The one sentence version
Dogs repeat behaviour that works for them and abandon behaviour that does not. That is nearly the whole of learning theory, and everything below is detail about what works means and how quickly a dog can figure it out.
The important consequence is that your dog is always learning, whether or not you meant to teach anything. A dog who barks at the door and gets it opened has learned something. A dog who jumps up and gets pushed away with attention has learned something. Training is not a separate activity from life; it is just the part where you decide what gets reinforced.
That reframing is why experienced trainers spend so much time on management. Preventing a behaviour from being practised is often faster than trying to un-teach it after it has been rehearsed a hundred times.
The four kinds of consequence, without the jargon
Learning science splits consequences into four kinds, along two axes: whether something is added or removed, and whether the behaviour goes up or down afterwards. The textbook labels are unhelpfully technical, so here they are in plain terms.
| What happens | Effect on the behaviour | Everyday example |
|---|---|---|
| Something good is added | Happens more | Sit, get a treat |
| Something is removed the moment the dog complies | Happens more | Leash pressure stops when the dog steps toward you |
| Something the dog dislikes is added | Happens less | A startling noise when they raid the counter |
| Something good is taken away | Happens less | Play stops the moment teeth touch skin |
Most good training lives in the first and fourth rows, with a fair amount of the second in leash and long-line work, where the dog learns that following the pressure makes it stop. The distinction that matters in practice is not which box a consequence belongs in, but whether the dog can tell what caused what.
That is worth stressing because arguments about method are often really arguments about clarity. A consequence the dog cannot connect to their own behaviour teaches nothing regardless of which box it came from.
Timing is the whole game
A dog connects a consequence with whatever they were doing in roughly the last second. Not the last ten seconds, and certainly not the thing they did before you got home.
That is why a reward delivered four seconds after a sit does not reinforce the sit. It reinforces whatever happened four seconds later, which is usually standing up and looking at your pocket. Owners then conclude their dog does not hold a sit, when what actually happened is that they trained standing up.
- Mark the moment. A word like yes, or a clicker, freezes the exact instant so the reward can arrive a second later.
- The marker has to be trained first: mark, then feed, thirty or forty times, until the word alone gets a head turn.
- Once marked, the delivery of the reward is less urgent, which makes everything easier.
- Late marking is worse than no marking, because it labels the wrong moment precisely.
This is also the argument for keeping rewards physically accessible. A treat you have to dig out of a coat pocket arrives too late to mean anything, which is why a pouch that opens one-handed is a genuine training tool rather than an accessory.
Value is decided by the dog, not by you
A reward is only a reward if the dog wants it more than the alternative available at that moment. Dry kibble is a fine reward in a kitchen and worthless next to a squirrel.
The practical approach is to build a ranked list for your own dog and spend the good stuff where it is needed. Most owners have this backwards: high-value food in easy situations, and nothing but praise in the hard ones.
| Tier | Typical | Use for |
|---|---|---|
| Everyday | Kibble, dry biscuit | Known behaviours at home |
| Good | Soft training treats | Practising in mildly distracting places |
| High | Chicken, cheese, liver | New behaviours, and real distractions |
| Jackpot | A long-lasting chew, a favourite toy | Breakthroughs and hard recalls |
| Non-food | A game of tug, a thrown ball | High-drive dogs who work for movement |
The bottom row matters more than people realise. Plenty of working-line dogs are indifferent to food and will work all day for a ball. Insisting on treats with those dogs is a slow way to conclude they are difficult to train.
Criteria: raise one thing at a time
Every behaviour has several dimensions, and a dog can usually only handle one of them getting harder at a time.
- Duration: how long they hold it.
- Distance: how far away you are.
- Distraction: what else is happening.
- Environment: where you are.
- Handler: who is asking.
A dog who holds a five-second sit in the kitchen is not failing when they break a five-second sit in a park. Two dimensions changed at once, and the correct response is to make the sit shorter in the new place rather than to repeat the cue more firmly.
When something falls apart, the general rule is to drop the dimension you last raised, get three or four easy repetitions to rebuild confidence, and then raise it again in a smaller step. Trainers call this splitting, and it is the difference between steady progress and a dog who becomes hesitant.
Why behaviours come back after they seem gone
A behaviour that stops being reinforced fades, but rarely in a straight line, and the shape of that curve explains a lot of owner frustration.
When a previously rewarded behaviour suddenly stops paying, most dogs try harder before giving up. A dog who used to get attention for barking will bark more, and louder, for a while. That spike is a normal part of the process and it is exactly the moment most people give in, which teaches the dog that persistence works.
It also explains why intermittent reinforcement makes behaviour so durable. A behaviour that pays occasionally and unpredictably is far harder to extinguish than one that paid every time, which is why the one time in twenty that begging at the table works keeps it alive indefinitely.
The practical version: decide what the rule is, make sure everyone in the household applies it, and expect it to get briefly worse before it improves.
What dogs are not doing
A few common interpretations get in the way of training, and dropping them makes everything easier.
| Common belief | What is more likely happening |
|---|---|
| The dog is being spiteful | The behaviour was reinforced by something, often attention |
| The dog knows they did wrong | They are reading your body language right now, not recalling this morning |
| The dog is challenging my authority | The behaviour works, so it is being repeated |
| The dog is being difficult on purpose | The criteria moved faster than they could follow |
| They understand the word, they are ignoring it | The cue was learned in one context and has not generalised |
The last row accounts for an enormous share of everyday frustration. A dog who sits perfectly in the kitchen and ignores the same word at the park has not learned the word in the sense we mean it. They have learned a kitchen behaviour, and the park version has to be taught separately.
Putting it into a session
A good session is short, frequent, and ends on purpose rather than on failure.
- Five to ten minutes, once or twice a day. Frequency beats duration by a wide margin.
- Decide the one criterion you are working on before you start.
- Get an easy repetition first so the dog starts with a win.
- Mark and reward on time, every time, while a behaviour is new.
- If two repetitions in a row fail, make it easier rather than repeating it.
- Stop while the dog is still keen, not when they are tired of it.
That last point is the one most people ignore, and it is the one that determines whether your dog looks forward to training or drifts off halfway through. A dog who wanted one more repetition when you stopped is a dog who arrives eager tomorrow.
Timing beats everything, and timing needs a reward you can reach
A pouch that opens one-handed, and a chew worth working for.
INVIROX Dog Treat Pouch
Magnetic one-handed opening, 5 x 5.5 x 3 inches, low profile when closed, 3 ways to wear it. Keeps the reward available in the half-second it matters.
INVIROX Bully Sticks (5 pack)
Single-ingredient grass-fed beef pizzle, 83% protein, fully digestible, no hormones or grains. The high-value reward most INVIROX owners move to.
Frequently asked questions
How do dogs actually learn?
By consequence. Behaviour that produces something the dog values happens more, and behaviour that produces nothing fades. Your dog is learning constantly, whether or not you intended to teach anything.
How quickly does a reward need to arrive?
Within about a second of the behaviour, or the dog connects it to whatever they were doing when it arrived. A marker word trained in advance lets you freeze the right moment and deliver the reward slightly later.
What is a marker word and do I need one?
A word like yes that means a reward is coming. Train it by saying it and immediately feeding, thirty or forty times, until the word alone gets a head turn. It makes every future session faster and more precise.
Why does my dog obey at home but not outside?
Because dogs learn cues in context. A sit trained in the kitchen is a kitchen behaviour to your dog. Each new environment and each new distraction has to be taught rather than assumed.
Why did the behaviour get worse when I started ignoring it?
That is a normal extinction burst. When a previously rewarded behaviour stops paying, most dogs try harder before giving up. Giving in during that spike teaches the dog that persistence works.
Is my dog being spiteful or dominant?
Almost never. The behaviour is being repeated because something reinforces it, usually attention. Looking for what pays off is far more productive than interpreting motive.
