How dogs learn: the science behind marker training and reward-based basics

Good dog training is not about dominance, relationship status, or making the dog understand who is in charge. It is about communication and association. Dogs are extraordinarily good at learning what behaviour produces a pleasant outcome — and they repeat those behaviours. Everything in positive-reinforcement training follows from that single insight.

The AVSAB's position statement on humane dog training is unambiguous: reward-based methods have the strongest evidence base for effectiveness and the lowest risk of welfare harm. Punishment-based techniques are associated with increased fear, aggression, and the breakdown of the owner-dog relationship. The approach in this guide — operant conditioning through positive reinforcement, paired with classical conditioning for marker training — is the professional consensus of the CCPDT, APDT, IAABC, and AVSAB.

This article explains the core mechanisms so you can apply them intelligently, not just mechanically follow steps.

How dogs learn: the basics of operant conditioning

Operant conditioning is learning through consequences. When a behaviour is followed by something the dog finds rewarding, that behaviour becomes more likely in the future. This is positive reinforcement — adding something good to increase a behaviour. The key insight is that the dog decides what is rewarding, not the owner. Food is the most universally effective reinforcer for most dogs, but play, social access, sniffing opportunities, and being let off lead are all valid reinforcers for the right dog.

• Positive reinforcement (+R): add something the dog likes → behaviour increases. The foundation of reward-based training.
• Negative punishment (−P): remove something the dog likes → behaviour decreases. For example, turning away and ending interaction when a dog jumps. No physical discomfort involved.
• Positive punishment (+P): add something unpleasant → behaviour decreases. The AVSAB recommends against this as a first-line tool due to welfare risks.
• Negative reinforcement (−R): remove something unpleasant → behaviour increases. Avoid designing training around this quadrant — use +R instead.

Why timing is everything

Dogs make associations between a behaviour and a consequence within approximately one to two seconds. Beyond that, the connection is lost. This is why "good dog" said three seconds after a sit has landed is nearly useless — the dog has already done something else in the interim and that is what gets reinforced. Precise timing is the single most important skill in training, and the marker — a clicker or a verbal "yes" — exists to solve this problem.

What a marker is and how it works

A marker is a brief, precise signal that tells the dog "that exact thing you just did has earned a reward." The clicker is the classic tool — a small box that makes a consistent, distinctive click — but a short verbal marker such as "yes!" works equally well for most owners. The marker is not the reward itself; it is a promise that a reward is coming, and it bridges the timing gap between behaviour and treat delivery.

Before using a marker in training, it must be charged — associated with reward through repetition. This is classical conditioning: click, then treat, repeated 10–20 times in a neutral setting until the click alone produces an anticipatory response (ears perk, dog orients towards you). From this point forward, the marker carries predictive value and can be used to mark behaviours precisely.

• To charge a clicker: in a calm environment, click once → immediately deliver a treat, regardless of what the dog is doing. Repeat 10–20 times in two to three sessions.
• One click = one treat, always. Never click without treating — it erodes the predictive value of the marker.
• The marker fires at the peak of the behaviour: as the dog's bottom touches the ground on a sit, not before, not as they begin to rise.
• Reward delivery can be slower — take 2–3 seconds to reach the treat pouch. The marker has already told the dog they succeeded.

Choosing and using rewards effectively

Food rewards should be small — pea-sized pieces of soft, smelly treats work best for most dogs. Variety helps: use higher-value treats (small pieces of cooked chicken, cheese, or commercial training treats) for new or difficult skills in distracting environments, and lower-value treats (the dog's regular kibble) for practised behaviours in calm settings. This concept — variable reinforcement schedules and reward hierarchies — is well established in applied behaviour analysis.

How many repetitions before a dog learns a cue?

Dogs typically need 20–50 repetitions in a consistent context before a new behaviour is reliable in that setting, and considerably more before it generalises to other environments (this is proofing — covered in a separate guide). Short sessions of 3–5 minutes produce faster learning than long sessions, because a fresh, engaged dog makes associations more efficiently than a mentally fatigued one. End every session before the dog loses interest — on a success, not a failure.