How dogs learn: the plain-language guide to reinforcement and timing

• Behavior that is rewarded is repeated. Behavior that is ignored tends to fade. That is the whole engine.
• Timing is the skill most owners underestimate: a reward that arrives three seconds late often rewards the wrong thing.
• Dogs are poor generalisers — "sit" in the kitchen is a different lesson from "sit" in the park until you teach both.
• Short, frequent sessions beat long ones. Most dogs learn fastest in blocks of 3–5 minutes.

The core principle behind almost all dog training is operant conditioning: behavior is shaped by what happens right after it. If sitting at the door leads to the door opening, your dog sits at the door more. If jumping on guests leads to attention, your dog jumps more. Your dog is not being stubborn or clever when it repeats an annoying habit — it's doing exactly what has worked before.

That cuts both ways, and it's good news. Every interaction is a training session whether you meant it to be or not, which means you can change most everyday behavior by changing what it earns. Ask two questions about any habit you want more or less of: what does my dog get out of this, and can I make a better behavior pay off instead?

Argus focuses on the first two rows. Adding good things and withholding them teaches the dog what to do and keeps training something your dog wants to take part in. The positive reinforcement guide explains why the other two rows cause more problems than they solve.

Dogs connect a consequence to whatever they were doing at the moment it arrived. If you ask for a sit, the dog sits, and by the time you've fished a treat from your pocket it has stood up again, you've just rewarded standing up. Aim for the reward (or a marker that promises one) to land within about one second of the behavior you want.

That's hard to do with food alone, which is why trainers use a marker — a word like "yes" or a clicker — to pinpoint the exact moment the dog got it right. The marker buys you a few seconds to deliver the treat without losing clarity. The marker words and clickers guide covers how to set one up.

Alongside consequences, dogs constantly learn which things predict which. This is classical conditioning — the reason a dog runs to the kitchen at the sound of the food bin, or hides when the nail clippers come out. Nobody taught those behaviors directly. The dog simply noticed what tends to happen next.

Associations are why cues work at all: a word means nothing until it reliably predicts that a certain behavior will pay. They're also why a cue can be "poisoned." If "come" often means the end of fun, a bath, or a telling-off, your dog learns that "come" predicts bad news and starts hesitating. Protect your most important cues by making sure they almost always predict something good.

• Good associations to build: harness means a walk, the crate means a chew, the vet car park means treats.
• Associations to avoid: recall means the leash goes on and the park is over; handling means pain; the crate means being left alone for hours.

Dogs are poor generalisers. People assume that once a dog knows sit, it knows sit everywhere. In practice the dog learned something much narrower: sit, in the kitchen, facing you, when you're holding a treat and nothing else is going on. Change the room, your position, or the level of distraction, and to the dog it can feel like a brand-new task.

The fix is deliberate: once a behavior is reliable in one place, practice it in a second easy place, then a third, and only then add distractions. Trainers call this proofing. It's covered in every obedience guide in the library, because it's where most real-world reliability comes from.

• Keep sessions to 3–5 minutes for puppies and 5–10 for adults, and do several a day rather than one long one.
• Say a cue once. If the dog does not respond, make the task easier rather than repeating yourself.
• Reward the behavior you want as it happens, including when you did not ask for it.
• Manage the environment so the dog cannot rehearse the habits you are trying to replace.
• Change one thing at a time: more duration, more distance, or more distraction — never all three together.
• When something isn't working, assume the task is too hard or the reward too small before assuming the dog is stubborn.