Why Did My Dog Suddenly Stop Listening?

It is common, and it is not spite

If your dog has stopped answering a cue they used to answer, you are in ordinary company. Trained behaviour is not a file a dog saves and keeps. It is a habit held in place by practice and by consequences, and habits fade when the conditions that built them change.

The words owners reach for are stubborn, spiteful, or he knows better. They are worth resisting, because they point you at the wrong fix. A dog who hesitates at a cue is usually telling you that the cue is no longer clear, no longer worth answering, or no longer possible in this particular place. None of those are moral failures, and all three come back with a week or two of short sessions.

Karen Pryor built this into her Ten Laws of Shaping in Don't Shoot the Dog! (Bantam, 1985). Law 9 is blunt: if a behaviour deteriorates, go back to kindergarten. Deterioration is expected. It is part of how teaching works, not evidence that something has gone wrong with your dog.

The three things that usually changed

Almost every sudden loss of a trained behaviour traces back to one of three causes, and they are easy to tell apart once you know the shape of each.

• You raised the difficulty too fast. Pryor's Law 1 says to raise criteria in increments small enough that the dog always has a chance of being reinforced. Ask for a long stay when the dog reliably held a short one and the behaviour does not stretch, it snaps. The dog has not refused the short stay. They simply never got offered it again.
• The reward dried up entirely. Most people stop treating the moment a skill looks solid. That is not thinning reinforcement, it is ending it, and a behaviour that pays nothing at all eventually stops being offered. The difference is the whole subject of our variable reinforcement guide.
• The context changed and the behaviour was never generalised. A sit learned in the kitchen is, to the dog, a kitchen sit. A new house, a park, a visitor, a different handler, you standing instead of crouching: any of these can make a familiar cue unrecognisable.

There is a fourth possibility, less flattering to us: the cue itself has gone noisy. If "come" is repeated in a tightening voice and then ends the walk, the dog has learned exactly what it predicts.

Practical take: Write down the last three times the cue failed and what was different about each one. Don't drill the same cue harder in the same setting that is already not working.

Go back to kindergarten, and go further back than feels sensible

Pryor's ninth law (pp. 54-59) is the repair, and it is counterintuitive. When a behaviour falls apart you do not push harder. You return to a version so easy the dog cannot get it wrong, and you rebuild from there. The common mistake is regressing one notch below the level that broke. Go well below it: back to the setting, distance and duration where the dog was fluent and relaxed, not merely competent.

Concretely, for a recall that has stopped working outdoors, go back to calling the dog from a couple of paces away, indoors, with nothing else happening, and pay every single one. For a stay that collapsed, go back to a count of two. For a loose leash that fell apart on the pavement, go back to the hallway; our loose-leash walking guide takes the same route.

Two other laws matter while you rebuild. Law 4 says that when you introduce a new criterion, relax the old ones, so if you are adding distance, stop caring for now about a tidy sit at the end. Law 2 says train one aspect at a time: distance, duration and distraction are three separate skills that happen to share a cue word, and shaping for two at once is how people end up back at the start. The wider set is broken down in our guide to the ten laws of shaping.

Relearning is fast. A faded behaviour usually comes back in days, not in the weeks the original teaching took.

Rebuild the payoff before you raise the bar again

Once the easy version is fluent, the trap is repeating the error that caused the problem: going straight from paying every repetition to paying none. Pryor's Law 3 is to put the current level on a variable schedule before raising criteria at all. The dog still gets paid. They just stop being able to predict which repetition pays.

Unpredictable payoff is not a compromise, it is an upgrade. The argument Pryor makes across pp. 36-40 is the slot machine: an intermittent, unpredictable schedule makes a habit stronger and much harder to extinguish than paying every time. A dog who does not know which recall earns the cheese comes faster, not slower. Occasionally pay out of all proportion, the jackpot Pryor describes on p. 31, for a repetition that was genuinely better than the rest.

Keep sessions short. Pryor notes (p. 65) that appetite for reinforcement drops off after roughly a dozen reinforcements, with an upper bound of something like eighty in a day. That argues for several brief sessions rather than one long one, and for Law 10: stop while the dog is still keen, on a repetition that went well. Law 8 is its other half, since breaking off a session for no reason is itself a punishment.

If sequencing all of that by hand is more planning than you want, Tavi's training module orders the steps and shows each repetition, which makes the regression decision less of a guess.

When it is a stimulus control gap, not a memory problem

Sometimes nothing has decayed. The dog can still do the behaviour, still offers it at home, and simply does not produce it when you ask. That is a stimulus control problem, and going back to kindergarten on the behaviour itself will not touch it.

Pryor sets out four conditions for a behaviour being genuinely under stimulus control (p. 81): the behaviour always occurs on the cue; it never occurs in the absence of the cue; it never occurs in response to a different cue; and no other behaviour occurs in response to this cue. Most household cues quietly fail at least one of those, and nobody notices until a distraction exposes it.

The tell is that the failure is shaped by context rather than by difficulty. A dog who sits instantly in the kitchen and never in the park has not forgotten anything; the cue was only ever attached to the kitchen. A dog who offers a down when you ask for a sit has not been taught to discriminate between the two. The repair is deliberate rather than dramatic, and it is covered properly in our guide to stimulus control: teach it in one easy place, then retrain it in each new place, changing one variable at a time.

Expect the first repetition in a new location to be poor. That is generalisation working normally, not your dog starting from zero.

When a sudden change is medical rather than behavioural

Some patterns make the visit more pressing: responding to a hand signal but not to your voice, hesitating at stairs they used to take without thinking, sitting slowly or crookedly, startling when approached from behind, sleeping more, or getting stuck in corners. Behaviour change is often the first thing an owner notices when a dog is unwell, and it can appear before anything looks physically wrong.

None of that is a diagnosis, and none of it means the training explanation is mistaken. Both can be true at once: something hurts, the behaviour becomes harder, it stops being reinforced, the habit fades. Getting the physical half ruled out means your retraining is not quietly fighting something training cannot fix.

What not to do while you rebuild

The instinct when a dog stops listening is to get firmer. The evidence runs the other way, and where it is thin it is thin about magnitude, not about direction.

In a controlled trial by China, Mills and Cooper (Frontiers in Veterinary Science 7:508, 2020), 63 dogs were trained in three matched groups. Dogs trained by reward-based trainers responded to the first recall cue 82.5% of the time, against 71.0% for the electronic-collar group, and sat faster afterwards, 1.36 seconds against 1.67. Hiby, Rooney and Bradshaw's survey of 364 owners (Animal Welfare, 2004) found 66% using vocal punishment, 60% praise and 51% food rewards, and reported that the more tasks an owner trained by punishment, the more problem behaviours they described. That last finding is correlational, so it cannot tell you which came first. The Complete Dog Breed Book (DK, 2nd edn, 2020, p. 322) is blunter: harsh handling "is simply not effective" and is "likely to cause more problem behaviours, including aggression and anxiety".

Be clear about what those studies cover. They tested ordinary obedience, recall and sit. They are not evidence about severe aggression, which is a different problem and a job for a qualified behaviourist rather than a guide.

The other habit to drop is repeating yourself. Saying the cue four times teaches the dog that the first three are optional. Say it once. If nothing happens, that is useful information rather than defiance: the setting is too hard, so step back and make it easier. The mechanics of building a cue that holds up are in our positive reinforcement guide.

Frequently asked questions

Why did my dog suddenly stop listening to me?

Usually because something in the training changed rather than something in the dog. The three common causes are criteria raised too quickly, so the dog never got a repetition they could succeed at; rewards stopped entirely rather than being thinned out, so the behaviour no longer pays; or the cue moved to a new place and was never generalised there. A cue that gets repeated or used to end something fun also loses value. A genuinely abrupt change in a previously reliable adult dog is worth a vet check first.

Is my dog ignoring me on purpose?

Almost certainly not in the way it feels. Dogs do not hold grudges or plan disobedience, but they do learn what a cue reliably predicts. If your recall has often meant the end of a walk, a bath, or an irritated tone, the dog has learned that and is responding to what the cue actually signals. That is not spite, it is accurate learning, and it is fixable by changing what follows the cue rather than by insisting harder on the cue itself.

How far back should I go when a trained behaviour falls apart?

Further than feels reasonable. Karen Pryor's ninth law of shaping is to go back to kindergarten when a behaviour deteriorates, and the usual mistake is regressing just one notch below the level that broke. Instead return to a version where your dog was fluent and relaxed, not merely correct: same room, short distance, no distractions, and pay every repetition. Then rebuild one variable at a time. Relearning is much faster than the original teaching, so this generally costs days rather than weeks.

Should I stop giving treats once my dog knows a cue?

Not stop, thin out. Pryor's third law of shaping is to put the current level on a variable schedule before raising criteria, meaning the dog still gets paid but cannot predict which repetition pays. She argues that intermittent, unpredictable reinforcement makes a habit stronger and harder to extinguish than reinforcing every time. Dropping rewards altogether is one of the most common reasons a solid behaviour quietly disappears. Keep occasional payment in place permanently, with the occasional oversized jackpot for a particularly good repetition.

Could my dog's sudden disobedience be a medical problem?

It can be, and it is worth ruling out. Training problems tend to build gradually, so a reliable adult dog who changes over a few days is a different picture. Things a vet will consider include hearing loss, pain that makes sitting, lying down or running uncomfortable, vision changes, and cognitive decline in older dogs. Signs that make a visit more pressing include responding to hand signals but not to your voice, hesitating at stairs, startling when approached from behind, or getting stuck in corners.

Tavi's guides are informational and not a substitute for professional veterinary advice, diagnosis or treatment. When in doubt, call your veterinarian.

Sources

• China, Mills & Cooper, Frontiers in Veterinary Science 7:508 (2020)
• Hiby, Rooney & Bradshaw, Animal Welfare 13 (2004)
• Karen Pryor, Don't Shoot the Dog! (Bantam, 1985)
• DK, The Complete Dog Breed Book, 2nd edn (2020)