How Dogs Learn: Reinforcement, Timing and Criteria

Every reward-based training plan, from teaching a puppy to sit to rebuilding a failed recall, rests on a small set of ideas about how dogs learn. This lesson explains the three you will use every day: what reinforcement really means, why timing matters so much, and how to raise your criteria without losing the dog.

You do not need to memorize terminology. You need to understand these ideas well enough to predict what a dog will learn from a situation, and to explain it to an owner in one sentence.

Reinforcement is defined by what happens next

A behavior that is followed by something the dog values becomes more likely to happen again. That is reinforcement. The important part is the definition: reinforcement is judged by its effect on the behavior, not by what we intended.

This catches owners out constantly. Consider a dog that jumps up on its owner when she comes home. She pushes it down, says "off" firmly and looks at it. She thinks she is correcting the jumping. But if the jumping is becoming more frequent, something in that exchange is reinforcing it, and the most likely candidate is attention. To a dog that has been alone all day, being touched, spoken to and looked at can be well worth jumping for, even when the tone is stern.

The same idea works in reverse. A treat the dog does not care about in a busy park is not a reinforcer there, however good it looked on the packet. The dog decides what is rewarding, and it changes with the setting. Working trainers often carry several kinds of reward (food of different value, toys, the chance to sniff or greet) and choose the one that beats the competition in that moment.

Punishment is defined the same way: something that makes a behavior less likely. Modern practice avoids physical and frightening punishment because it carries well-known risks, including fear, damage to the relationship and aggression, and because it tells the dog what not to do without teaching what to do instead. The skill a trainer builds is arranging things so the behavior you want is the one that pays.

Timing and markers

Dogs learn about the moment just before the reward arrives. If you reward a sit as the dog is already standing up again, you have rewarded standing up. The gap between behavior and reward is where most training quietly goes wrong.

Because it is hard to deliver food at the exact instant, trainers use a marker: a short, distinct signal, either a clicker or a word such as "yes", that is first paired with food until the dog understands that it means "that was it, a reward is coming". Once charged, the marker bridges the gap. You can mark the precise moment the dog's elbows touch the floor and deliver the treat a moment later.

A few practical rules follow from this:

• Mark first, then move your hand toward the food. If your hand moves first, the dog watches your hand, not the lesson.
• Mark the behavior you want, not the one just after it.
• Use the marker only when you mean it. A marker said without a reward loses its meaning.

Good timing is a physical skill, and it is one reason reading about training is not enough. Many trainers practice on a friend, a video, or a game of catching a ball as it bounces, before relying on it with a client's dog.

Criteria: raising the bar one step at a time

The criteria are what the dog must do to earn a reward right now. Progress in training means raising them gradually: a slightly longer stay, a slightly bigger distraction, a slightly different location.

The working rule many trainers follow is to change one thing at a time, and to move on when the dog is getting it right most of the time at the current level. If the dog starts failing, the step was too big. Go back to where it was succeeding, then take a smaller step. Trainers call this splitting rather than lumping.

This explains a pattern you will see in almost every client: "He does it perfectly at home and ignores me outside." The dog has not learned "sit"; it has learned "sit in the kitchen when there is nothing else going on". Dogs do not generalize easily. A new place, a new person or a new distraction is a new level of difficulty, and the criteria have to drop for a while in each.

Putting the three ideas together

Before any training plan, ask three questions:

1. What is reinforcing the behavior the dog does now?
2. What will I reinforce instead, and can I mark it at the right instant?
3. What is the smallest step the dog can succeed at today?

If you can answer those three, you can build a plan for almost any everyday problem. The article positive reinforcement dog training, explained for new trainers goes further into markers, shaping and reinforcement schedules. Lesson 3 applies these questions to a real case.