My dog won't take treats or toys when he's nervous

Finn is the case this concept was built on, and the one where the search failed for weeks. He freezes when personal play is offered - the handler reaches to grab at him and instead of play drive he locks up and becomes frightened - and on a store field trip a fortnight later the handler says outright that his usual fallback is unavailable, because he has not figured out what this dog likes and is going to have to keep experimenting. What is available is prey, and it is deliberately set aside: it is the highest-value thing and the fastest to switch him on, and it is not under the handler's circumstantial control, so it is less useful for training. What he lands on instead is a jog. The mechanism is stated as chemistry rather than psychology, and then the build is described in three moves in a single passage: run for the feeling, lay the touch and the personal play onto the run so handling predicts it, and put the run at the end of an obedience repetition so the work buys it. The payoff is named too, and it is not sentimental - having a reward Finn genuinely wants is what lets the handler raise the difficulty and use negative reinforcement without huge fallout. The session also records the change as it happens: Finn lagged behind the first time the handler broke into a run and by the end takes off immediately. Against all of that sits a live constraint - this dog builds aversive associations fast, so the handler says he is scared of poisoning any command with him.

Kodak carries both halves of this concept, and the two halves were filmed months apart. On his first public walk the run is already doing double duty - it is a stress reducer in its own right, argued from brain chemistry at a moderate trotting pace, and it is stacked immediately behind the release word, where the handler says it raises the value of that marker and can charge it up. That is a second job nobody else in the corpus assigns to exercise. The same walk contains the hardest instruction in the whole concept: after a good recall from a dog who does not trust him, the handler does not praise and does not touch, and he names the temptation rather than pretending it was easy. He also refuses the standard justification. He does not say the comfort rewards the fear; he says that explanation does not make sense behaviourally, and then gives the definitional one - an aversive is whatever the animal wants to get away from, and for this dog right now that is loud praise and a hand. He demonstrates it on camera, praising loudly and pointing out that it frightened the dog. Then in September, four minutes into a session, the whole thing turns: Kodak wants a ball. The session plan is dropped, the stated aim becomes getting him out of his nerves and into a good place, and the rule is handed forward as something anyone can copy - carry the toys and actually try them, and do not decide in advance that a nervous dog will not play. What that looks like from outside is a dog staring a ball down in the heat, indifferent to the stranger holding it.

Lola is where the rule about touch is stated most cleanly and also where it is hardest to hold, because with her the answer might be touch. On her intake day the handler sits near her and refuses to pet her, and explains why in the same breath: people think petting a nervous dog will make her feel better, and if the dog is terrified and does not like the pet, the pet stops being a reward. In a later session the same search is running and the candidate is contact - the main thing is finding something that opens her up, and if cuddling is the only thing that does, he will sit in a room for two hours and cuddle with her. The reason opening up is a stage of the programme rather than a nicety is given plainly: if she cannot open up around him, he cannot teach her to open up around anyone else. What does work as a payment in her sessions is the release and the run that follows it - he says the clarity came from saying okay and letting her go immediately.

Louie's first session is the honest negative result, and it is recorded as a plan rather than as a failure. Ball, loosening up, a rope toy, a long line so he can move, a test of whether the line touching him is what is blocking the running - none of it works, and the handler stops rather than pushing. What he writes down is that it is early days, they are still figuring him out, and he is going to have to take the dog on some ordinary walks before he can get him to engage at this level at all. That puts engagement upstream of reward selection: with this dog there is nothing to choose between yet. The read of the dog's temperament that forms in the same minutes - a sensitive dog, spooked by the line - is what the rest of his June set agrees with.

Bo is the inverse case and it belongs here because it shows what the concept is actually about - having a channel, not having a toy. Bo's problem behaviour is that he loves contact and jumps on people. The handler's move is to put the jump on cue, under any word you like - uppies, up, jump - and punish it every other time. What that buys is a reward he did not otherwise have: for a dog who wants contact, the cue that permits contact is itself the payment. The annoying habit becomes the payment method, and the handler is explicit that the name of the cue is irrelevant.

Bullet gives the requirement in its shortest form, and as a floor rather than a preference. Off leash in a park, at range, the handler offers the ball, reads the refusal, says so out loud, and moves on - the dog likes to run, so as long as he can reward him somehow the session works. That clause, 'as long as I can reward him somehow', is the whole concept in eight words, and it is said in passing rather than as a lesson.

Roxy is the same shape applied to space rather than to food, and it is worth keeping because it shows the concept is general. She sticks to the handler, so there is no distance, so a recall cannot be set up at all - he says he cannot train her like this if she just sticks with him, and his first fix is to create the distance himself by walking away. The prerequisite that is missing is not a reward here but the room to run the exercise; the move is the same one, which is to go and manufacture the missing precondition instead of proceeding without it.

Quotation held: its full wording does not match the current transcript. Preserved-track provenance is recorded in the private audit.

if you take a look at the evolutionary history of humans and of dogs both were built to to run

Watch at 18:31 ↗ · Current transcript match · Audio confirmation pending

so it can increase it can increase the value of that marker

Watch at 43:25 ↗ · Current transcript match · Audio confirmation pending

Quotation held: its full wording does not match the current transcript. Preserved-track provenance is recorded in the private audit.

Quotation held: its full wording does not match the current transcript. Preserved-track provenance is recorded in the private audit.

Unresolved

Whether touch and praise can be a reward for a frightened dog. The same programme says both, about different dogs, and in one dog's file about the same dog in different weeks.

Nothing in the corpus reconciles these. The obvious reading is that the two statements are about different points on the same relationship, and the Lola files are consistent with that - the refusal is on her intake day, the cuddling proposal comes later. But no trainer says so, and the test that would settle it, which is whether the dog seeks the contact or endures it, is never named as the criterion.

Unresolved

Whether a nervous dog can be assumed not to play. Two handlers behave as though the answer is opposite, and one of them says the assumption itself is the error.

These are not strictly contradictory - one is about testing, the other about what to do when the test comes back negative - but the corpus gives no threshold between them, and the Kodak session is evidence that a negative result can be wrong for months. Finn's case was never resolved on film.

My dog won't take treats or toys when he's nervous. What do I do?

You change his body instead of trying to change his mind, and the specific thing the trainers do is jog with him on the leash. The argument they make is chemical rather than psychological: running releases happy chemicals, and the dose that does it is a trot or a jog rather than a sprint. The reason that works when nothing else does is that the good feeling comes from the dog's own body, so it does not require him to trust you first, which is the whole problem with a treat or a hand. Once the run is reliably good, you start putting yourself into it - a pat, a grab, a moment of rough play immediately before and during the run - so that being handled starts to predict the feeling. Then you move the run to the end of a command, so the command buys it.

Isn't it cruel not to comfort a frightened dog?

The trainers' answer is that it depends entirely on whether the dog trusts you yet, and that with a dog who does not, comforting is the cruel option rather than the kind one. The reasoning is definitional: a punishment is whatever the animal wants to get away from, and for a dog who is frightened of you, a hand and a raised voice are exactly that. One handler demonstrates it on camera - loud praise, lots of it, and the dog visibly more scared. He is careful not to use the usual argument. He does not say you are rewarding the fear; he says that explanation does not make sense behaviourally, and then gives the other one. And he says the temptation was strong for him too, which is worth hearing, because the instruction is hard to follow rather than hard to understand.

How do I find out what my dog will work for?

By actually trying things, and by not deciding the answer in advance. The clearest statement of it in the corpus comes out of a session that changed direction four minutes in because the dog turned out to want a ball. The handler says the rule as something anyone can copy: always take some stuff with you and try it, because it can totally change your approach - try tug, try ball, try stuff - and do not assume that because a dog is fearful or nervous he will not do it, because you do not know that. What that looked like in practice was a nervous dog staring a ball down in the heat, indifferent to the stranger holding it, in a session where nothing else was asked of him at all.

What if nothing works at all?

That happens in this corpus, and what the handlers do is stop offering and go back a step rather than push. On one dog's first session the whole inventory was tried - ball, a rope toy, a long line so he could move, a test of whether the line touching him was the blocker - and none of it worked; what got written down was that it was early days, they were still figuring him out, and he would need ordinary walks before he could be got to engage at that level at all. That is engagement put upstream of reward choice. On another dog the search was still running weeks in, with the handler saying plainly that he had not figured out what this dog liked and would have to keep experimenting. So the honest answer is that the search can fail, and the answer to a failed search is more ordinary time with the dog, not a bigger reward.

He goes mad for squirrels. Can I use that?

Probably not as your main reward, and the reason is worth understanding because it is the opposite of what most people assume. A trainer looking at exactly that in one dog says it is the highest-value thing available and the fastest way to switch him on - and then sets it aside, because he cannot produce it when he wants it. A reward you cannot summon is not a tool. The useful reward is the available one, not the strongest one. That is also why a jog, which looks like a weak reward next to a squirrel, does more work: you can start it any time, anywhere, on a leash.

My dog jumps on everybody. Is there anything good about that?

Yes, and one handler builds a whole reward channel out of it. For a dog who badly wants contact, the cue that permits contact is itself a payment - so the jump goes on a word, any word you like, and is punished every other time. That makes it allowed when you ask for it and correctable when you do not, and it hands you a currency you did not have. It is the inverse of the fearful case on this page and it makes the point the page is actually about: what matters is having a channel you can open at will, not having a particular toy.

Comforting a frightened dog - stroking him, talking to him in a soothing voice, telling him it is okay. For a dog who does not yet trust the person doing it, that is the thing he is trying to get away from, which is the definition of a punishment.

Deciding in advance that a nervous dog will not play, and therefore never trying. The one session in this corpus that turned on a single discovery turned because somebody brought a ball and threw it.

Carrying on offering a refused toy. A refusal is information; the handlers in this corpus say it out loud and switch channel rather than repeating the offer.

Reaching for the dog's strongest motivator rather than his most available one. Prey is the fastest thing to switch Finn on and it is set aside precisely because the handler cannot produce it on demand.

Restraining a frightened dog close to you when something alarming passes and assuming the alarming thing did the damage. One handler names his own holding-back, together with a loud noise, as part of what sent the dog into a panic - and then says the restraint was still necessary for safety.

Asking the dog to come a second time so you can praise him properly. On a fearful dog the handler's own correction is that he should have released the dog off the first return immediately - the release is the payment, and asking again spends it.

You offer a toy and he goes still rather than towards it - or he takes a step back when you reach to play with him.

He takes food at home and will not take it anywhere else, so nothing you have works in the place where you actually need it.

Stroking him or telling him it is all right makes him worse rather than better - he pulls away, or he tries to get behind you.

The only time he looks like a normal happy dog is when he is moving fast - on a run, or chasing something.

He never leaves your side, so there is no situation in which he could get it wrong and nothing for you to reward.

The thing he wants most is one you cannot produce - squirrels, birds, other dogs at a distance - so it is useless as payment even though it works instantly.