How to teach a dog anything, one agreed rule at a time

A family we know spent a winter watching training videos. The dog will sit in the kitchen, beautifully, for the person holding the treat pouch. He will not sit on the sidewalk, will not sit for the teenager, and puts his front paws on every visitor who comes through the door. Nobody in that house is doing anything wrong, exactly. They are just each doing something slightly different.

Almost every stalled training project looks like this from the inside. The method is fine. What is missing is the boring part underneath it: the same word, the same criteria, the same consequence, from everyone, over and over, until the dog can predict what happens next. A mediocre technique applied identically by four people beats an excellent one applied four different ways.

One word per behavior, and everyone uses it

Watch a household for an hour and count the words aimed at a dog on the couch. Off. Down. Get down. No. Move. Out. To the dog these are six unrelated noises, one of which — down — you also use when you want him to lie on the floor. He is not being stubborn. He is being asked a question in a language that changes every twenty minutes.

The fix is a five-minute conversation and a list on the fridge. Pick one word for each thing, write down who says what, and include the hand signal if there is one, because most dogs read the hand long before the word.

• The word for getting off furniture, and the word for lying down. Never the same word.
• Whether the dog is allowed on the couch. Not sometimes, when it is cold, and not just for the person who lets him.
• What happens when he jumps up: everybody turns away and waits, or everybody says something. One or the other.
• Who feeds, when, and whether anything at all comes off a plate at the table.
• The recall word, which is used only when you can back it up — never to end a good time and never twice in a row.

Say the cue once. If you say sit three times because the first two did nothing, the dog learns the cue is actually sit-sit-sit, and eventually sit-sit-SIT. If one repetition gets no response, the exercise was too hard for the situation, which is information rather than defiance.

Timing does more work than the reward does

Dogs connect what they are doing with what happens roughly a second later. Anything slower and you are rewarding whatever came next. You call the dog, he arrives, you spend four seconds digging in a pocket while he wanders off to sniff the fence, and the food lands on wandering off.

This is what a marker fixes. Pick a short sound — yes, or a click — and pair it with food a few dozen times until the sound alone makes the dog's head snap around. From then on the marker does the precise timing and the treat can take its time arriving. It is the difference between telling a dog he was right and hoping he works out which part you liked.

When it falls apart, make it easier

The universal instinct when a dog fails is to repeat the same request more firmly. The useful move is the opposite: drop the difficulty until he is right four times out of five, then add one thing back. Distance, duration and distraction each count as a separate thing, and you raise one at a time.

1. Ask for the stay in a quiet room with you standing right there. Two seconds. Release, reward.
2. Same room, same two seconds, but take one step back and return before releasing.
3. Build to about ten seconds at one step, then go back to two seconds and try three steps.
4. Move to the hallway and start over at two seconds and no steps, which will feel like a waste of time and is not.
5. Add a mild distraction — someone walking past — and again cut the duration in half.

A dog who holds a stay for thirty seconds in the living room and breaks instantly in the yard has not forgotten anything. Dogs are poor generalizers; a behavior belongs to the place it was learned until you deliberately teach it somewhere else. Practicing in five ordinary locations does more than another week of practice in the kitchen.

Manage what you have not trained yet

Training takes weeks. In the meantime, prevention is not cheating. A latching trash can, a baby gate across the stairs, shoes off the floor, a leash clipped on before the door opens. Every rehearsal of a habit you dislike makes it stronger, so the cheapest training tool in the house is usually a closed door.

Short sessions, and knowing when to stop

Two minutes, three or four times a day, beats a heroic Saturday session. Stop while the dog still wants more, and finish on something he already does well rather than on the hard new thing. Training that ends in frustration teaches him that the training game gets worse the longer it runs.

One boundary worth naming: fear, guarding and anything that has involved teeth are not consistency problems, and they get worse with the wrong homework. That is work for a qualified behavior professional. A sudden change in a dog who was previously fine — a house-trained dog having accidents, a friendly dog snapping when touched — deserves a vet visit first, because discomfort shows up as behavior long before it shows up any other way.

The part that surprises people is how ordinary the successful households look. Nobody in them is a gifted trainer. They just settled on the words in February and were still using the same ones in May.