Is my dog confused or is he refusing?

Bo's handover is where the fork is set out formally, and it is given twice - once with a worked example halfway through and once as the closing answer to the owner's only stated worry, which tells you what the trainer most wanted her to leave with. The framing is clarity first: most of dog training is about clarity, and clarity to the dog rather than to you, because "they're not telepathic." Then the read. If he is trying and short - the example given is a dog who sits when asked to down and then looks at you - he gets helped in, with the words to say while you do it, and he does not get punished. If he knew and is off doing something else, he gets the marker word, which is described as the word that predicts that something bad is about to happen. What makes Bo's file more than a rule is that the observational half of it is on camera weeks earlier, in a public-access clip: the trainer cues the down, gets nothing, and narrates the read live - you can see the look on his face where he knows what I want.

Deuce's file gives the same fork in the most actionable form anywhere in the corpus, because it tells the owner which hand to reach for. The dog has started treating sit and down as interchangeable - Connor's reading is that he thinks maybe he can just do either - and the instruction is explicit: no collar for this one, just pull him up into the position, because hands are how he learned it. Elsewhere in the same program the opposite case is read off the dog's face just as confidently: that look, Connor says, is not the look of I don't know what that means, it is the blowing-you-off look, and that one gets the e-collar. The case also carries the warning that makes the whole read matter. On a confident-looking breed, a dog who locks up reads as stubborn and is often confused, and pushing on the wrong reading is how you "hand poison a command".

Stormie's file contains the experiment, which is the part of this concept an owner can actually run. A command that was working has stopped working, and rather than assume avoidance the handler drops the word entirely and starts producing the position with the leash. His reasoning is that if the failure was about clarity, that fixes it right then and there - and he names the cause of the clarity failure as his own schedule rather than the dog: teaching a couple of other commands in between is what broke this one. Later in the same session the test comes back with the other answer, and he names the physical tell that decides it. She is yanking against him with her neck, which a confused dog does not do, so this time it is avoidance and she just does not want to.

King's file is the staff version, and it generalises the read into a two-question diagnostic applied to every failure: what is the dog's emotional state, and how clear was it to him. The symptoms are named - flat, or desperate and throwing out behaviours, or simply doing the wrong things - and the teaching is done by analogy, on the junior he is talking to: if I stood over you shouting at you to just read the board, would that help, or would it induce panic. From that follows the instruction that most distinguishes this case from the handovers: never default to more pressure, default to suspecting you should play with the dog. And if the dog has gone backwards, go backwards with him and re-clarify the criteria from there. The same session answers the obvious objection that using help is cheating, with an analogy about training wheels: nobody teaches a child to ride with nothing, and the wheels come off soon enough.

Lola is a nervous dog, and her file turns the read into a live dial rather than a one-time verdict. Working a recall, the trainer says he is watching her emotional state and changing what he does based on it: a dog who has simply blown him off gets told; a dog who starts back and then hesitates gets cheerleading, because you do not punish uncertainty. His reason is the best analogy in the corpus for what confusion feels like from inside - a parent telling a child to clean that, and the child asking what, and the parent repeating it louder. He also names the moment to switch, which is what makes it a dial: you can see the shift between a dog saying I am not doing this and a dog saying I am not sure, and you flip to the encouragement side the instant it happens. Her other session adds a third outcome that a two-way model has no room for. A behaviour he reads as fear-based gets nothing at all - not corrected and not rewarded - and the next correct behaviour gets paid even while he is still unsure what caused it.

Dutchess supplies the possibility that neither branch of the fork covers, and the trainer thinks it important enough to stop the session and define a word for it. She was trying to hold the stay - he says he could see her trying - and her energy and her degree of self-control would not permit it. He then spells out what permit means, because the word is doing the work: not that she declined, but that it was not within her ability at that time. What follows is the part that matters, and it is a choice most handlers do not make: he shortened the exercise to what she could do and raised it as the ability arrived, rather than raising the pressure. A separate park session adds the simpler reading on the same dog - sometimes on the down she just gets a little confused - and between them her file is the reason nothing in her case reads as a battle of wills.

Roxy's file is the vocabulary changing on camera inside a single day, and it is worth preserving rather than tidying because the words decide what happens next. In the morning session the trainer reaches for stubborn, says he does not want to use that word, and uses it anyway. By the afternoon he has rejected it outright: not stubborn - avoiding, which makes more sense, because she knew what she was doing and wanted to go away. The distinction is not cosmetic. A dog who did not understand needs clearer teaching; a dog who is avoiding needs the avoidance to stop working. The same morning session also gives the corpus's most honest description of physical help still attached to a command - training wheels on her, pulling her down while he says the word - which is what the concept looks like when it is only half finished.

Bullet adds the least flattering possibility on the list, offered without being asked for and about an exercise that is never reported as solved. Working the heel turns, the handler runs through the candidates out loud: maybe it is his own fault and he was not clear enough when he first taught it, maybe the dog is having a hard time learning it, or maybe he just finds it boring - which, he says, would not be shocking for a high-octane dog with a lot of other things he would rather be doing. Nothing in the existing concept list permits boredom as a diagnosis. It belongs here because it is the fourth branch, and because a handler who will say it about his own session is a handler who is genuinely reading rather than defending.

What to reach for when a rep fails. Never more pressure, or the marker and the collar.

The two are probably addressed to different situations - the play-first rule is staff-facing and is about a rep that has stopped working, while the marker-and-consequence rule is owner-facing and is about a known command being declined - but no transcript says so, and the King session states its rule as a default rather than as a special case. An owner who has watched both will not know which one they have been given. This is the sharpest unresolved thing on this page.

What to do with a failure you cannot classify.

Not a contradiction so much as two different tools for the same gap, and it is worth noticing that neither handler pretends to have resolved the ambiguity. The Lola session is the only place in this corpus where a handler acts under acknowledged uncertainty and says so on camera. Whether the withhold-everything option is house practice or one trainer's habit with nervous dogs is not established.

A dog who breaks a known position: could not, or would not.

Different dogs and different handlers, so this is not a disagreement about a case. It is listed because the two readings produce opposite actions from the same visible event - a dog who takes a position and leaves it - and because only one side supplies a tell. Stormie's file names the physical sign for avoidance, the neck yanking against the collar; nothing in Dutchess's file names a sign for inability beyond the handler's impression that she was trying. That gap is where owners will guess.

How do I tell whether my dog is confused or ignoring me?

Read the attempt rather than the outcome. A dog who is confused is usually doing something - offering the wrong position, getting halfway, then looking at you for the rest of the answer. That is the picture the trainers describe as trying and not quite getting there, and it is the one that gets helped rather than corrected. A dog who is declining is typically doing something else entirely: sniffing, looking away, engaged with the world. The handlers here also claim to recognise a particular expression - one of them describes it as not the look of I don't know what that means, but the blowing-you-off look - which is real but is exactly the judgement an owner is least equipped to make. If you are not sure, treat it as confusion, because the cost is asymmetric: helping a dog who was actually refusing loses you a repetition, while pressing a dog who was confused can cost you the word.

Is there a way to test it rather than guess?

Yes, and it is the most useful thing on this page. Stop saying the word altogether and simply produce the position with the leash - guide him into the heel, or the down, without naming it. If the failure was about clarity, that fixes it right then and there, and you have learned that the word had come loose rather than that the dog had. If he braces and fights the guidance instead, you have your other answer, and one handler names the tell he uses: a dog yanking against the collar with his neck is a dog who knows what is going on and does not want to do it. Note where the trainers point the blame when the test comes back as confusion. The named cause is the teaching schedule - other commands taught in between broke this one - not the dog.

My dog is just stubborn. Isn't he?

One of the trainers here says that word, dislikes himself for it, and comes back a few hours later having rejected it: not stubborn - avoiding. The objection is not that stubborn is unkind. It is that it answers neither half of the question you actually need answered. Did he understand, and did he choose. Avoiding says both - she knew what was being asked and wanted to go the other way - and it tells you what to do next, which is to make the avoidance stop working rather than to teach the command again. If you catch yourself reaching for stubborn, try finishing this instead: he knew, and he wanted to ___. If you cannot finish it, you do not yet know that he knew.

He holds the stay and then breaks it every time. Is he testing me?

Possibly not, and there is a third answer worth knowing about. One trainer stops a session specifically to define a word for it: the dog was trying to stay - he says he could see her trying - and her energy and her degree of self-control would not permit it. Permit meaning, as he spells out, that it was simply not within her ability at that moment. What he did next is the part to copy. He shortened the exercise to something she could actually hold, and lengthened it as the ability arrived. He did not add pressure to a requirement she could not meet. If your dog breaks at about the same point every time, and breaks worse on the days he is wound up, this is the reading to try first.

Isn't helping him just teaching him he doesn't have to do it himself?

The trainers get this objection and answer it with an analogy about bicycles. Nobody teaches a child to ride with no training wheels on the grounds that wheels are cheating; you put them on to get started and you take them off soon after. The same trainer who says that is candid elsewhere about a dog still on them - training wheels on her, pulling her down while he says the word - which is what a half-finished command honestly looks like. The thing that makes help temporary is not refusing to give it. It is noticing when it is no longer needed and stopping.

What if he hesitates - starts to come and then stops?

That one gets encouragement, not pressure, and the trainers treat the switch between the two as something you make mid-repetition rather than deciding in advance. On a nervous dog, the handler describes watching her emotional state and changing what he does based on it: a dog who has simply blown him off gets told about it, and a dog who starts back and then falters gets cheerleading - come on, come here - at which point she speeds up and arrives. His reason is that you do not punish uncertainty, because a dog who is unsure has nothing to comply with. And he names the moment to flip: you can see the shift between a dog saying I am not doing this and a dog saying I am not sure, and the instant it happens you move to the encouraging side.

What happens if I get this wrong?

It depends which way you get it wrong, and that asymmetry is the practical heart of the concept. Help a dog who was actually declining and you have lost a repetition, which you will get back. Press a dog who was confused and you do damage that has a name in this corpus - you poison the command, meaning the word itself starts to predict something unpleasant and the dog gets worse at it rather than better. The risk is highest on dogs who look confident while they are confused, which is exactly when a handler feels most justified in pushing. The senior trainer's version of the argument is aimed at his junior rather than at a dog: if I stood over you shouting at you to just read the board, would that help you learn, or would it induce panic.

Correcting a dog who was trying and fell short. He offered the wrong position and then looked at you, which is the picture of a dog attempting the task.

Reading a confident dog's lock-up as toughness. On some dogs that face is confusion, and pressing it poisons the command by hand.

Reaching for the e-collar when the dog is confused between two positions. He was taught by hand, so the hands are what he will understand.

Assuming avoidance when a working command stops working. The trainers' first suspect is their own teaching schedule - other commands taught in between - and the test costs nothing to run.

Punishing hesitation. A dog who is unsure has nothing to comply with, and the moment she begins to give you flip to encouragement rather than continuing.

Holding a criterion the dog's state cannot carry. Where the energy or the self-control is not there, the exercise gets shorter rather than firmer.

Calling the dog stubborn. The word answers neither half of the question, and one trainer rejects it on camera and replaces it with one that does.

He offers the wrong position - sits when you said down - and then looks at you, waiting.

He looks straight at you and does nothing, and you would swear from his face that he knows exactly what you asked.

He plants and braces when you try to guide him into position, pulling against the collar with his neck.

A command he had is suddenly gone, and the timing lines up with you having taught him something new.

He goes flat and switches off, or the opposite - he throws every behaviour he knows at you in a panic.

He starts towards you, then falters halfway and stalls - which looks like changing his mind and is more often nerve.

He holds the stay visibly trying, then breaks - and he breaks at roughly the same point every time, on the days he is wound up.

He is good at everything except one exercise, and with that one he switches off the moment it starts.

He did it perfectly two days ago and is acting now as though he has never heard the word.