At the first lesson for Lola and her housemate Archie, Kelton reaches into his own pocket, then says 'Archie, sit', and asks the owner what he did first. The answer changes everything the family thought they had taught. Archie does not know the word. He knows the hand. It is the moment a lesson stops being about the dog.
Archie sit what did I do first yeah and then I said Archie said
The chapter in one paragraph
A dog answers the earliest reliable signal, not the word the handler thinks he is giving. That signal might be the hand going to a pocket, a step away before 'okay', slack in the lead, the handler's settled gaze, or the fact that he is standing in the room. So until the hand, the body, the leash, the reward and the handler himself have been taken out, nobody knows what the dog knows. The trainers find out cheaply, with hands in pockets, the reward on the far side of the body, the leash out of it and the door closed. They also break their own patterns on purpose so the dog has to listen to the word. Once a command has been given, it is finished in full: all the way down, all the way in. What you accept is what you will be offered next time. A behaviour the dog offers on his own is not held to that. A down offered for a sit is usually a guess between two words he knows, and he is shown the difference calmly. The bar itself moves in small steps. Start with a rough version the dog can win. Say the criterion out loud before the rep. Lower it when he keeps failing, but never fake a success. Teach in a set order, one layer at a time. How good is good enough is set by the dog's job and the family's life, and reliability counts only as far as it has been tested.
Principles
The dog answers the earliest reliable signal in the chain, not the word the handler intends. A reach into the pocket, a step before 'okay' or a hand on the lead becomes the real cue if it keeps coming first. So the discipline is the handler's: say the word first, with the body still, and move only afterwards.
Said as a rule on filmOn film
you have to do the first thing in the chain has to be the one that you want to mean something okay right so you have to kind of discipline your body to not do other stuff
So, you always want it to be the word. Same thing with like sit down, place, whatever it is. You want to make sure that you don't do other stuff first.
So that important principle is that the first thing in the chain is what becomes the predictor for the dog, right?
they track whatever you do first.
anticipation like that is just always based on prediction so the only way to get rid of the anticipation is to remove the antecedent
Concepts: Hidden help, False reps: the dog reads the chain, not the word
Exceptions
- Help is not cheating. The trainers call it training wheels and say so while they use it (Roxy, watch 08:25). The rule is about order: the word goes first and the helper after, so the word comes to predict the helper and can then stand alone.
You find out what a dog knows by taking the help away, one piece at a time. Hands in pockets and say the word once. Carry the reward on the far side of your body. Take the leash out of it. Then take yourself out: leave the room, close the door, watch through a gap. Expect it to look worse while the help comes off, and don't read that as harm.
Said as a rule on filmOn film
that's not good it's supposed to be verbal only like my hands in my pockets I say sit and she's supposed to say you know
uh didn't use the ball as a lure or anything like that it was in my pocket um uh on the opposite side of my body so if he was looking for the ball he would have come across my body
with a toy or something like that um if they need guidance every time they don't know how to do the command on their own and that means that they don't know it right
Now when we go off leash, of course We're not actively like only using leaf pressure
What I worry sometimes is that maybe people listen to you because I'm here.
sometimes it looks ugly at first and then it looks smoother as you add help. And then sometimes you give less help and it looks ugly again,
Concepts: Hidden help
Exceptions
- Place may keep its point, because the cot moves around the house (Dutchess, watch 19:58). Whether every visual aid must go is open.
- For a very leash-sensitive dog the trainer keeps leash pressure out of the cue from the start rather than fading it, because the leash could quietly become the command (Mazi, watch 06:04).
- Whether to push through the ugly stretch or stop and come back to it elsewhere is open (see conflicts).
Because dogs pattern-match, the trainers break the pattern on purpose. They alternate commands, ask for the one the setting argues against, change their own posture and position, and vary the reward. A rep the dog could have predicted is not a rep of the word. The trainers call the deliberate pattern-breakers false reps. They are for proofing commands the dog owns, not for teaching new ones, and the dose is small.
Said as a rule on filmOn film
you gotta hit it with some like like false reps, right? So some ones that break that prediction, you gotta break that pattern
you're doing them one after the other so he has to like it's kind of like you know that game where they do like like Simon says or something like that
Try to make it as random as possible.
one thing to avoid is doing something like this. I throw the ball. He brings it back. I make him do a command.
so if you find yourself you did like 20 reps unless the session was like an hour long
Concepts: False reps: the dog reads the chain, not the word
Exceptions
- A predictable rep asked as a measurement, not as training, is legitimate. Luna's handler asks for a down he knows she will give in order to read her state (watch 07:22).
- What to do when a dog anticipates (refuse it, redirect, or leave it) has no house rule. See conflicts.
A command you gave is finished in full, because what you accept is what you will be offered next time. Mark the partial, ask again and wait, rather than paying the almost or helping him the rest of the way. Standards held when it does not matter are what make the response there when it does. Slow is not wrong: a tired dog gets more time, not a lower standard. And the criterion applies only to commanded reps, not to what the dog offers on his own.
Said as a rule on filmOn film
want you to notice there is is that uh if you want it to like look correct and you want him to do it right and come all the way to you, you can't accept less.
Zeus. Down. Down. No. We've practiced this so much. Down. Come on. All the way. All the way, Zeus. All the way. All the way. We're going to keep doing it until you go.
So then it's worth it to me to punish, to penalize when it wouldn't be a big deal so that when it is a big deal, they listen.
merely just couldn't let him win that exchange because then he kind of develops a little bit of a thing that i'm like oh i can just kind of like not do anything and it works
her a while to sit down or lay down. So, um now if she like sat and immediately went down, no, I wouldn't allow that. But she's doing good right now.
okay I'll allow it if I had actually put her into a legitimate like sit then I would have put it up so like now I'm gonna put her in a legit sit
Concepts: Don't accept the almost
Exceptions
- Stormie's trainer draws a workable line: reject a rep whose shape would change the word's meaning (a down that rolls into a belly-up), and don't enforce limb placement (watch 13:32). It is one trainer's line, not a stated house rule.
A dog who lies down when asked to sit usually knows both words and is guessing. The cause is over-practice of the down, recency, or reading the pattern. The fix is clarity, not force: mark it softly, reset him to a stand, ask again, and release the instant his rear touches the ground. Mix sits and downs so he has to listen to the word.
Said as a rule on filmOn film
The dogs can confuse those two commands all the time. So we just want to make sure and refine that for King and see how it goes.
And you notice that like my tone is not like, like really authoritative and like, no, you did the wrong thing. because it's just a mistake right um she's obviously trying to comply
as soon as his butt hits the ground before he has a chance to lay down as soon as his butt hits the ground release right because then what you're getting is you say sit he completes the position properly and gets released
I had him down too much in our last training at the walks, so I think that's one of the the things sometimes it's kind of like learning spanish and french at the same time
so like in the house maybe like instead of maybe you just do it down a lot right because right there if you say sit and she goes into it down you will teach her that she can do both
Concepts: Swapped commands: keep sit and down distinct
Exceptions
- Thor's family heard three answers on film ('fine for a pet', a gentle correction, 'the down always has to be a down').
- Rule out a physical or context cause before calling a persistent swap a knowledge gap. Shadow the Siberian Husky always held his sit indoors (watch 00:12).
Before each rep the trainer decides what one good repetition looks like and says it: one thing, small enough to win. After the rep he says what happened and what the dog learned. When a block goes badly, the next rep is made easier and shorter, not repeated.
Said as a rule on filmOn film
All my criteria was that he stopped when I stopped.
walk with him i'm gonna lower the criteria next time get a little bit of a shorter rep in
Just defining the criteria for a roddy like him.
in proofing it's an opportunity to learn right so what did he learn he learned that look you have to
Concepts: Say the criterion out loud before the rep
Exceptions
- It is not known whether saying the criterion aloud is taught to staff or is a habit of the handlers seen here.
Raise the bar in small steps. Start with an unfinished picture the dog can win, and raise it only a little at a time. When he keeps failing, drop it as far as needed for a real success. Never fake a success by releasing him just before he fails. Each session starts below yesterday's peak.
Said as a rule on filmOn film
like the maximum criteria this is exactly what i want right the finished picture i start with an unfinished picture and i teach that first
That's kind of the rule. If a dog's failing at it, just lower the criteria as much as you need to to get them to succeed.
As long as you don't smoke and mirrors it, as long as they actually succeed.
So what I wanted to give her there is, one, it's a successive approximation, so I'm not forcing her to do, you know, like a straight to me recall right away
you're probably you know let's say you trained yesterday right your peak is your the start of your training the next day is not going to be equivalent to the peak of your training a day before
Concepts: Raise the criterion in steps
Exceptions
- When a partial is accepted as a step and when it is refused (principle 4) is not stated on film. Kelton has not yet answered it (see conflicts).
- One element can be held firm while the rest is loose. Lucia's recall accepted a rough approach but required the finish: a sit and one second of eye contact (watch 05:23).
The order is part of the method. Walking without pulling and giving to leash pressure come first, holding commands next, and heel last, because heel is the least natural command and is built from pressure and focus. Within one command the phases are learning, clarity and proofing, with distance, duration and distraction added one at a time. With a nervous dog, one fear at a time.
Said as a rule on filmOn film
before we do any introduction to heel or like long distance down and stays or sit and stays or big like store situations because if we jump right to that then the foundation won't be done
because it makes it easier to teach them everything else if when they feel pressure they go with it
so it's always the last command that i teach them yeah um and i think it's the most unnatural command for a dog
what people do is they think a dog knows something but the dog doesn't really
Of course, when we train a nervous dog, we do not introduce all of this at once.
Concepts: One layer at a time: the order is part of the method
Exceptions
- How fast to move through the phases differs. Levi's handler budgets about a week per phase, while Thumper's coach wants leash help gone as fast as possible.
How good is good enough is set by what the dog is for. A family dog does not have to be perfect, but he has to be really good at the things the family's life needs: stopping when told, coming back, and a walk that isn't a fight. A service or protection prospect is held to that job's stricter bar.
Said as a rule on filmOn film
This is a pet dog. It does not have to be perfect, but it does need to be really good.
if we were in a competition we failed but we're not in a competition we're thinking about live functionality for the owners
action but clip so it's gonna be a service dog so just cuz she sees some some action, eh eh, doesn't mean she can break command.
And I tell all my trainers that. I'm like, look, you need to prioritize those things.
Concepts: The standard is set by the job
Exceptions
- Who sets the standard, the household or the trainer, is layered on film. The family chooses what matters and the trainer how well, but no file says so.
A dog is only as reliable as the hardest situation he has actually been tested in. A good run in easy conditions (night, a fenced yard, the trainer's own hands) is discounted until it is repeated where he could fail. The test is built to match the real deployment, and the backup is not taken off on a hunch: 'you want to know not hope'.
Said as a rule on filmOn film
sometimes a dog can look really good because nothing's happened. You know what I mean? So like maybe two out of 10 times the dog will bolt. But the thing is if you only see the eight out of 10 times you're like yeah good.
it needs to be in a place where he could just leave and because they know they know if they can just leave or not right dogs are not stupid right so like if he's in the training yard he knows he can't get away
have him go in and out of the crate for them and follow commands with different like skill levels for handlers so not all professional dog trainers
it's not something you just kind of like try out you want to be confident before you do something like that you want to know not hope you want to know not hope
Concepts: Reliability is what you have not tested
Exceptions
- Where failure is dangerous or can't be undone, the trainers refuse to find out rather than stage it. Zeus stayed on leash because of his size and the liability.
The trainers grade help rep by rep and do not count a helped rep as a success. Sage's sessions use six steps, from 'doesn't know it' to 'verbal and reliable'. A handler redoes a down she 'smoke and mirrored'. Another points out the rep that happened with nothing attached.
Shown, not saidOn film
she knows the word but she's not confident that she knows the word she still needs escape yeah to get on she's not in an avoidance paradigm yet
So I feel like I accidentally smoke and mirrored right there, so I'm gonna do it again. Phoebe.
So he was chasing a bird right there. Teddy, sit. Okay. Okay. But he recalled right off of it with no leash tension and no e-collar.
That right there was was verbal only. Um, I didn't have to do much of nothing and he figured it out real quick.
Concepts: Hidden help
Exceptions
- The grading scale is shown on Sage and the correction count on Dolce. No trainer states either as a house instrument.
What owners are told
Find out which commands your dog really knows. Hands in your pockets, stand still, say the word once. Keep any reward on the far side of your body, take the leash out of it, then leave the room and check. Say the word before you move, every time. When you ask for something, get all of it: all the way down, all the way to you, back to the spot if he broke. Mark the almost and ask again rather than helping him the rest of the way. A tired dog gets more time, not less standard. If he lies down when you ask for a sit, stay calm, bring him back to standing and release the moment his rear touches the ground. Mix sit and down so he has to listen. Whether you accept a down for a sit at home is your call. Keep your sessions short and your repetitions few. Expect it to look worse as help comes off, and don't read that as harm. Don't count a good week as proof: test him where he could fail, with other people holding the leash, and keep the long line on until you know.
On film
okay again, but I said it first, right? Verbal first.
less. Because if you do, then he'll be like, "Maybe I can just come up to there and then like lay down and she'll be cool with that." Cuz that worked last time.
A little pull towards me, she got excited, but that's okay. There you go, that's why we're testing it out.
What trainers know
Grade help, don't assume it. Anything that indicates or guides is help: the handler's body, a hand resting on the remote, slight line tension. Word first, then the helper, then the word alone now and then to check. Wait a beat after the cue before helping. Don't count helped reps. Check the release as well, since 'okay' must not depend on the handler walking off. Take the leash out of the cue before go-home. Add distance, duration and distraction one at a time, and only after verbal fluency. Fade the handler: other room, closed door, a gap to watch through. Deliberate variation is for proofing, not teaching. Twenty reps in fifteen minutes is too many. Mark and re-ask on an established cue, and don't pay or help a partial. 'All the way' works as a separate cue for the missing part. Repair a broken position where it was asked. Reject a shape that changes the word's meaning, and don't enforce limb placement. A half-done response is a diagnosis of what the dog is keying on. For sit and down: separate over-practice, recency and anticipation. Reset, don't place, and release on completion. Keep working words that could collide apart. Before each rep, one winnable criterion. Define it before enforcing it. Declare the terms before a staged test and the limit before the attempt. Lower and shorten after a failed block. Build in order: no pulling, yield, holding commands, heel last. For pressure-sensitive dogs, pressure goes first onto a known low-stress cue. For nervous dogs, one layer at a time. Test in the real deployment with non-professional handlers before calling a dog reliable.
On film
that would be help because all of those indicate to her that she needs to do the thing And some of those actually guide her into doing this thing.
so that's how I didn't do any least pressure just said it and gave her a minute to give her a second to kind of process it
I probably will not have time to start introducing heel today just because introducing all that at one go is a lot for most dogs and I don't want to mess them up.
but what do you want him to think in this case is the right thing to trigger him going into that motion into the down it's the word down not me getting up
In the dogs
Lola
At the family's first lesson Kelton uses Lola's housemate Archie to show that the sit they thought he knew was a sit on the reach into a pocket. At the second lesson he warns that eight clean reps can hide the two in ten when Lola bolts, and calls her in at a narrow point instead of risking the chase.
On film
no don't look at me like that uh but yeah so um really the question is do they know the word or do they know something else right
Roxy
The whole ladder in one household. The handler gives up pointing, names the pull-down as training wheels, and sets 'verbal only, hands in my pockets' as the finished standard. Then he corrects Roxy from behind a closed door, watching through the door hole.
On film
I just happen to be looking through the little door hole this time when she went to get off
Kodak
Two copyable self-checks: the ball in the pocket on the far side of the body, and a warning that behaviour looks ugly, then smoother, then ugly again as help comes off. Kelton calls the second ugly stretch an extinction burst, and says it is not a sign of hurting the dog.
On film
and it's important that you don't give up at that point or think that you're hurting him or or that, you know, you're scarring him or something like that because really all you're doing is clarifying it
Sage
The measuring instrument, six grades of help from 'doesn't know it' to 'verbal and reliable', and the cleanest sit-and-down fix: don't place her, reset, and release the instant the right position is complete. Sit and down alternate like Simon Says.
On film
as soon as his butt hits the ground before he has a chance to lay down as soon as his butt hits the ground release right
Bo
Bo's file holds most of the chapter: an audit of which behaviours still need help, a stated criterion ('he stopped when I stopped'), 'you can't accept less' on a recall that stopped short, and a mock boarding kennel with handlers who barely know him, because boarding is what the family needs.
On film
Bo. Down. Nope. Down. All the way. Down. Nope. Go down.
Bullet
It also has a place test run out of sight, and the most useful line on proofing mistakes: an error is a lesson.
On film
release so that's a handler error i can remove that by just not saying that before
Thor
One family's dog lies down when asked to sit, and over three visits the family hears three answers: fine for a pet, corrected gently, and never. Teach the distinction and leave the standard to the owner.
On film
work the duration here sit i'm okay he's a pet dog it's functional obedience here if he folds into a down it's fine
I, the down always has to be a down.
Whiskey
A handler working Whiskey's stay off the cot admits she feels bad she can't give the dog a single success. A colleague hands her the rule on camera, with its one condition: the success has to be real.
On film
I know. I just feel bad because I'm not able to give her any success.
Zeus
Only the whole behaviour counts: 'all the way' until he goes. Progress is narrated as help falling rep by rep, until the down comes on the verbal alone. After his best session the trainer thinks Zeus would listen off leash and leaves the leash on anyway.
On film
and based upon his reaction there I think that he would have listened to me pretty well even if he was off leash but I just you know he's a very big dog and there's liability issues I'm not going to test that
Bun
The dog the term false reps comes from. If every 'come' ends in a down, the two pieces weld together, so the trainer throws in recalls that break the prediction. His first week had one rule, no pulling, before any holding commands.
On film
is that if you do the same pattern over and over and over and over again like that, they will learn it like it is a command, right?
Dutchess
The criterion applied only to commanded reps: a sit she offers at the end of a recall is allowed, while a sit he had asked for would have been held. For a sit that blurs into a down, the family gets a structural fix instead of a correction: only 'down' in the house.
On film
i'm taking her out duchess come sit because it's the recall i'll allow that if i had specifically
Stormy
Halfway through a lesson in the owners' street, a dog who has braced and yelped all along starts going with the leash. The trainer points out that everything else just became easier to teach.
On film
with it right and so you can see now that it's becoming easier to teach with the pressure now
Anna
A very nervous dog, and one fear at a time. The short leash is added only once Anna can take a little pressure.
On film
So we can't introduce that all at once, but she's getting some confidence now to where
Where the trainers differ
On film
Yeah, lower the criteria. Get her a couple wins, and then slowly raise the criteria.
On film
want you to notice there is is that uh if you want it to like look correct and you want him to do it right and come all the way to you, you can't accept less.
On film
and it's important that you don't give up at that point or think that you're hurting him
On film
I'm not going to go for any more A-T-E-L because I'm going to start recapping that and pushing that again in another play session, not here on the walk.
On film
Now, the reason that I'm telling him to go is because he's kind of like fell into a heel by anticipating, and I don't want him to anticipate it that way.
On film
hey, down, oh, okay, well, good job, I guess, okay, so that was, that was a predicted down, so she didn't really, like, think she already heard me say it
On film
What I find interesting is that she's going there and sitting in the ground before you go.
On film
that's not good it's supposed to be verbal only like my hands in my pockets
On film
usually those are like visual aids that that we don't use um most of the time when you start training
On film
command i'm not gonna like push it over and over we're kind of gonna like walk around for a minute
On film
And this is gonna be a standoff until you lay down.
As a course unit
The learner can find out which of a dog's commands are really verbal. That means fading help in order (body, reward, leash, handler), keeping the word first in every chain, breaking a pattern on purpose without breaking a command that is still being learned, holding a given command to its finished form, and setting, naming, raising and lowering a criterion honestly. They can also lay out the order of a program and say what 'good enough' and 'reliable' mean for this dog.
Lessons, in order
- Hidden help
- False reps: the dog reads the chain, not the word
- One layer at a time: the order is part of the method
- Say the criterion out loud before the rep
- Raise the criterion in steps
- Don't accept the almost
- Swapped commands: keep sit and down distinct
- The standard is set by the job
- Reliability is what you have not tested
Drills
- The audit, as run across Bo, Dolce and Roxy: list the dog's commands. For each one, hands in pockets, say it once and wait a second. Score what it needed (nothing, a look, a point, a tug, more). Next session, word first and the smallest help after it. Then try the word alone again.
- The absent handler, as with Roxy and Bo: for duration commands, leave the room, close the door, and watch through a gap. If he breaks, come back and reset.
- Simon Says, as run with Sage: sit and down in an order decided before you start, reset to a stand on a wrong answer, and release the instant the right one is complete.
- All the way, as run with Bo, Zeus and Dolce: one command. On a partial, mark 'nope' and repeat, or add 'all the way'. Praise and release only when it is complete.
- Working sets, as run with Josephine: find the stay at which she breaks, drop a little below it so the next rep succeeds, and spread five to ten such stays across a walk. Test the maximum only now and then.
- The real test, as run for Bo and Pepper: build the situation the dog will actually face, with an unfamiliar crate or handlers who aren't trainers, and watch.
Assessment
- Which of your dog's commands work with your hands in your pockets and you out of the room?
- Your dog sits every time in the kitchen and never at the park. Before you blame the park, name the three things you do in the kitchen that you don't do at the park.
- When you last asked for a down and got three-quarters of one, what did you do next?
- He downs on your sit. What are your next three actions, and whose decision is it whether that down is acceptable at home?
- Your dog broke a 60-second stay three times in a row. What do you ask for next?
- He has come back every time in your fenced garden. What would a real test of his recall look like?
The ideas it draws on
- Hidden help
- False reps: the dog reads the chain, not the word
- Don't accept the almost
- Swapped commands: keep sit and down distinct
- Say the criterion out loud before the rep
- Raise the criterion in steps
- One layer at a time: the order is part of the method
- The standard is set by the job
- Reliability is what you have not tested