Why doesn't this trainer use treats?

Dutchess's go-home lesson carries the fullest statement of the reliability argument, and it is delivered to a family whose own intake form named food as one of her two motivators. The trainer volunteers the history - they used to use treats a lot more and do not any more - and gives the reason as a property of the reward rather than a moral position: a treat is the most reliable thing there is while you hold it and while the dog does not want something more, and it becomes insanely unreliable the moment the dog even slightly cares about something else. Then he puts the failure in the dog's imagined voice, which is the version an owner remembers: you are out of treats, you say sit, and the dog finds that funny, because it cannot see a treat in your hand, so the relationship is transactional. Her earlier in-program lesson supplies the general rule underneath it, credited to an online course the trainer took - the question is whether the reward for not listening is greater than the reward for listening - with the punch-in-the-face analogy attached to make it concrete.

The Kodak household material contains different explanations of food and play. In the owner-lesson passage, the trainer uses an example that explicitly names Kodak while explaining his preference for engagement through play rather than attention directed at visible food. His friendship analogy expresses that preference; it does not independently establish the effects of food across dogs or prove the owner report is the strongest evidence in the archive. A separate Rocko session, TSAWDf3ifr8, describes occasional food rewards and distinguishes responding to a spoken cue from responding to a displayed treat. Read these as two approaches described in these lessons, not as a controlled comparison or proof of an organization-wide policy. The source caveat identifies iZO2WNrinFg as partly involving Rocko, so its individual passages must be attributed by context rather than assigning the entire recording to one dog.

Stormie's two sessions show the policy in its weakest and its strongest form, a fortnight apart and from two different people. In the in-house session it is barely a statement at all: the handler has failed to get a down for three days, tries tapping the nose, then thinks he may as well try a treat, and the treat is refused too - and the sentence that begins to explain the house preference is cut off mid-phrase by the other speaker offering food anyway. At the home lesson the same policy arrives as a complete argument, and it is the funniest phrasing of it in the corpus: they do not use treats, because a treat-trained dog ends up thinking you told me to sit but I do not see a treat, which is funny, because I only sit for treats. What makes that session valuable is that the limit comes first - you really cannot train without some kind of reward, the programme uses a lot of pressure, and the dogs are not going to enjoy themselves if it is all pressure and nothing positive, which would be old-school training, and they are balanced. So the no-food rule is stated inside a commitment to having a reward at all. Her family's intake answer sits oddly against the policy: their reported motivator is play, and their reported state of her sit is that she will do it if there is food in the hand to lure her.

Roxy is the case where the experiment is actually run, and that is what makes her file the most honest evidence on this page. At session six the trainer introduces treats for a few minutes to see what she needs, the sit and the down come without the e-collar, he notes that he did not need it - and then he stops, says that is enough treats, and gives the reason. It is not that food failed. Treats can be a good thing and treats can also be a bad thing, and the specific bad thing he is avoiding is the handover: give her a treat once and she may go home, look at her owners, and wait to see the food before doing something she already knows. The decision is made entirely about what the family will get after he leaves. The same reasoning disposes of a clicker, on the more prosaic ground that the house uses verbal markers and a clicker is one more thing to carry. Her own intake lists food first among her motivators.

Lola's first in-program lesson is where the relational argument is made at length and where the replacement is specified, which is the part most versions of this argument skip. The objection starts with the labour - you have to get a treat pouch, fill it with tasty treats and carry it everywhere - and then turns on what it does to the dog: it makes the thing transactional, it makes the handler a bank, and it produces a dog sitting there looking at you instead of going into the bushes, which the trainer calls creepy and, more to the point, not the picture the family described wanting. The phrase he lands on is buddies, not bank. What replaces food is release to freedom and games, and he then says what makes a game payable, which is the most useful passage on this page: a game needs a shared objective, real strategies, and a real chance of losing. If he beats her every time there is no contest and it is no fun; if he lets her win every time by using a cheat code it is also no fun; and if it is too predictable it is no fun either. For Lola herself none of it resolves inside the corpus - the ball frightens her, tug is never established, and what actually pays her is freedom, sniffing and touch - and another session in her file says that is allowed to take time, because sometimes you do not find the interesting thing until the dog has opened up.

Deuce's lessons supply the economics that the rest of this concept rests on, and they are taught to the owners as theory rather than stated as policy. Disobedience is described as a calculation the dog runs - command, do I want to do this, or is there something I want to do more, and does doing this make me happier than the punishment - and then made concrete with an analogy that recurs word for word in another dog's lesson: nobody lets a stranger punch them in the face for fifty cents, and quite a lot of people would consider it for five million dollars. The criterion that follows is borrowed, and credited as borrowed, from a month-long course one of the trainers took: make sure the dog's reward for listening is greater than his reward for not listening. Two operational rules come out of it. The toy's excitement is a quantity to be built up rather than a preference to indulge, because the more excitement it carries the more likely it is to outrank whatever else is competing. And the payment is deliberately unpredictable - do not use the tug every time, and specifically not every other time, because he will find that pattern; in a later session the handler says the tug comes out at random times as a matter of course.

Zeus's sessions give the claim its most careful form, which is a claim about a condition rather than a dismissal. Reward-only training is said to fall flat and fail specifically when the distraction is high, and the reason offered is that the dog has never been reinforced either way in that state - which is why obedience has to be worked when the dog is ramped up, amped and wanting something rather than in a calm room where nothing is happening. His second lesson adds the other half, and it opens with a concession rather than an objection: with a piece of pepperoni the trainer could teach a dog to lie down, and that is simply not how he taught it that day. Two reasons are given. It is often slow. And - the larger one - it does not teach the lessons that teaching with pressure does, because there is no obligation in a lure: with pressure the dog is told he is obliged to do the thing, then the pressure goes away and a positive reward follows, which could be a treat, could be play, could be freedom. That passage is the only place in the corpus where the positive reward at the end of a pressure sequence is allowed to be a treat.

Luna's first in-program lesson repeats the borrowed criterion almost word for word from a different dog's lesson - the reward for listening has to be greater than the reward for not listening, credited to a course the speaker took a while back - which is what makes it a fixed piece of the owner curriculum rather than one handler's improvisation. Her file matters most here for the other side of the conflict. Her family's intake form lists food first among her motivators, and reports that the way she performs a down, and the way she responds to the yes marker, is with food in the hand to lure her. The programme's answer to that is to remove the currency the family has actually demonstrated and replace it with pressure released into exploring.

Finn is the demonstration rather than the argument, and it is included because it shows the reliability claim happening rather than being explained. In a hardware store a stranger offers him a treat as part of a planned greeting, and he will not take it. The handler does not push it - he says the dog does not care much about the treat, and then, plainly, that if he does not want it he does not want it. On a dog whose whole case is fear, a treat is not a reward that can be counted on to be there when it is needed.

expedience the accident so we used to use treats a lot more we don't use them anymore because they're the most reliable if you have the treat or if the dog doesn't want something more than the treat right but they're like it becomes insanely unreliable if the dog even slightly cares about something more than the treat

Watch at 01:43 ↗ · Current transcript match · Audio confirmation pending

Let's say the trainer has a treat in front of them before he even says, like, sit. to the dog he's gonna get used to the treats saying oh when i see treats i sit or down or whatever you know they don't wait for the command

Watch at 08:20 ↗ · Current transcript match · Audio confirmation pending

Like, if there's one time I give her a treat, and let's say she goes back to her owners and then she just doesn't do it and she just looks at you, but she knows it, she's like, well, where's my food? I need to see my food first.

Watch at 12:47 ↗ · Current transcript match · Audio confirmation pending

psychology of a dog's brain. Their brain usually works with like as far as commands go okay command, do I want to do this or is there something I want to do more? Okay, I want to do this more. Does doing this make me happier than the the punishment?

Watch at 15:46 ↗ · Current transcript match · Audio confirmation pending

So we try to get the dog to be as excited about the toy as possible because the higher level of excitement you have from the toy then the more it's likely to to outrank the other distractions

Watch at 15:07 ↗ · 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.

Unresolved

How strong the ban actually is. The corpus records it as a flat house rule and as a sparse top-up, and the handler who uses food says on camera that his colleagues do not.

The Zeus passage is on this side only in part: it names a treat as one of the acceptable positive rewards at the end of a pressure sequence, which the flat statements do not allow. The corpus contains no session in which the two positions are put to each other, and the handler who uses food is the one who flags the disagreement, which is the only reason it is on the record at all.

Unresolved

The families say food and the programme says no food. Several intake forms name food as the dog's motivator, and two of them report that the dog's existing commands only happen with food in the hand - which the trainers treat as proof of the problem rather than as the family's evidence about their own dog.

This is laid out and not settled. The two readings of the same fact are both available: that a dog who only works with food in the hand is the failure mode the trainers describe, or that food is the one currency the family has demonstrated works on their dog and the programme discards it. What the corpus does not contain is any session in which a family's food answer is discussed with them as evidence rather than as a habit to be corrected. Worth noting on the record: Stormie's form names play, not food, as her motivator, and in her in-house session the treat was refused - so at least one dog's evidence does not support either side.

Unresolved

Whether food fails or is merely slow and narrow. The concession and the ban are made by the same programme in the same month.

These are reconcilable if 'does not work' is read as 'does not work under distraction', which is how the same speaker qualifies it in the same breath. They are recorded separately here because the unqualified version is the one that reaches a viewer, and because the pepperoni passage is on low-confidence audio and so is the weaker of the two on the record.

Why doesn't this trainer use treats?

Three reasons, given separately across different dogs. Reliability: a treat is the most reliable reward there is while you are holding it and while the dog does not want anything more than it, and it becomes, in the trainer's own word, insanely unreliable the moment the dog even slightly cares about something else - which is exactly the moment you needed it. The relationship: a treat-trained dog looks at the hand rather than the person, and a handler with an empty hand stops being anyone the dog owes anything to. And the handover: a dog paid with visible food can go home and wait to see the food before doing something she already knows, and her family will call that a training failure. What they do not say is that food does not work. One of them says the opposite - that a piece of pepperoni would teach a dog to lie down perfectly well.

My dog only listens if he can see the food. How did that happen?

Almost certainly because the food was visible before the word. The handler in this corpus who uses treats most is also the one who explains this most exactly: if there is a treat in front of the dog before anyone says sit, the dog learns that the sight of treats means sit, and he stops waiting for the command. His fix is a sequence rather than a ban. Wait until the dog is out of position, say the word, and produce the treat only after he has answered - so that what he responded to was your voice. He also does not pay every time; roughly one command in five, as a top-up for motivation rather than as the teaching mechanism.

So what do you use instead?

Release to freedom and games, in that order, with the reward chosen from what the individual dog actually wants. The important and rarely stated part is what makes a game work as payment. A game needs a shared objective, real strategies and a real possibility of losing: if the handler wins every time there is no contest and it is no fun, if he lets the dog win every time by using what he calls a cheat code it is also no fun, and if it is too predictable it is not a game either. On top of that, the value of the reward is treated as something you build rather than something you have - the dog is made as excited about the toy as possible, precisely so that the toy can outrank whatever else is competing for him. And it is paid out irregularly, because a dog who can predict the schedule starts choosing which repetitions are worth answering.

Is positive reinforcement-only training wrong, then?

The claim the corpus actually makes is narrower than that, and it is worth quoting precisely because the short version travels badly. What is said is that rewarding the right behaviour on its own often falls flat and fails when there are higher distractions - and the reason given is not that reward is weak but that the dog has never been reinforced either way in that state, which is why the work has to be done when the dog is ramped up rather than in a calm room. The second objection is about content rather than power: teaching with a lure is often slow, and it does not teach the lessons pressure teaches, because there is no obligation in it. Teaching with pressure says you are obliged to do this, then the pressure goes away, and then a positive reward follows - and in that passage the reward is allowed to be a treat.

Why won't he do it for a treat when there's another dog around?

Because he is running a comparison, and on that occasion your treat lost. The trainers teach this to owners as arithmetic rather than character. The dog's question is whether he wants what you are asking for more than the thing he would rather be doing, and whether refusing costs him anything. The version they hand families is a borrowed one, credited to a course one of them took: make sure the reward for listening is greater than the reward for not listening. And the analogy they use twice, in two different dogs' lessons, is being offered fifty cents to be punched in the face, against five million for the same punch. Neither answer is about how good a dog he is.

But I told you on the form that food is what motivates my dog.

You did, and so did most of the families on this page - food appears among the named motivators on the intake forms of six of these nine dogs, and two forms report that the dog's existing commands only happen with food in the hand to lure him. The programme reads that second answer as the problem rather than as the solution: a dog who performs only with food visible is precisely the dog the trainers describe when they talk about checking the hand. That is a real disagreement and this page does not settle it. What the corpus does not contain is a session where a family's food answer is discussed with them as evidence about their own dog, rather than as a habit to be replaced. What it does contain, on one dog, is the experiment: treats were introduced, the sit and the down came without the collar, the trainer said so - and then he stopped anyway, for reasons about your household rather than about the session.

Should I use a clicker?

The house answer is no, and the reason is unglamorous: they use verbal markers, so a clicker is one more thing to carry. It came up because one of the families had one, and the trainer's reply was that he never uses them. That is worth knowing mainly because it tells you what the marker system is for in this method - the words do the clicker's job, which means the thing you have to get right is the timing of your own voice rather than the possession of a device.

Holding the treat out before saying the word. That teaches the dog that the sight of food means sit, so he answers your hand rather than your voice and stops waiting for the command at all.

Paying every time, or paying every other time. A predictable schedule lets the dog work out which repetitions are worth answering, and every other time is the pattern he finds fastest.

Training only where the dog is already calm. Obedience built in a quiet room has never been reinforced in the state the dog will actually be in when you need it, so it gets blown off.

Reading a refusal as defiance. The trainers describe it as a price comparison the dog has run and lost interest in - the question to ask is whether what you are offering beats what he would rather be doing, and whether refusing costs him anything.

Pushing a refused treat, particularly on a frightened dog. The handler in the store simply stops - if he does not want it, he does not want it.

Playing a game the dog cannot win, or cannot lose. If you beat him every time there is no contest, if you let him win every time it is a cheat code, and if it is too predictable it is not a game - and a game that is not a game does not pay.

He does everything you ask in the kitchen and nothing you ask in the street.

He looks at your hand, or at your pocket, rather than at your face when you speak to him.

You say sit with nothing in your hand and he stands there looking at you as though you have not finished the sentence.

He has got fussy about his own dinner because he has learned that holding out sometimes produces something better.

He will not take food at all in the places that frighten him, which is exactly where you needed it to work.

You have worked out that he answers about every other time you ask, which usually means he has worked out your schedule.