Humans are REALLY bad at reading what’s happening with dogs. We ignore really bad stuff and we stop, even punish, perfectly normal behavior. This is true even when you’re really experienced; I have spent thousands of hours studying this stuff and I am more convinced that I am an idiot when it comes to dog behavior than I was when I began. Maybe when I’m sixty I’ll start interfering, but right now I know perfectly well how dumb I am.
One of the reasons we screw up interactions has to do with the fact that, just like we humans do, dogs have communications that have beginnings, middles, and ends. The beginning is the series of behaviors that initiate the interaction, the middle is the interaction itself, and the end is when the dogs resolve or end the interaction and move back away from each other.
That entire cycle is VERY important. If it is not completed, the next time the dog or dogs tries to interact the “transaction” won’t go as smoothly. Some of the social lubrication will have been lost.
If interactions are routinely truncated, two bad things happen: First, the dogs involved don’t get to finish the conversation, so they get out of practice in how to finish interactions. This is a lot more dangerous than it sounds–every dog interaction is a finely honed and subtle meeting of two animals perfectly prepared to kill each other. Predator-to-predator transactions are not exactly natural, and dogs have evolved an incredibly complex series of behaviors to keep things from escalating into an attempt to physically harm. If they are bad at those behaviors–if instead of suave and smooth talkers they’ve become awkward and tend to say the wrong thing–they are in genuine danger of falling from normal transaction into a situation where one or the other of them will make a move toward a killing attempt.
New puppy arrives and is cute and wonderful. For a few days or weeks she toddles around and falls over adorably and snoozes everywhere and plays her funny little puppy games and everyone, including all the adult dogs in the house, smiles indulgently and allows her many liberties that they would never allow an adult dog.
This is normal and good; it’s how the pack bonds with and learns to protect the puppies that come into it.
But then at some point, say at twelve or sixteen weeks, even earlier for the quick maturers, the little soft fuzzy schnookums-wookums becomes a growing dog, and her little games start to involve using her teeth in a real and deliberate way. And instead of bumbling into the adult dogs’ heads and falling over, she’s lying in wait and then barreling over and jumping on their heads.
The adult dogs decide they’ve had enough, and they begin to punish her for this rude behavior. If she jumps on them they roar, they knock her with their mouths, they send her ki-yi-yi-ing into the next room. When she has play interactions with them they don’t hold back anymore; they pin her and knock her over and she yelps and rolls away.
The human says “Oh no! Poor Gladys! They’re being rough with her!” and they begin to supervise the play. Every time the adult dogs get “rough” they are stopped or disciplined. If they continue to “victimize” the puppy they are totally separated; she plays alone and they play alone.
THIS IS SUCH A BAD IDEA.
Puppies learn from adult dogs. A vital and absolutely incontrovertible role of a healthy adult dog is to teach the puppy how to be a good and polite dog. The adult teaches–yes, by physical punishment, though that punishment is not cruel–how to interact with other dogs, how to live in a pack, how to ask permission, how to back off, etc. If you stop that from happening, not only does the puppy grow up with SERIOUS issues that will hurt her chances of being a normal dog who can get along with other dogs, you build resentment between the two dogs. If the adult dog is never allowed to complete a lesson, he will try harder and sooner the next time. If he’s stopped again and again, pretty soon he will decide that the only way to deal with this is to remove the puppy from the picture entirely.