Why does my dog only listen when I'm pulling the leash?

Bo is where the mechanic is stated most exactly, because his sessions are early and the trainer is narrating what he is installing rather than using something already installed. The e-collar and leash pressure are introduced together on purpose, so that the dog learns one pressure system rather than two, and the collar's first use is not a correction. In the park session the word 'come' is attached at the instant Bo is already coming in off the line - and the trainer points out, in the same session, the first time the verbal cue lands on its own. The down is where the same dog shows the method's limits: physical manipulation stalls, so the trainer names and uses low continuous pressure - 'the nagging method' - and then says on camera that the rep was ugly, that he bent his own principle to make the dog do it once, and that he would like to hear a better way to train a down without food.

Lola is the nervous case, and in her file pressure and relief is taught twice over: once to her and once to her family. On day one the content is not obedience at all but what the trainer calls a baseline concept - that a leash pulling in a direction means move, rather than brace - and he defends the time it takes by pointing out that an e-collar tap carries pressure with no direction, so a directional channel has to exist first. With her bolting, the same loop is aimed at panic: running produces more pressure, the return ends it, and the return is then flooded with relief for a long stretch. Her lesson is also the only place an owner is warned that the method will look like it is hurting the dog before it works - the command gets worse, the dog looks flat, and the trainer puts a five-or-six-week timescale on the other side of that.

Zeus is the dog the loop is demonstrated on when it is already working. His sessions supply the corpus's cleanest statement that the two halves are one thing - negative reinforcement to get the position, a release of freedom on the back end every time - and its cleanest demonstration that the dog is answering felt pressure rather than leash length. In the possession games the currency is the toy and the trainer narrates the contingency to the dog while the dog is still refusing, then hands him the switch: do what is asked and the game comes back on. And where a command is not yet reliable, the decision is made out loud and without softening: he does not like going into the down, so it is going to have to be made worse for him.

Roxy's file shows the first two stages laid out as separate jobs on the same afternoon. Stage one produces the movement with a light tug and no word at all - the trainer says so explicitly - then a break, then the word. Within the same session the aid has faded to almost nothing, and he is honest that it took two forced reps to get there. Her handover contains the corpus's most unusual application of the loop, to something that cannot be commanded: at a fixed time she goes to the same spot in the yard on a ten-foot leash, where the handler stands and does nothing and she is allowed nowhere else, and going is what buys a minute or two of running. The pressure is boredom, the relief is freedom, nothing is asked and nothing is corrected - and the honest rider is attached, that the first time you may be standing there forty-five minutes.

Bullet, who arrived knowing nothing at all, is the dog on whom the pressure is stripped of equipment entirely. On the place cot the trainer stands still and holds the dog's gaze - social pressure, named as such - and the criterion he is waiting for is acceptance, not stillness: the dog getting off to come to the handler is scored as the same escape as the dog getting off to leave. The relief is the release straight into play with the helper dog, and the contingency is said out loud while Bullet is still holding. His owners get the ladder in the form they can execute: marker sound and the command again, then putting him back, then the collar tap, and the reason the first rung has to be free - it is counting to three with a child.

Deuce is ninety-six pounds of adolescent with no idea how strong he is, and his file is where the mechanism gets named without euphemism - straight negative reinforcement, said in those words, while the heel is being taught. The leash-pressure session is the loop reduced to its bones: constant random direction changes, the collar firing at the instant he hits the end of the line, markers deliberately left out so nothing stands between the end of the leash and the stimulation. The trainer hears his own language the way an owner would and answers it rather than avoiding it, and reports the dog's emotional state in the middle of it. At the handover the loop is handed to the family in one instruction: say the recall and pull at the exact same time, so the word and the obligation arrive together.

Dolce is the protection dog whose obedience ran months ahead of her confidence, and her sessions are where the loop is most explicitly defined by where the relief lives rather than by how much pressure is applied. The heel is not a place she is held; it is the place the pressure ends - 'if you're next to me you're good' - and when she gets stuck the answer is a smaller ask rather than a bigger push: lower the criterion to simply being beside the handler. Her handler names his own error in the same breath, that he gives up a little too fast. On the sleeve the same shape is paid in the only currency that works there: she gets to keep it, and only for the whole sleeve held.

Dutchess is curious rather than resistant, and her sessions are the best record of the timing question - not how much, but when. The release lands the instant she matches pace, and the reason is given immediately afterwards: she associates fast, so whatever the relief lands on is what she learns. Her file also contains the corrective to the commonest e-collar error, that a level which produced nothing may mean the dog was too distracted to feel it rather than that the level was too low; a different session has the same handler raising the level when it is genuinely not registering. And at her handover the same tap is shown carrying two opposite meanings depending on the word attached to it - 'nope' plus a tap marks something he does not want, the same tap while asking for place marks something he is trying to produce, and is followed by okay and praise.

Finn is the case that shows what the loop needs and what defeats it. His whole problem is context-bound: he is a different dog at the park and freezes on basics he knows at home and in the facility. So the pressure is used for a narrow purpose the trainer names precisely - to close off an option rather than to produce a behaviour - and the relief is the pool or freedom. His file also contains the corpus's clearest statement that physically placing a dog is not the same as teaching him: the hand on the hindquarters is not there to teach the sit, it is there to show that the position is unavoidable, so complying becomes the cheap route. And the constraint on all of it is stated: with this dog especially, poisoning a command is a live risk, because he builds aversive associations very quickly.

King arrived as a standard dog who simply did not know his commands, and his file is where the loop is taught to an owner as theory before it is shown on a dog: the four quadrants walked through with human examples - a present, an unannounced dinner, the seatbelt chime, a tax bill - and then translated into a leash-and-collar down followed by the release and the ball. It is in that lesson that the failure case is spelled out most usefully for an owner: keep pulling after the dog is already down and nothing was reinforced. His phobia work in a store shows the release word being spent rather than built - the trainer says he has a lot banked in it and can use it as a positive feeling next to the thing the dog is afraid of.

The shared Rocko/Kodak owner lesson explains the trainer's view that consequences must depend on something the learner can understand and control. In a separate Kodak walk, he attributes a disrupted heel repetition to barking dogs and describes pressure from both the environment and his handling. He interprets Kodak's return to play as the lesson he wants the dog to learn: that a stressful moment can be followed by recovery and a valued activity. This is the trainer's explanation of this sequence, not proof of a lasting emotional change from one repetition. The final place-work passage is a comparison from Rocko's session, where the handler describes waiting for settling before release; it is not another event in Kodak's training history.

Louie is liked by every handler in the corpus and was defensive enough that one of them wore bike gloves. His file is where the dose is visibly dialled down for the individual: the house method for the down is a pressured sit followed by a lure, and the trainer says in the same breath that with this dog he does not want to put a ton of pressure on, so he leans on the lure instead. The pressure that is used is held rather than escalated: Louie's first reaction to feeling it was to freeze and offer a fight, and the answer is to neither withdraw the request nor raise it, but to hold it until yielding ends it. The trainer's account of why he does not avoid the confrontation is unusually direct - with a dog like this, progress arrives in a handful of conflict moments rather than spread across reps. He also states plainly that the correction word's weight is manufactured, and that he is not a fan of it.

Luna was typed on day one as environmentally rather than genetically nervous, and her file is where the choice of pressure is matched to what the dog wants. She loves to explore, so when she fights the sit the answer is not a higher collar level but locking the leash so exploring is unavailable - and the trainer says out loud that what he is taking away is freedom. Her hardest session is also the one he calls good: generalising a known down to a new place, she tests him harder than she ever has, and he does not reduce the ask, because a test is the only thing there is to reinforce. The pressure is held explicitly - he is not going to let up on it - and the pay is free sniffing. He also declines to be certain why the fight was worse, offering generalisation as his best guess and refusing to commit to it.

Sage is the dog whose file states the quality bar. Her place work draws the line between a dog who has learned to stay up there because she understands and a dog who is sitting in what the trainer calls a suppression zone, and the release is withheld until she settles rather than delivered on a complaint. Her group session shows the aid engineered to be skipped: the leash tug comes before the hand on purpose, so she learns that the tug predicts physical manipulation and pre-empts it. And her go-home lesson is where the loop is handed over in a form a family can run - ask, tap if the second ask fails, then 'O and then a K', then loving - with the reason it matters for this dog in particular: she comes back to ask whether she did well, and somebody has to answer.

Stormie is soft, sweet and so physical that contact gets in the way, and her file contains the corpus's best example of solving a stuck rep by geometry instead of force. Three days of failing to get a down ended in about two minutes when a second voice worked out that from a sit she has something to brace against, and said seven words: don't let the sit happen. The pressure was not raised; the escape was removed. Her loose-leash session states the contract the dog is being taught in the plainest terms anywhere - listen and it is brief and you are fine - and draws the line between correcting slowness and correcting a bite, which are not given the same level. Her handler is also candid about having got nowhere, and about how little time he has actually had with her.

My dog only listens when the leash is tight. How do I get him to listen to the word?

Put the word where the relief is. In Bo's park session the trainer lets him reach the end of the long line, reels him in, and says the word at the instant the dog is already coming in off the pressure - so the word is learned on top of a behaviour that is happening rather than being spent on one the dog cannot produce yet. In the same session the trainer points out the first time the verbal cue lands on its own. Roxy's first session takes it a step further and separates the two jobs completely: the movement is produced with a light tug and no word at all, then a break, then the word is attached. If the word has only ever arrived while you were pulling, the pulling is the cue you have taught.

Isn't this just negative reinforcement? Doesn't that mean the dog is only working to avoid something?

Negative reinforcement is what the trainers call it themselves, without euphemism - one of them says 'straight negative reinforcement here' while teaching a heel. But their account of a repetition has two halves, not one. The pressure comes off, and then something the dog wants arrives: sniffing, a ball, a tug he gets to win, the run of the yard. A handler puts it as a rule - there is always positive reinforcement on the back end. In Zeus's possession games the dog is explicitly taught that he owns the switch: doing the thing turns the game back on. What the corpus also contains is the counter-evidence people ask for, reported in the middle of correction-based sessions: the dog's emotional state named as good, a dog getting the zoomies after a command, a trainer saying he measures attitude under pressure rather than compliance.

How much pressure is the right amount?

The corpus never gives a number, and it is honest about that. What it gives instead is a set of tests. The dose is read off the dog in the moment: a handler describes what he wants as low and continuous, another says the level is scaled to the behaviour rather than fixed per dog, and a third dials it down explicitly for a dog he does not want to load up. Where a rep stalls, the first move on film is often not more pressure at all but removing the dog's escape - the seven words that solved three days of failed downs were 'don't let the sit happen'. And the upper bound is stated as a principle rather than a setting: the pressure has to be contingent on something the dog controls, because a dog who cannot tell why it happened or how to avoid it is being traumatised rather than taught.

My dog got worse after we started correcting him. Did we break something?

On film that is described as a normal stage rather than damage. A trainer warns Lola's owners that adding negative reinforcement to a command the dog already knows will often make it look worse - the dog can look depressed, can panic slightly, can try a different option - because there is now more pressure inside the puzzle he has to solve. His claim is that the harder repetition is worth more than the easier one that looked better, and he puts a timescale on it: stack a few over five or six weeks and the dog figures things out faster and holds them longer. Kodak's session shows the same thing from the inside, where a repetition falls apart because dogs barked nearby and the pressure is arriving from two directions at once. What the corpus does not contain is any way to tell in advance which dog will come out the far side, which is why these sessions are narrated as judgement calls rather than as a protocol.

Will I be pulling on the leash forever?

The record says no, and shows the aid shrinking inside single sessions. Two forced repetitions into Roxy's first sit, the trainer is already down to a tug he describes as barely anything. Sage's group session builds the fade in deliberately: the leash tug comes before the hand every time so the dog learns the tug predicts the hand and skips ahead of it. And Zeus demonstrates why leash length stops mattering - he is answering the felt pressure, not the amount of line, so shortening it anywhere gets the same response. In Dutchess's sessions the handler says out loud each time no pressure was needed at all.

My dog freezes instead of giving to the pressure. Do I push harder or stop?

Neither, on the evidence in Louie's file. His first reaction to feeling pressure was to freeze and offer a fight, and the instruction is to hold the request without withdrawing it and without escalating it, until he finds that yielding is what ends it, and to mark the moment he does. The trainer's reason for not going around the confrontation is given plainly: with a dog like this, the progress is concentrated in a handful of conflict moments rather than spread evenly across repetitions. Finn's file shows the other half of the same judgement - with a dog who builds aversive associations very quickly, the risk of poisoning the command is named as the thing that sets the pace.

Keeping the pressure on after the dog has already complied, so the rep teaches nothing.

Running the pressure with nothing on the far side of it - no release word, no freedom, no game - so compliance buys the dog nothing.

Leaving the rules implied and then treating the dog as if he knows better - the dog cannot follow criteria that were never defined.

Reading a dog who comes to you off the place cot as obedient. The dog leaving to come to you is the same escape as the dog leaving to go away.

Concluding the training is hurting the dog at the exact moment a harder rep is about to pay off - the command looks worse and the dog looks flat before it gets better.

Reading panting on a cot as thirst or heat and letting the dog off for it.

Settling for the absence of pulling. A dog living at the end of the line all walk is still chaotic.

Following the dog around while trying to teach a recall, so there is another cue in the picture and the pull never means anything.

He walks fine while the leash is tight against him and drifts the moment it goes slack - he is answering the tension, not you.

He braces against the leash instead of moving with it - leans into it, plants, pulls back with his neck.

He knows the word - he has done it a hundred times - and in one particular room or one particular yard he freezes and gives you nothing.

He got worse, not better, after you started being firmer about a command he already knew - flatter, more hesitant, trying other things.

He does the position but you can see he is holding still to stay out of trouble rather than because he understands what you asked.

The command works for the trainer, or for one person in the house, and falls apart for you.