top of page
Search

Training or Testing? Why the Difference Matters

Nov 9
10 min read

There is a phrase I hear quite often in dog training: “Let’s see if he’ll do it.”


Sometimes that is exactly the right thing to do. At some point, after all, we do need to find out whether the behaviour we have been working on actually holds up when the support starts to disappear. The problem is that owners often move into testing without quite realising they have done it. They think they are still training, when in reality they have changed the question from “Can I help my dog learn this?” to “Can my dog now do this without the help?”



That distinction matters, because it changes how we should think about both success and failure.


I tend to think of it a little like learning to ride a bike. In the early stages, there is quite a lot of support. Someone may be holding the saddle. You may have stabilisers. You are probably somewhere flat, quiet and fairly forgiving. As confidence grows, some of that support starts to disappear. The stabilisers come off. The person holding the bike lets go for longer. You ride a little further, then somewhere slightly more difficult, then perhaps somewhere with more going on around you.


Eventually, of course, you ride on your own.


That unsupported ride is useful because it tells you whether the skill is really there. What would be less useful would be removing all the support before the rider was ready and then drawing grand conclusions from the fact that they fell off.


We do something similar with dogs more often than we perhaps realise.


Training is rehearsal


When we are training a behaviour, our job is not simply to ask for it repeatedly and hope that repetition alone will make it stronger. We are trying to create conditions in which the dog understands what we want and then gradually learns to do the same thing under more difficult circumstances.


That means making the exercise harder over time, but in a measured way.


Take place training. You might begin by helping the dog understand that getting onto the bed and remaining there is what earns reinforcement. Once that is clear, you add a little duration. Then you begin to move away. Later, you walk around the room. At some point, you may introduce a visitor, open a cupboard, carry something across the room or ask the dog to remain there while dinner is being prepared.



The important point is that the behaviour is not suddenly being asked to survive everything at once. You are building it in layers.


Some trainers would describe part of that process as proofing, and that is perfectly reasonable terminology. Personally, I think owners often understand it more clearly if we keep the distinction simpler. Training includes teaching the behaviour and then strengthening it by gradually reducing support and introducing real-world difficulty. Testing is the point where we ask whether that work now stands up with much less help from us.


You could call it rehearsal and performance if you prefer. The labels matter less than being clear about which one you are doing.


The temptation to test too soon


Owners understandably want to know whether the training is working. If your dog has been recalling nicely on a long line for the last couple of weeks, there is a strong temptation to unclip it and find out whether the recall “really works”. If the dog can hold place for several minutes while you move around the kitchen, you may be curious to see whether the same thing happens when friends come over. If a reactive dog has been calm at fifteen metres from another dog, you may decide to move closer and find out whether they can handle it.


Sometimes that works perfectly well. Sometimes it does not.


The issue is not that failure itself is terrible. Dogs are not machines, and training is never a completely smooth line from not knowing something to performing it perfectly. The issue is what we conclude from failure, particularly when we have quietly changed the difficulty more than we realise.


A dog that recalls reliably on a long line in a quiet field has not necessarily learned exactly the same thing as a dog recalling away from wildlife at fifty metres. The cue may be the same and the physical movement may look the same, but the task is very different.


Likewise, a dog that can remain on place while you stand a few feet away has not automatically learned to remain there while three visitors move around the room and another dog wanders past.


This is where people often slip into phrases such as, “He knows it, he just chose not to do it,” or, “She was being stubborn today.”


That may sometimes be true, but it is worth checking whether we have actually trained the behaviour at the level we just demanded before deciding that the failure tells us something about the dog’s attitude.


The Three Ds are useful for a reason


Distance, duration and distraction are useful because they give us a simple way to see how much an exercise has changed.


Imagine your dog can remain on place for two minutes while you stand nearby in a quiet room. Then one evening you ask for ten minutes while you are in the kitchen and friends are moving around the house. You have increased duration, distance and distraction at the same time.


That is no longer a small variation on the original exercise. It is a much harder version.


The same thing happens with recall. Perhaps the dog reliably comes back from ten metres when not much is happening. Then one day we ask for fifty metres while another dog is running nearby. Again, the behaviour has not changed, but the difficulty has.


This is why I tend to prefer small increases. If you add more distraction, perhaps keep the distance shorter. If you increase the duration, reduce some of the other difficulty around it. If you move into a more challenging environment, give the dog a little more support initially.


There is rarely any great advantage in making three things harder at once.


Good training is not about finding the biggest possible jump the dog can cope with. It is about moving the difficulty forward far enough to create learning without making the exercise unnecessarily fragile.


Good trainers are generally conservative


That may not sound very exciting, but sensible dog training usually involves rather less gambling than people imagine.


If I am working on recall around significant distractions, I would rather keep the long line on slightly longer than remove it slightly too early. If I am working with a reactive dog, I would rather stay a little further from the trigger and get several good repetitions than push closer simply to find out where the dog’s limit is.


Likewise, if a dog has a history of chasing livestock, I am not going to release that dog in a field full of sheep because the last three training sessions happened to go well. There is nothing to be gained from turning a training decision into a wager.



People sometimes worry that leaving a safety net in place means the dog has not really learned the behaviour. I do not think that follows at all. If the dog is responding correctly and the line is loose, the line may not be helping in any practical sense. It is simply there because, if our judgement turns out to be wrong, we still have a way of preventing the mistake from becoming a much bigger one.


The cost of failure matters too. If I test a place command and the dog steps off the bed a few seconds earlier than expected, very little has happened. I have learned something about the limits of the behaviour and can adjust the next repetition.


If I make the same error of judgement with an off-lead recall beside a road, livestock, wildlife or another dog, the consequences may be very different. The level of caution should reflect what is actually at stake.


Failure can be useful


None of this means that we should engineer training sessions in which the dog can never possibly get anything wrong. If we make everything so easy that success is guaranteed, we may never learn where the behaviour is beginning to weaken.


At some point, we need to ask a little more.


The key is that the mistake should give us useful information rather than simply create a bad rehearsal.


If I ask a dog to remain on place while somebody walks past and the dog gets up, that tells me something. Perhaps movement at that distance is still too distracting. Perhaps I added too much duration first. Perhaps the behaviour is solid in the kitchen but not yet in the living room.


That is all useful information because I can change the next repetition accordingly.


What I would not want to do is repeat exactly the same failed exercise six more times and hope that, by sheer persistence, the dog suddenly starts getting it right. At that point, the dog is rehearsing the behaviour I do not want.


A dog that repeatedly ignores recall is practising not coming back. A dog that repeatedly leaves place is getting repetitions of leaving place. A reactive dog repeatedly taken too close to another dog is getting more practice at reacting.


Rehearsal works both ways.


Recall is probably the clearest example


Recall is one of the easiest places to see the difference between training and testing because owners quite reasonably want to know when they can finally trust the dog off lead.


The long line often starts to feel like something the dog needs to “graduate” from, as though keeping it on somehow means the training is incomplete. I do not see it that way.


The long line is simply a tool that allows you to practise the behaviour while retaining control over the outcome. Over time, you can use more distance, move into more interesting environments and introduce stronger distractions. The line can start to trail rather than being actively held. Eventually, there may be long stretches of a session where it is doing absolutely nothing.


By that point, removing it should feel like a very small change.



If unclipping the line suddenly turns the session into a leap of faith, I would question whether the dog was ready to lose it. The transition should be boring, because the dog has already shown you repeatedly that the support is no longer doing much.


That principle applies more broadly too. We should not remove support because we are impatient to prove it is no longer necessary. We should remove it because the training suggests it has already become largely redundant.


Reactivity makes the same point in a different way


With reactive dogs, there can be a temptation to keep moving closer to whatever triggers the reaction because progress becomes associated with distance.


If the dog is comfortable at fifteen metres, we try twelve. If twelve goes well, perhaps we try eight. Eventually the whole exercise can start to feel like a race towards the trigger.


That is not really the objective.


Distance is one of the things helping the dog remain calm enough to think, respond and disengage. Reducing that distance gradually may absolutely be part of the training, but there is no prize for getting closer quickly.



If a dog can observe another dog calmly at twelve metres, remain engaged with the handler and recover easily, I would rather build a strong history of success there before deciding that seven metres must automatically be better.


We are trying to change the dog’s response, not discover the precise distance at which the dog explodes.


That distinction is important because one approach is about building capability while the other can easily become experimentation for the sake of seeing what happens.


So when should we test?


There is no magic number of repetitions that tells you a dog is ready, and different behaviours carry very different levels of risk. What I am generally looking for is consistency rather than occasional success.


Has the dog performed the behaviour reliably over a reasonable period of time? Can they do it in more than one environment? Have you already added meaningful levels of distraction? Are they responding to the first cue, or are you still having to repeat yourself and negotiate the outcome? Has support been reduced gradually, or are you planning to remove most of it in one go?


The final question is probably the most important: what happens if your judgement is wrong?


If the dog simply steps off place and you calmly reset them, you can afford to explore the edges a little more. If getting it wrong means the dog disappears after a deer, charges towards another dog or ends up somewhere dangerous, then I would want considerably more evidence before taking away the control I still have.


Testing should tell you something useful. It should not require you to cross your fingers.


And if the dog fails?


Before deciding the dog has been stubborn, disobedient or deliberately difficult, I think it is worth looking at the training first.


That does not mean dogs never make choices we would rather they did not make. Of course they do. Nor does it mean that accountability has no place in training. It does, but accountability only really makes sense when the dog genuinely understands the task under the conditions in which we are asking for it.


So look at what changed.


Was the distraction significantly stronger? Did you increase the distance? Did you ask for much more duration? Did you move into a new environment? Was the dog already highly aroused before the cue was given? Was the behaviour genuinely as well established as you thought?


Sometimes you will conclude that the dog understood perfectly well and made a poor choice anyway. That happens.


It is still worth earning the right to reach that conclusion rather than making it the default explanation whenever something falls apart.


Training before the final


The simplest way I can think about all of this is to separate rehearsal from performance.


Training is the rehearsal. You teach the behaviour, strengthen it, change the environment, add distraction and gradually remove the support that helped the dog succeed in the first place.


Testing is the point where you find out whether that work now stands up on its own.

We should test our dogs. Behaviours eventually need to work outside a structured training session or they are not much use to us. The mistake is not testing; it is turning the test into an unnecessary gamble.


Keep the support for as long as it is useful. Make the changes gradually. Be particularly conservative where the consequence of a mistake matters.


Then, when you do eventually think, “Let’s see if he’ll do it,” you should already have a pretty good idea of what is going to happen.

 
 
 

Comments


©2025 by Much Ado About Dogs. Proudly created with Wix.com

bottom of page