A lot of what I have read about dog training, involves creating commands or cues which the dog will then follow. Or teaching it to behave in a specific situation.

However, I would in theory, like to do the opposite. I.e., I want to train my dog to communicate it's wishes to me.

For example, if it has to go to the bathroom, I want it to do one action. If it just wants to walk for fun, I want it to do another action. If it wants to stop playing, I want it to do something else. The goal here is purely for it to communicate a desire to me, so that I may decide what to do, not for it to go ahead and act on that desire.

To quip, I'd like to train the dog to give me commands.

  • 6
    To quip, I'd like to train the dog to give me commands. To quip back: have you considered getting a cat? Commented Mar 13, 2014 at 10:35

1 Answer 1


If we were to ask the dogs, they would probably say that's what they do all the time!

So in a way it's up to us to learn what they are already saying. I've found that I can learn a lot just by observing dogs, not just our dog(s), when they play, walk, meet other people, etc. So their wishes are all there, just behind our eyes (or nose...).

If your dog wants to go out to the bathroom he will probably do something anyway, be more active, run to the door and back or whatever. So it would in principle be possible to add a cue that we find more clear (see below).

However we should remain a few things about dogs and how they learn:

  • dogs live in the moment without explicitly referring to actions in the past. Their level of "consciousness" is not the same as our;

  • as a consequence the learning is mostly simple associative learning. That means the way to teach them things is done by conditioning (classical or operant conditioning);

  • we should not adopt a point of view that is too anthropocentric and expect elaborate conscious processes.

The last point means: the dog won't elaborate a full scenario like we would when we want to obtain something from someone. To use the bathroom analogy again: the dog would first need to be fully aware of his need, then he would need to remember a similar situation in the past, and its full context (which is the difficult part), and he would then need to perform some specific action in order to make us let him go out.

Knowing the limitations can help us have realistic expectations and derive a proper way to obtain them.

In this case my answer would then be that yes, it is possible but primarily through conditioning. A few principles from the theory of learning, applied to dogs are useful in this case: reinforced behaviours increase in frequency, high probability behaviours tend to increase the frequency of low probability behaviours (Premack's principle).

Summarising my answer taking the example of a dog that wants to go out, but can't open/unlock the door (which is usually the case).

From the dog's point of view he's already telling you very clearly. But we need to add a cue which is more clear to us. For example ringing a bell or pulling on something, or simply barking in front of the door. The reinforcer here is you opening the door. That's the high probability behaviours, the dog want to go out and every time you open the door you reinforce him to do so. So you want to add a lower probability behaviour, ringing a bell next to the door. You can then train the dog to ring the bell, using the opening of the door as a reinforcer.

That way you used a mix of classical and operant conditioning: the stimulus is the feeling that the dog wants to pee and as he's house trained the response is that he wants to go out. The operant conditioning is that ringing the bell is a reinforced behaviour that is rewarded all the time. That's also a behaviour chain, were the first behaviour is ringing the bell, it is reinforced by another behaviour which is in turn reinforced by an actual reinforcer (primary reinforcer) in this case peeing in the garden.

As a side note this is a good opportunity to point out one of the advantage of positive reinforcement training (as opposed to punishment based training): we reward the dog for performing behaviours but we never inhibit behaviours with punishment. That means the dog always feels free to perform new behaviours (worst case they won't be reinforced and will extinct). The usual example is the 101 things to do with a box training session: you introduce a box to the dog and you reward him each time he does something. But the rule is: you reward only new behaviour. That's in a way teaching "creativity". The implication here is that if your dog is free to perform different kind of behaviour in front of you, most of the time you just have to choose what to reinforce. Imagine he wants his toy. If he start barking, you do nothing, then he'll try something else (say sit in front of you), you reinforce by giving the toy, and you achieve your goal.

This video ("Dog Training: How to Teach your Dog to ring a bell to be let outside") covers. I don't fully agree with the first part where he teaches the dog to ring the bell using a food reward. I think we can go directly with the actual reward, that will help the dog associate the bell with the door, and not ringing the bell as yet another trick.

Last thing: dogs can associate the present context with their action. For example they can probably associate that ringing the bell only works in your presence. However at first I would remove the bell when you're not there, to prevent the behaviour from being unrewarded.

  • How would you distinguish between ringing a bell to go outside for a pee, and licking his leash to go for a walk?
    – avi
    Commented Mar 13, 2014 at 10:09
  • I think simple associations would work. For example, go out for a walk only from the front door and let him go in the garden only from the back door. Then setup the leash in a way that he can take it and have him do that every time, before going for a walk.
    – Cedric H.
    Commented Mar 13, 2014 at 10:13
  • 1
    I must admit that I have no practical experience with that, but I think many owners report that their dog is doing such things. Usually it focusses on a very tiny behaviour and a lot of interpretation from the owner, but being consistent it is possible to strengthen the behaviour. Actually now I'm very curious about that, I'll start doing the bell thing with my dog, I'll let you know.
    – Cedric H.
    Commented Mar 13, 2014 at 10:15
  • 2
    But don't they pee every time they go out. There is no difference (from the dog's point of view) between fun walk and pee walk. Then again, observing my dog I can tell if he has a full bladder in need of relief, but I fail to notice if they are giving any signals for a fun walk. Either it is me, or my dogs, but some signals go unnoticed. Commented Mar 13, 2014 at 17:17
  • 2
    Good answer. Yes this can be taught and your dog is probably already doing some of it but you may not be good at reading it yet. The person that wrote "Dogs can sign, too" took it to a literal extreme. But there many good books and blogs on dog behavior that will help you. Just stick with folks like Temple Grandin and Patricia McConnell and stay away from people who focus on dominance theory and wolf ancestry.
    – Beth Lang
    Commented Mar 14, 2014 at 6:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.