next ten years!

[waffles]

Ourwully- you may benefit from putting Cally back on a long line. This makes it so she has to listen and you have to reinforce- she doesn't come when called, give a little tug of the lead to remind her, if she still doesn't then start reeling her in. She no longer gets to choose to run off after a squirrel.

 

You can buy 50 and 100 ft sections of line at Lowes or Home Depot-parachute cord or thin rope, any light weight will do. A leash clip is about $2, just tie the rope to it and I like to wrap electrical tape along the first few inches after knotting.

[ourwully]

Thanks Waffles. I have already done that. I put her on the long line this morning. It really makes me sad to do this but for her own long term safety I don't see any other choice.

[geonni banner]

Dear Doggers,

 

'snip'

 

Donald McCaig

 

 

 

What he said. In spades. Can this be made permanently accessible on the forums somehow?

[D'Elle]

Although everyone here probably knows I am a big fan of Karen Pryor and not at all a fan of Koehler, I agree with most of what Mr. McCaig has said. My method of teaching Wait At The Door, minus the cap-swatting, is virtually identical, (break from the Wait and you will run Smack into a closed door), and all of my dogs are very good about that. Same with wait-to-greet when someone comes to the door. What I have come to think about training is that the best approach is more of an Action-Consequences concept than a Punishment one. If I am angry at the dog or want to punish, I have no business doing anything but going away until I cool off. My dogs learn that this behavior results in happy things happening for dogs, and that behavior gets them nothing that they want, and may get them something they do not want. This seems to work pretty well.

[chrisandgabe]

Also teach "wait" in a similar fashion, although use more body positioning to get the dog to move, and I just conceptualize the dog running into the door as withholding the functional reward (going through the door) instead of a true correction (more of a negative reinforcement than positive punishment). But yes, you don't want the dog blowing you off, and effective corrections are an important part of that. Thanks for the great summary.

[Maralynn]

Can you expound on why you think corrections are more useful than praise?

 

They both teach the dog - one what you want, the other what you don't want. It seems to me that much of what you've said about corrections can be applied to praise. With both timing and context is paramount.

[GentleLake]

Can you expound on why you think corrections are more useful than praise?

 

I would be interested in hearing more about this as well.

 

Or, more accurately, I'm interested in the reasoning behind the assertion that praise is "less useful" than corrections.

 

How can reinforcing desired behavior be less helpful, especially in a context where, unlike working livestock, the desired behavior isn't necessarily self rewarding?

[Donald McCaig]

Dear Doggers,

Thanks for your question. Jake and I are leaving tomorrow for Seattle: family and the Vashon Island SDT. But I'll answer as soon and as best as I can. Donald

[Root Beer]

I'd like to ask a question, too. A separate question!!

 

I trained Bandit's recall without corrections. No leash or long line pops, no verbal reprimands, no shock, nothing that communicated, "wrong". In the early stages of his training, when he didn't come when I called, I simply went to him and - very pleasantly - said his name and had him come along with me, and reinforced him (usually with play) for coming along with me. The early stages didn't last all that long. He got the idea pretty fast.

 

We hiked off leash in the woods this summer and he played off leash on the beach, and his recall was the fastest and most reliable of any of my dogs. He is now allowed off leash in the unfenced side yard (for some reason he has chosen that area as his favorite toilet) and he easily calls off the neighbor's cats - and he is VERY interested in those cats! But the second I call him, he's back with me.

 

I'm interested - when you have time - in knowing how you can say that reinforcement (in his case it was food and play more than praise, although praise is what he gets now when he recalls off of things in real life) isn't incredibly effective.

 

According to the theory that corrections are necessary in order to train a reliable recall, I shouldn't have anything resembling a reliable recall. But I have one so good on this dog that he surprises me sometimes.

 

If toys/treats/play aren't plenty effective to get the job done, then how did the job get done so well? I'm not trying to be snarky here, nor to be combative. I really am interested in your answer because I am trying to understand where you are coming from.

 

In the meantime, have a wonderful trip!

[geonni banner]

While Mr. Mc Caig is sheepdogging in WA I'll venture this:

 

Praise may teach a dog what you want just fine, but a correction teaches them that you expect to get it - even if he wants something else.

 

Corrections and punishment are not the same thing. Corrections illustrate what you don't want. Punishment illustrates that you are a jerk - and nothing else. A correction conveys information. It does not hurt, frighten or undermine the dog's trust in you. If it does any of these things, you are not doing it right. Punishment does hurt, frighten and/or undermine the dog's trust in you, and tells him nothing useful.

[D'Elle]

I think Root Beer is right in that positive reinforcement is incredibly effective. I believe in it strongly, and it is my first approach.

And, sometimes it is also good to utilize a correction if needed.

 

My experience is that positive reinforcement is the way to go when training a new behavior of what to do. Corrections are sometimes needed in training what not to do.

For example, I do Musical Canine Freestyle. I teach my dogs their moves using *only* positive reinforcement, never a correction. Corrections of any kind are counter productive when training something like freestyle, and are unnecessary. If the dog doesn't do what I want I just assume he doesn't understand yet, and take a step back to the point where the dog was doing what I wanted and work forward again. If the dog blows me off it means he is tired of this for now and we stop until later. Freestyle needs to be only fun.

 

A different example: Teaching a recall, I use positive reinforcement.... the dogs learn that when I blow the whistle it means something delicious is being handed out and the sooner they get to me the sooner they get the food. It works great, and my dogs have good recall (even when not reinforced with food). So, the positive reinforcement trained the dog what I want him or her to do.

However, if one day a dog decided that what he is doing is more important than to come get a treat and he blows me off, I go get him gently but firmly by the collar, walk him back to the house or vehicle, and he gets no more running free for that time period. This is negative reinforcement, or correction. This I do because recall is important; it is not just for fun.

 

I don't use a lot of corrections, and most of the time my corrections are nothing more than a low-voiced "uh-uh" or just "ah".

[CptJack]

By my definition of 'correction', going and getting the dog IS a correction. It's a consequence. He was made, however gently and pleasantly, to come along with you when he didn't want to. Likewise, reeling the dog in on a long line, even if you're crouching and calling and gentle, is a correction. Withholding a treat can be a correction for some dogs. No reward markers. Whatever.

It doesn't all have to be force, pain, fear, compulsion, or typically aversive things.

[Root Beer]

Sure, it's a consequence - I agree there. But hearing your name said in a happy, excited, inviting tone, and then walking along companionably together, and then getting to play - it's a GOOD consequence. Instead of making him come when he didn't want to, I changed the picture so he now wanted to.

 

I once asked a group of "balanced" trainers if they would consider this a correction: I cue my dog to spin. The dog stands still and stares at me. I take out a treat and lure the spin, and give the treat. Then I cue the spin again, the dog does it and I give another treat. Was that a "correction"? After all, I "made the dog correct" and I got him to do the behavior. You could even go so far as to say I "made him do it". But . . . the answer - a resounding "no - that's not a correction"! And I agreed with them. Now, it is certainly information, but I doubt there are many who would define that choice of delivering information to be a "correction".

 

Personally, I don't consider consequences that the dog enjoys to be "corrections", even when those consequences convey information. I know there are people who do, but it doesn't make logical sense to me. And the line has to be drawn somewhere or any cue or directive given to a dog becomes a "correction". And, of course, that's not the case.

 

So, I stand by my assertion that I taught Bandit's recall sans corrections. I would not say "without consequences", but the consequences were always highly desirable consequences - consequences that communicated, "it is very much worth your while to do this". Consequences that I would actually call . . . reinforcement.

[ourwully]

I am finding this back and forth banter quite interesting because I have been trying to train Juno in the most positive way I can. For the recall, I have used roast beef, chicken and liver and I've jumped up and down like I've won the lottery. I've followed the Really Reliable Recall to the letter and I have tried every suggestion given to me on this forum. You name it I've tried it but the squirrel still wins more often than not! How can this be? The simple answer is that I am inexperienced. My timing has been off, my consistency has been questionable, etc.

 

I am convinced that positive reinforcement is the best way to train a dog, but I am also convinced that, for some dogs and for inexperienced trainers, it is a very difficult way of doing things. When Root Beer asks how it is that she has been able to train Bandit with reinforcements only, I would say that it is probably because she is a very competent and experienced trainer!

 

I would also surmise, from reading many of his posts, that Donald McCaig, despite his advocacy for corrections, would be considered to be very positive in the eyes of most dog owners.

[geonni banner]

I once asked a group of "balanced" trainers if they would consider this a correction: I cue my dog to spin. The dog stands still and stares at me. I take out a treat and lure the spin, and give the treat. Then I cue the spin again, the dog does it and I give another treat. Was that a "correction"? After all, I "made the dog correct" and I got him to do the behavior. You could even go so far as to say I "made him do it". But . . . the answer - a resounding "no - that's not a correction"! And I agreed with them. Now, it is certainly information, but I doubt there are many who would define that choice of delivering information to be a "correction".

Does the dog being cued to spin know the cue? Has it mastered the behavior? Or is it a dog that has never been taught to spin on cue?

[CptJack]

For me, the luring and spinning to treat wouldn't be a correction, either, but going to retrieve the dog and bring him back, however happily, IS. I am not sure why the distinction in my brain, exactly, though, and couldn't begin to explain it to you except that to me going and getting the dog, however happy you may be, says 'nope, you've got to do this' in a way that simply luring the spin around again doesn't. And again: I don't know why, but frankly since 98% of my training is done with nothing more harsh than a no reward marker, it probably doesn't matter overly much, anyway.

 

I am finding this back and forth banter quite interesting because I have been trying to train Juno in the most positive way I can. For the recall, I have used roast beef, chicken and liver and I've jumped up and down like I've won the lottery. I've followed the Really Reliable Recall to the letter and I have tried every suggestion given to me on this forum. You name it I've tried it but the squirrel still wins more often than not! How can this be? The simple answer is that I am inexperienced. My timing has been off, my consistency has been questionable, etc.

 

I am convinced that positive reinforcement is the best way to train a dog, but I am also convinced that, for some dogs and for inexperienced trainers, it is a very difficult way of doing things. When Root Beer asks how it is that she has been able to train Bandit with reinforcements only, I would say that it is probably because she is a very competent and experienced trainer!

 

I would also surmise, from reading many of his posts, that Donald McCaig, despite his advocacy for corrections, would be considered to be very positive in the eyes of most dog owners.

 

I'll let you in on a little secret. I am an experienced dog owner and have done quite a bit of training with dogs over the years, HOWEVER:

I've never done an ounce of official recall training with 4 of my 5 dogs. The one dog I did any with was the GSD-X, and that amounted to putting him on a long line if he blew a recall and using it to reinforce (bring him in and treat) for a week or two after he blew it. He hasn't done that in probably a year now.

The others? They're velcro dogs. They want to be close to me. I'd randomly reward attentiveness to me in general, and when off leash if they checked out, I'd simply say 'Bye' and walk away from them. Sometimes, I'd outright hide from them. They were dogs who weren't happy about losing me, work hard to find me (or catch up to me) , and then when they did I'd praise and reward and carry on with the day. That was literally all I had to do with any of them.

 

That wouldn't work with a dog who was more prey driven than handler focused. Some of the equation is the DOG and their drives and desires, too.

 

Basically what I'm saying here is don't necessarily give those who have dogs with perfect shot out of a canon recalls too much credit. Sometimes it's the trainer, and sometimes it's just the DOG.

[rushdoggie]

I am finding this back and forth banter quite interesting because I have been trying to train Juno in the most positive way I can. For the recall, I have used roast beef, chicken and liver and I've jumped up and down like I've won the lottery. I've followed the Really Reliable Recall to the letter and I have tried every suggestion given to me on this forum. You name it I've tried it but the squirrel still wins more often than not! How can this be? The simple answer is that I am inexperienced. My timing has been off, my consistency has been questionable, etc.

 

I am convinced that positive reinforcement is the best way to train a dog, but I am also convinced that, for some dogs and for inexperienced trainers, it is a very difficult way of doing things. When Root Beer asks how it is that she has been able to train Bandit with reinforcements only, I would say that it is probably because she is a very competent and experienced trainer!

 

I would also surmise, from reading many of his posts, that Donald McCaig, despite his advocacy for corrections, would be considered to be very positive in the eyes of most dog owners.

 

Also, some dogs take time. I have owned and trained a lot of dogs. Like, more than 12 (I used to be one of those crazy has 6 dogs at a time ladies). I have also fostered literally dozens of dogs. Teaching a reliable recall was a primary goal.

 

The dogs I trained with a long line and a collar correction learned a recall pretty quick. The dogs I taught with cookies and then a follow up correction (usually walking them down) learned a reliable recall pretty quick. The dogs I trained with cookies only learned a reliable recall pretty quick.

 

Then there was Jasper. He would not come in from the yard when called.He came in a formal context, he came when dragging alight line on outings, he came in the house. I did it all right for a year. I tried a walking him down with disastrous results (he freaked out and wouldn't come near me and I see shadows of that to this day). I fed the other dogs for coming in front of him. I rewarded lavishly. I got mad. I teach dog training classes and compete with high scores in competitive obedience. I had not one but TWO Siberian huskies with a reliable recall (in most contexts, I am not stupid). This is a smart, trainable dog. WTH?

 

Then one day, after many frustrated attempts and feeling at wits end, I called him and he came. Right away with a smile and a wag. The next time, he came. And now he ALWAYS comes, and can be called off an opossum, the neighbors kids engaging him through the fence, his favorite neighbor lab buddy engaging him, off a cat the streaked across my front yard in front of him.

 

It just took time.

 

Dogs are not identical, they don't come programmed. They are individuals and sometimes learning doesn't happen overnight.

[CptJack]

^ that actually said what I was trying to say, or a facet of what I was trying to say, only much better and more clearly.

There is *nothing* I would like more than to take credit for the fact that Molly has had a beautiful recall and successfully called off birds and squirrels and cats and ducks and deer and everything in the world since she was about 6 months old.

I can't.

I didn't work at that, I didn't train that. Molly just did that. . She's never blown a single recall in her life and it's not because I'm an amazing trainer, it's just because it's what Molly *does*. We don't even have a formal recall command, just a sort of casual 'come on, then'.

 

If it was all down to me being an incredible trainer, we wouldn't have a dozen holes in our training I could point to right now - and she wouldn't be a reactive nightmare around other dogs.

[Blackdawgs]

I know that a lot of people are not fans of this site, but I think that this guy makes some very interesting and unfortunately valid points including the need for the right dog: http:www.leerburg.com/allpositive.htm

 

I think that for purely positive (eg correction-free) training to have even a chance working (and I am talking total reliability in all environments), the trainer must have an excellent understanding of reinforcers and the ability/patience to completely control the environment. The dog is never allowed to have the thrill of the chase to begin with, so it just doesn't think about taking off after a rabbit. I think that most people (including myself), and certainly not the average pet owner, are probably not willing/ able to control the environment to the extent needed. And in the end, one still needs the right dog.

 

For a time, I worked with a cross over trainer for my rescue Border Collie. I don't think that this trainer (who had OTCH'ed some dogs) ever worked with a dog who would run thru a wall of fire to get what was on the other side. One day, the dog appeared to be committed to the dog walk--had her front feet on it and then she whirled, bolted, run thru a wire fence, and started chasing some sheep. In the end, I tackled her.

 

I used to scour the grocery store looking for a magical food that would somehow overshadow the dog's interest in the environment and predation.

 

I seriously considered using a shock collar on her.

 

Over time (years), her recall improved on its own. I stopped trying to train it. These days, I can frequently call her off the fence fighting neighbor dogs. She just started wanting to be with me.

 

But, I still would never in a million years trust her around livestock.

[rushdoggie]

I also have found, that many (certainly not all) people who go the more forceful method (and no not every correction is forceful) do so because they want it NOW. They don't want to take the time to build that rapport, that bond, that trust, that communication that creates (as discussed in another thread) a mannerly dog.

 

You can get that with one perfectly well timed and appropriate correction for some things without fallout. You can't for others.

 

FWIW, as a "cross over trainer" who started with a choke chain and scoffing at cookies 20 something years ago when I started training dogs, that I don't get any behaviors more reliably when taught via corrections vs cookies. I reliability with relationship.

[Blackdawgs]

I agree with the need for NOW. There is a "balanced" trainer in town who is pretty much putting prongs AND shock collars on everything. The shock collar remote is given to the same inexperienced owner who would be a timing disaster with a clicker (but of course without the fallout) and the net result is a non surgically lobotomized dog. But the owner is thrilled because the former pain in the ass dog is now "behaving" after 1 training session.

 

Just like the patient who wants a magic pill because it is easier than diet and exercise

[geonni banner]

"I also have found, that many (certainly not all) people who go the more forceful method (and no not every correction is forceful) do so because they want it NOW. They don't want to take the time to build that rapport, that bond, that trust, that communication that creates (as discussed in another thread) a mannerly dog."

 

This is interesting to me. I have been criticized for using what amounts to the Koehler method on my dogs as being "so mean!" and various other things of variable printability. But those same individuals, many times in the same conversation, will remark on how thoroughly my dogs seem to trust me. Many of these dogs were rescue dogs who came to me with a considerable baggage of fear issues. They are free-spirited dogs with a sense of humor, who have a fun time with training and in general. The very last descriptive that would apply to them is "robotic" or "lobotomized." But they are reliably obedient, and generally well-mannered.

 

And no, I don't see how teaching a behavior in thirty sessions is better than teaching it in four if you have the skills to do that, and you end up with a happy dog with rock-solid trust and reliability.

 

It is absolutely not the same as the patient who wants a miracle pill for fatness (or fat-headedness).

 

I choose my dogs carefully, and with an eye to a character and temperament that will be most suited to the way I train. In cases of dogs that were "a pig in a poke," (usually a dog between a rock and a hard place) I keep them if it works, or place them with someone who will be better suited to their ways than I, if they have a character incompatible to me and my training style.

 

In the end, I suspect, there are as many ways to effectively train dogs as there are dogs to train, and people to train them. I say, use what truly works for the dog in front of you, and let others do the same. When I got my latest, I observed her closely both up close and from a distance for a few hours. She looked like a really solid, tough-minded dog who had been through some bad treatment, (read: stupid and inappropriate - heavy-handed rather than actually brutal) and was a bit shaken and unsure because of it. So far, she has borne out my assessment. After three weeks of decompression I started her training. She is good, and having fun. She's a quick study and a joy to train. What else matters? Move over, Just Jumpy!

[Root Beer]

Does the dog being cued to spin know the cue? Has it mastered the behavior? Or is it a dog that has never been taught to spin on cue?

 

Either!!

 

Now, typically if the dog knows the cue, I would actually toss a treat to break off what we are doing, and then cue again. If the dog still didn't respond (and I were positive that there were not a physical issue in play - with Dean, for instance, if his hip is sore, he won't spin, and I know that's because he's feeling sore and won't ask him again), then I would use a lure or two (or a big hand cue) to remind the dog what I am asking for.

 

For a dog who is just learning the cue, I am going to be mixing up good clear hand signals (not usually food lures at the cue stage) with the verbal.