My dog ignored me and then came back - can I tell him off?

Roxy is where the rule is stated as a rule. It arrives in the middle of the go-home lesson, inside the two-factor recipe for maintaining her off-leash recall - reinforce the return every time, sometimes big and sometimes just a good job and keep walking; use the collar when she does not come - and then the handler stops the recipe to put one absolute on top of it. He says it twice, with the reason attached, and he includes himself in the failure mode by naming his own dog as one that does it to him. Her case makes it the highest-stakes rule in the handover, because the off-leash recall is the headline result the family is taking home, and the second after a failed recall finally succeeds is the moment most likely to destroy it.

Finn is the case that makes the rule urgent rather than tidy. He is a flight risk whose program was built around the recall, so the one lesson he must never learn is that coming back does not work - and that is exactly what Kelton says a punished return teaches, describing it in the dog's own voice: he ran, it failed, he came back, and that went terribly too. The instruction to the family is emotional rather than technical, and it concedes the cost to them: you are angry, say good job anyway. The same case supplies the mechanical half, in the second in-program lesson, where the owner is told not to stack repetitions with a stim on each one, and where the failure state is demonstrated sarcastically - yell sit four times and hit it each time if you want to shut him down. And it supplies the repair: after pushing him too far at Home Depot, the handler says the word several times with nothing following it, partly to take the pressure off the word and partly as a diagnostic for whether the dog is shut down or the word is spoiled.

Bo has the softer half of the concept, and it is the half nobody else in the corpus states. On one recall the handler deliberately does not ask for the down he had been working, and says why: he does not want Bo to associate coming back with being made to do something. The sit stays, but only because he wants it as the recall's finishing move, and even that is not asked for every time. Read with a session from four days earlier - where an informal recall is paid with affection and no pressure at all, and coming to the handler is framed as the start of the fun rather than the end of it - Bo's two files state the same rule from both ends: make the return pay, and do not let it cost.

Luna's first in-program lesson gives the mechanism a price. Teaching the four quadrants to her owner, the handler lands on the order of operations - the word, then the marker, then the collar, never the collar on the heels of the word - and spells out what the dog concludes if you get it wrong: every time he says sit I get shocked, which means sit is a punisher. Then he costs it. Unpoisoning a command takes about three weeks, and the program is four. That is the sharpest argument in the corpus for the rule, because it is stated as lost program time rather than as a principle, and it is said to a family who are about to be handed the remote.

Lola's go-home lesson supplies the same rule in its most compressed form, and the reason it is worth having twice is that it is told against the handler himself. Asked to show what they do, owners give the cue and the correction together - sit, boom - and he names what that teaches in one clause. He then gives the two observables: a dog who hears the cue and does not know what to do, with his legs shaking; and the first dog he ever trained, a dachshund who bolted at the word sit because he had taught it the wrong way round. He says he felt really bad. That is the only place in this material where the rule is backed by someone's own failure with a name and a species attached.

Deuce is the counterweight, and he is in this file for that reason. His recall is chained to a sit on every single repetition, deliberately and over and over, until the two fuse and he offers the sit without being told - which the handler calls a conditioning thing rather than a command, and which is checkable three weeks later in a field, where he drops into it by himself and the handler remarks on it. That is the opposite policy to Bo's deliberately empty arrivals. Nobody in the corpus notices the disagreement, and both handlers are describing a recall they are happy with.

Whether a return is allowed to carry a demand at all.

Two handlers, two dogs, and neither addresses the other. It is not obviously reconcilable by stage either: Bo's withheld demand is early-program teaching and Deuce's chaining is early-program teaching too. A reader could guess that the sit is cheap and a down is expensive, or that it depends on whether the dog finds the demand unwelcome, but nobody on film says so. Left open.

Whether the cue may be repeated, and where the collar goes when it is.

These are not flatly contradictory - one is about latency and the other about obligation, and Kodak's handler also supplies the escape hatch of the informal cue. But they pull opposite ways in the second a family is standing in a park with a dog who has not come, and no session in this material puts the two rules side by side. The gap an owner falls into is the length of the beat: nobody names it in seconds.

He blew me off for four minutes. Am I really supposed to praise him?

Yes, and the trainers do not pretend it is easy. The instruction given to one family is to channel your inner Buddha: you are angry, you say good job anyway, even though the dog has just irritated and, in the handler's word, tortured you. The reasoning is about sequence rather than fairness. The dog is learning from the moment he arrives, not from the four minutes before it, so anger on arrival attaches to the arrival. Another handler puts the same thing as punishing her for doing the right thing, because she finally came back. What you are allowed to be annoyed about is the refusal, and that is dealt with where it happened - by making the recall real at the moment it was ignored, not by settling up afterwards.

Won't he learn he can ignore me if nothing happens when he comes back?

Nothing happening when he comes back is not the same as nothing happening at all. The correction in this method is attached to the refusal, at the moment of the refusal - the word, then the marker, then the collar if the word is not answered - and by the time he is on his way to you that transaction is finished. What the handlers refuse is a second bill presented on arrival. The risk they are guarding against is specific and, on their account, expensive: a dog who has learned that neither leaving nor returning works has no good option left, and for a dog who bolts that is the one lesson that cannot be allowed.

I say come and press the button at the same time. Is that wrong?

On this method, yes, and it is the commonest way a good recall gets spoiled. The order is the word first, then the mark, then the pressure - the pressure attaching to the fact that he did not come, not to the sound of the word. Run the other way round, the dog draws the conclusion one handler says out loud on behalf of his dog: every time he says sit, I get shocked, so sit is the punishment. He then puts a price on the repair: about three weeks, inside a four-week program. Owners are told the marker first is not a preference but a must.

Should I make him sit every time he comes back?

The corpus disagrees with itself here and it is worth knowing that before you pick. One handler chains a sit onto every single repetition until the dog offers it uncued, and three weeks later the dog drops into it by himself in a field. Another deliberately skips the demand on some arrivals so that coming back does not come to mean being made to do something, and keeps the sit only because he wants it as the recall's finishing move - not on every return. Both describe recalls they are satisfied with. If you want the finishing sit, the safer reading of the two is to ask for it often and leave some arrivals free, which is what the second handler actually does.

I think I have already ruined the word. What now?

There is one repair in this material and it is deliberately unimpressive: say the word several times with nothing following it - no pressure, no consequence, no demand - and watch what the dog does with it. It is run on Finn at Home Depot after his handler says out loud that he pushed him too hard that day, and it does two jobs at once. It takes the prediction off the word, and it tells the handler which problem he has: a dog who is temporarily shut down, or a word that has gone bad. The rest of that session is the other part of the answer - stop asking, walk, and let the dog come back to baseline before anything else is tried.

Being angry when the dog finally arrives - which charges him for the one behaviour you cannot afford to lose.

Repeating the cue and pressing the collar on each repetition, which is named on film as the way to shut a nervous dog down.

Giving the cue and the correction together, so the dog learns the word is the punishment rather than the instruction.

Making every return the beginning of work, so that coming back becomes a toll booth.

Pushing a half-trained word until the dog goes quiet, and reading the quiet as compliance.

He comes back eventually, and you notice you are no longer pleased to see him by the time he arrives.

He used to come and now he hesitates, or he stops a few feet short of you.

You say a command and he looks like he does not know what to do - and his legs shake.

He moves away from you when you say the word, rather than towards you.

Every time you call him you then put him somewhere or take something off him, and he has started taking longer to arrive.