Do I need to teach my dog to stay?

Bo's go-home lesson is where the rule gets its fullest statement to a client, and the argument it makes is about the owner rather than the dog. The trainer runs the whole command list, gets to place, and then stops to say that stay is not in it and never will be - and the reason he gives is that the word creates a problem rather than solving one. He is explicit about the arithmetic from the dog's side: a dog who hears stay a hundred times cannot know how many repetitions are binding. Then he gives the break protocol in the same breath - if the dog gets up before the release word, you tell him nope and put him back - so that the release keeps its monopoly on ending the command. The same trainer explains the same rule to a stranger on a park trail months earlier, unprompted, which is the best evidence in this file that it is not a script for clients but how he actually thinks about it. Bo's own record is also where the corpus has filmed this most and read it least: his unread half contains a ten-part series called Implied stays.

Roxy's first in-program lesson gives the rule its theory and its best joke. The command is described as a window: the word opens it, the release marks its end, and nothing in between is needed. It is the only place in this cluster where the rule is extended out loud to a moving command - even a heel is held until okay, and there is no heel heel heel. It also contains the one stated mechanism anybody offers, which is a claim about what repetition teaches: say it over and over and the dog learns the position is temporary and can be left in a couple of seconds, which the trainer points out is fatal for place, where duration is the entire point. The illustration is somebody else's dinner party, with the dog fighting demons while his owner says stay every ten seconds. Roxy's second lesson adds the other half of the same idea - most of the repeating owners do is caused by not leaving the dog time to process the first command - and her later training session shows the rule bending in practice, because there the handler repeats come until the dog arrives.

Luna is a nervous case, and her file is where the rule is stated as an absolute and then immediately complicated. The absolute is the cleanest sentence in the cluster: to get a dog to stay, you never say stay. The complication is that the release word is defined by use and not by intent. In one session the trainer catches Luna releasing herself on good girl and says so on camera; in another, the trainer stops the owner mid-rep to tell him that the word he just said - release - is not her release word. Her file also carries the tolerances nobody else gives: dropping straight from a sit into a down is not allowed, dropping after three to five seconds is, because she is holding the position and getting tired, and coming up into a sit out of a down is never allowed at all. And it carries the gentlest break protocol in the corpus. Asked by the owner how he would punish her for breaking, the trainer says he would not, because sometimes it is too much and she will not know what to do with the information; he would move her out of it and put her back. The proofing exercise that goes with all this is the most copyable thing an owner is asked to do anywhere: the handler runs across the room while the dog holds, because a dog who reads excitement as a release has learned the wrong cue.

Zeus gets the rule twice, from two handlers, in two stints, and it comes out almost word for word both times - which makes it one of the few house rules the corpus confirms independently rather than by repetition of a single voice. The early solo session states the systems argument bare: the release word is the only thing that hands freedom back, so to keep him in anything you simply do not say it. The owner lesson two months later says the same thing with the handler demonstrating on himself - if I sit down I don't have to say stay - and sets it inside the wider frame the lesson is built on, where every command is either telling him to do a thing or telling him to do nothing. Because okay is the only exit, the trainer coaches the owner to make it clearer: a release the dog is not sure he heard leaves him still in the command.

A different dog with the same name - a Cane Corso from another household, owner-confirmed as a separate case after the intake comparison - and a different trainer, who names Kelton in the third person. He opens his owner lesson by saying he will show her how the implied stay works because it applies to all of their commands. That is the value of this file: the rule reaching a second family, through a handler who is neither of the two who state it most often, in the same words. He also frames the whole session the way the rule is framed - it is for the owner rather than the dog, so that she knows what she is getting into rather than having the dog dropped off with have fun.

Dolce is where the rule is built rather than explained, and where the vocabulary shifts under it. On her first public-access outing the implied stay is the named exercise and it is tiny: sit, a couple of steps back, a few seconds, break - and her release word in that session is break, not okay, which no other session of hers uses. Pressure, not a correction, is what puts her back, and the stated aim is that she succeed. Seven weeks later at the owner lesson the same idea is stated as a finished thing - once she goes to position it is an automatic stay until she hears okay - and it is paired with the correction marker so the family gets both ends of the window at once: eh-eh means settle back into the position, okay means the obligation is over and she may do whatever she likes. Hers is also the clearest statement in the cluster of what the release actually buys: total freedom, including going to meet people, so long as she comes back when asked.

Louie's park session is the rule used rather than taught - the shortest and most ordinary appearance it makes anywhere, and useful for exactly that reason. The handler notices the dog thinking about getting up to investigate something, tells him to sit, and names what is now happening: an implied stay. He is explicit that the position is held without a treat being shown, and the block ends with the release framed as the reward - a nice release to being able to explore again. It is the rule as working vocabulary in a nervous dog's proofing session, not as a piece of doctrine being handed to a family.

Dutchess is the dog for whom the implied stay is the last item outstanding, and her session is run deliberately as a dress rehearsal for go-home day. The trainer names it as the only thing left and says how he intends to get it - a sentence that is also the clearest terminology slip in this cluster, because what follows is not negative reinforcement in any technical sense but a correction: she gets up, she is put back, and if she does it again he hits it immediately. The transcript renders implied stay as implied state, which is worth knowing before anybody searches for the phrase. What makes her file valuable despite the slip is the posture around the correction: her failures are read throughout her record as confusion or impulse rather than refusal, and the rule is stated to her as a rule rather than as a fight - that's the rules with her, it's just the rule.

What you do when the dog breaks: correct it, or simply reset it.

The two sides are not addressed to each other and neither handler mentions the other's practice. The visible pattern - correction for the dog who is testing and reset for the dog who is nervous or still learning - is this reader's reading, not a rule anybody states. Dolce appears on both sides seven weeks apart, which is either the same distinction applied over time or two handlers doing it differently.

Whether the command is ever repeated.

Both are Roxy, weeks apart, and the second may be the difference between a held position and a moving recall in progress rather than a contradiction. Nobody on film draws that distinction, so it is left standing. See `should-i-repeat-the-command`, where the same tension is the whole subject.

So how do I tell my dog to stay?

You do not. You tell him to sit, or down, or place, and then you say nothing. In this system the position carries its own duration, so the command you already gave is the instruction to stay: he is expected to hold it until he hears the release word. The whole gain is that you now have exactly one thing to remember instead of a running commentary - one command, then silence, then okay. One trainer compresses the rule into a single line he clearly enjoys saying: to get a dog to stay, you never say stay.

What do I do if he gets up before I release him?

You put him back where he was, and the one thing you do not do is say the command again. On the go-home dog the handling is marked with nope and then the dog is returned to the position and released properly, so that the release is still the only thing that ends the exercise. How hard the marking is varies by dog and this is the one place the corpus genuinely disagrees with itself. With the nervous dog, the trainer is asked directly whether he would punish a break and says no, because sometimes it is too much and she will not know what to do with the information - he moves her out of it and puts her back instead. With the dog who is testing, the correction is immediate. If you are unsure which yours is, reset first: resetting costs you nothing but time.

Does this really apply to heel as well, or only to sit, down and place?

To all of them, and the heel is the case that makes the principle visible. A heel is a moving command, so there is no position to hold - and the trainers still treat it as held. The dog is in the heel from the word until okay, and there is no heel heel heel along the way. The same lesson points out why this matters most for place: if you have to keep repeating the word, you have taught the dog that the position lasts a couple of seconds, and place is the one command where duration is the entire point.

What if I accidentally say something else and he gets up - is that his fault?

No, and this is the part of the rule that catches owners who are otherwise doing everything right. The release word is whatever word you actually use, not the one you meant to use. One trainer watches a dog release herself on good girl and names it on the spot; another stops an owner mid-rep to point out that the word he had just used - release - is not the dog's release word, which is why she did not move. Pick one word, use only that word to end work, and be careful what you say while he is holding: praise, relief and a sudden burst of energy all end up as cues if you let them.

My dog already knows stay. Do I have to untrain it?

Nobody in this record answers that, and it is worth saying so plainly rather than guessing. Every statement of the rule in the corpus is made either about a dog the trainers started themselves or to a family receiving a dog who was trained this way from the beginning. What the record does show is the direction of travel - the rule is introduced as a removal, the deletion of a word and of the habit around it - so the practical version is probably to stop using the word and start using the release, and to expect the dog to need the release taught as the thing that now ends the work. That last sentence is inference. It is not in anybody's mouth on film.

Saying stay, and then saying it again, until neither of you knows which repetition was the binding one. This is the failure the whole rule exists to delete.

Ending the command with a word you did not mean as a release - praise, a sigh of relief, or the word 'release' itself - and then wondering why the dog gets up early.

Letting your own excitement be the release. Break into a run or get hyped up and a dog who has learned the pattern reads it as permission, so the cue becomes your body rather than your word.

Using the commands interchangeably, so that the dog is answering a different question from the one you think you asked - the complaint that most often arrives at a follow-up as 'he's not listening'.

Re-cueing the command when the dog breaks, instead of marking it and putting him back where he was. Re-cueing pays the break with a fresh command and teaches him the position restarts on its own.

You catch yourself saying stay repeatedly - at the door, at dinner, whenever guests are over - and he still gets up within a few seconds.

He holds the position perfectly while you stand still and breaks the moment you move away, or the moment you get excited.

He releases himself when you praise him, so a good boy ends the command as reliably as the release word does.

He is fine in a sit but slides into a down after a few seconds, and you do not know whether that counts as breaking.

He does a command you ask for and then wanders off, and you had assumed that was the end of it - there is no moment in your routine that means the job is over.