Free shaping a dog with a marker, clicker or word

Quick answer

Free shaping means reinforcing small steps towards a behaviour without luring, pointing, guiding or saying anything. The dog offers, you mark, the dog works out what earned the mark, and the behaviour is built out of the dog's own choices.

Free shaping asks the dog to solve the problem. Your job is to be silent, still and precise.

The marker is the whole mechanism. It can be a clicker or it can be a word, and for most household dogs a word works well, but whichever you choose it has to mean exactly one thing: the food is coming, and it is coming because of what you just did at that instant.

The most common reason shaping fails is not the dog. It is criteria that jump too far, a rate of reinforcement that is too low, and a human who cannot stop helping.

• Charge the marker first: marker, then food, dozens of times, with no behaviour required.
• Pick one observable criterion at a time and stay on it until it is reliable.
• If you go more than a few seconds without being able to mark, your criterion is too big. Split it.
• Add the cue only after the behaviour is fluent, never at the start.
• End the session while the dog still wants more, not when it gives up.

Decide in 60 seconds

Why a marker works at all

Dogs learn from consequences, but only from consequences that arrive close enough in time to be connected to the behaviour.

The problem is mechanical. By the time your hand reaches your pocket, finds a piece of food and delivers it, the dog has moved, looked away, taken a step and shifted its weight. The food arrives attached to whatever the dog was doing when it arrived, not to what you intended to reward.

A marker solves this by being instantaneous. A sound that has been paired with food many times becomes a predictor of food, and a reliable predictor takes on the properties of the thing it predicts. That is what a conditioned reinforcer is. In practical terms it lets you say "that, exactly then" from across the room, and buy yourself the two seconds you need to deliver.

Three consequences follow from that mechanism, and they explain almost every rule in shaping.

The marker must be precise. Its value comes from marking one instant. A drawn-out "gooood boyyy" marks a second and a half, which in dog-learning terms covers three or four different behaviours.

The marker must always be paid. Its value comes entirely from the reliability of the prediction. Mark without delivering and you have weakened it. Do that repeatedly and it stops working.

The marker is information, not a summons. Using it to get your dog's attention teaches it to mean "look at me", which destroys its use as a behaviour marker.

Clicker or marker word

Both work. They differ in ways that matter for specific dogs.

A clicker is uniform every time, extremely brief, mechanically identical regardless of your mood, and unlike anything else the dog hears in the house. That gives it the cleanest possible signal. It also occupies a hand and it is easy to leave in the wrong coat.

A marker word is always with you and costs nothing. Choose a short, sharp word you do not otherwise use, and say it the same way every time. The cost is precision, since a word takes longer to say and your voice varies. In practice this matters most for fast, fine-grained behaviours and hardly at all for a mat settle.

Reasons to avoid a clicker specifically. Sound-sensitive dogs, and dogs with noise phobia, can find the click genuinely aversive. Signs are flinching, ears back, moving away from the click, or shutting down. Test it once at a distance in a pocket before committing. Quieter clickers exist, and a tongue click or a pen click are quieter still.

Deaf dogs and dogs losing hearing. Use a visual marker: a thumbs-up, a small hand flick, or a torch flash held consistently in the same place. Everything else in this guide applies unchanged, but the marker has to be within the dog's field of view at the moment of the behaviour, which changes where you stand.

Charging the marker

Before any shaping, the marker has to mean something.

Sit with the dog and a supply of small food. Mark, then deliver. Do nothing else. Do not ask for a behaviour, do not wait for one, and do not care what the dog is doing.

Keep the order absolutely fixed: the marker happens, and only then does your hand move. If your hand moves first, the hand becomes the real marker and the sound becomes decoration. This is the most common error in the whole exercise and it is invisible unless you film yourself.

You know it has worked when the sound produces an immediate head snap towards you, even when the dog was not paying attention.

Recharge for a few repetitions at the start of any session after a long gap.

Running a shaping session

Set up so you cannot help.

Sit or stand still, hands neutral, food out of sight. The whole method depends on the dog generating the behaviour, and any pointing, leaning, tapping or encouraging turns free shaping into luring.

Silence is part of the technique. Talking to a dog that is problem-solving interrupts the problem-solving.

Choose one criterion, and make it observable.

"Looks at the mat" is a criterion. "Engages with the mat" is not, because you cannot decide in a quarter of a second whether it happened. If you cannot judge it instantly, you will mark inconsistently, and inconsistent marking is what confuses dogs.

Keep the rate of reinforcement high.

Early on, marks should come thick and fast. If several seconds pass and you have not been able to mark anything, the criterion is too demanding. Split it into something smaller. There is no prize for a hard first step.

Use reinforcement placement to reset the dog.

Where you put the food decides where the next repetition starts. Toss it away from the mat and the dog has to come back to it, which gives you a clean approach to mark. Feed it in position and you build duration in place. Both are useful; choose deliberately rather than by accident.

Work in loops.

Stay on one criterion until the dog is offering it confidently several times in a row. Then raise it once. If the dog fails twice at the new level, go back, get three easy wins, and try again with a smaller increase.

Expect the burst.

When you stop paying a level the dog has been paid for, it will first do more of that thing: faster, harder, more emphatically. That is an extinction burst and it is not a failure. It is where variation appears, and variation is what you are fishing for. Owners very often decide the dog "does not understand" and go back to luring during exactly this window.

Stop early.

End while the dog is still keen, on a repetition that succeeded. Several very short sessions across a day beat one long one, and a dog that has learned sessions end in frustration becomes hard to start next time.

A defined prop, a clear space and nothing else going on. Free shaping needs a boring room.

A worked example: mat settle

The criteria below are the whole plan. Each line is a level, and you only move on when the current one is easy.

Throw the food off the mat for steps one to seven, so the dog has to leave and return and you get a fresh approach each time. From step nine onwards, deliver the food between the dog's front paws, which pays the position and encourages staying.

Do not name it yet.

Adding the cue

The cue goes on last. This is not a stylistic preference, it is the difference between a reliable behaviour and a fragile one.

Wait until the dog is offering the finished behaviour promptly and repeatedly without hesitation. Then say the cue as the dog begins to do it, so the word overlaps the behaviour. After a number of repetitions, say the cue slightly earlier, just before the dog starts. Eventually say it while the dog is doing something else and see whether it responds.

Once the cue works, stop paying the behaviour when it is offered without the cue. That is what teaches the dog the word is doing the work.

If you attach the cue while the behaviour is still rough, the cue permanently means the rough version, and you will spend longer fixing that than you would have spent shaping properly.

Where it goes wrong

Marking late.

The classic version: the dog sits, and by the time you mark it has already begun to stand. You have just reinforced standing. Filming a minute of a session on a phone shows this immediately, and almost nothing else will.

Marking without paying.

Every unpaid marker weakens the prediction. If you mark by mistake, deliver anyway. An accidentally reinforced behaviour costs one repetition; a devalued marker costs the whole system.

Lumping.

Jumping from "one paw on the mat" to "lie down on the mat" because the dog did it once. One success is noise. Three or four in a row is a pattern.

Helping.

Pointing at the mat, tapping it, moving it closer, leaning towards it, or saying "go on". Each of these gives the dog a shortcut that will not be there later, and the behaviour collapses the moment you stop.

Reinforcers that break the rate.

Hard biscuits take chewing, and chewing kills the tempo. Use something small and soft that disappears in one swallow. Save the exciting food for genuinely hard steps.

Training a tired, full or overheated dog.

A dog that has just eaten will not work for food. A dog at the end of a long walk has nothing left. A panting dog cannot take food comfortably and is not learning.

Using the marker as a recall.

It becomes an attention cue and stops functioning as information about a behaviour.

Punishing wrong guesses.

Free shaping runs entirely on the dog being willing to try things. Any correction for a wrong attempt reduces offering, and a dog that stops offering cannot be shaped.

Loose body, soft face, and a dog that keeps offering. That is what a session going well looks like.

Dogs that will not offer anything

Some dogs sit and stare, waiting to be told. This is common in two groups: dogs that have only ever been trained by luring and physical prompting, and dogs whose initiative has been corrected in the past.

The fix is to lower the bar until it is impossible to fail.

Put a prop on the floor: a cardboard box, an upturned tub, a mat. Novel objects invite investigation.

Then mark literally any movement. A head turn. A blink towards the object. A shift of weight. A single step in any direction. The goal for the first several sessions is not the target behaviour at all, it is teaching the dog that doing something makes food appear.

Keep sessions extremely short. Two minutes is plenty, and finishing while the dog is still curious is what brings it back next time.

Expect this to take days rather than one session. Once a dog discovers that its own behaviour controls the outcome, offering usually arrives suddenly and then never goes away.

What changes with the dog in front of you

Herding breeds and many gundogs offer readily and can escalate into frantic, high-arousal guessing. Slow your food delivery, use calmer reinforcers, choose criteria that involve stillness, and take breaks before the dog tips over rather than after.

Independent scent hounds and primitive breeds often take longer to start offering and need genuinely valuable food. Slow is not the same as unwilling.

Guardian breeds frequently work at a more deliberate pace, and pressure produces disengagement rather than speed. Let the session be slow.

Puppies are ideal shaping candidates because nothing is imposed on them physically, but their sessions need to be very short and their criteria very small.

Brachycephalic dogs overheat and struggle to breathe while excited. Keep sessions short and low-arousal, work in a cool room, and stop at the first sign of noisy breathing.

Senior and arthritic dogs should not be shaped through repeated sitting, standing, jumping or turning. Choose nose targets, chin rests and paw targets, which cost the joints nothing and are the most useful husbandry behaviours anyway.

Fearful and anxious dogs often do better with shaping than with any other method, because nothing is done to them and they control every step. Keep the rate of reinforcement high so there is never ambiguity, and never move towards them to deliver food; toss it or put it down.

Dogs losing hearing need the marker moved to a visual signal before the hearing goes, not after.

What never to do

Do not correct, scold or say "no" during a shaping session. The whole method depends on the dog volunteering behaviour.

Do not use the marker as a no-reward marker or a way to end the session.

Do not shape in a room with another dog waiting, with food on the floor, or near anything the dog would rather be doing.

Do not shape a behaviour that involves the dog putting weight on a slippery floor, jumping onto furniture, or working near stairs.

Do not keep going because it is going well. Ending a good session early is what makes the next one good.

Do not switch to luring in the middle of a shaping session. If you need to lure, that is a different plan for a different day.

When to get help

Free shaping is a low-risk activity and most problems are solved by splitting criteria further. Get help from a qualified, reward-based trainer or a veterinary behaviourist in three situations.

A dog that shows fear of the clicker or of training itself: flinching, hiding, drooling, refusing food it normally likes, or leaving the room when you pick up the treats.

A dog that will not offer any behaviour at all after several patient sessions with a novel prop, which sometimes reflects a history of punishment and sometimes an underlying pain or medical problem.

Any behaviour problem involving aggression, guarding or genuine fear. Shaping is a tool that can help within a plan, but the plan itself needs professional assessment first, and a dog in pain is not a dog that can learn.

The finished product: a behaviour the dog chose, on a cue added at the end, that holds up when you are not looking.