I found this discussion on James Gorman's Scientific American blog on dogs and of special note is one particular comment that I've copied in its entirety below. Of course we should always post an asterisk besides these kinds of accounts since they can't be directly verified, but for me it holds the ring of truth and interestingly on such a prestigious site no one has in fact come forward to challenge it. It also is resonant with the "Covenant of the Tongue" worked out between Orcas and Aboriginal whalers in the Bay of Eden which gives it added credibility. At any rate, through training dogs to bite and hunt for criminals, and then coming to understand how everything dogs do is but a variant on this overarching theme of synchronized movements in the hunt; from mounting, to loving car rides, to howling, giving paw, playing fetch, this is the same conclusion I arrived at in formulating my theory. I predict in twenty years proto-dog as hunting companion to early humans will become the mainstream consensus of Dogdom in complete contravention to what is currently being taught, and yet, the attempt will be made to enfold it into the current paradigm, as for example we see with the dominance idea. Today consensus science says there's no such thing as a dominance impulse within the individual, but rather a dominance relationship that emerges from competitive interactions between individuals. This is just as wrong, if not more so, as the original idea of a dominant pack leader. One can't say that once in a while the sun revolves around the earth. The purpose of sociability is not companionship, that's a wonderful emergent benefit, the purpose is to improve the configuration, and the oldest configuration between living beings is between hunter and the hunted.
"When Europeans first settled Australia, dingoes were observed to be semi-domesticated camp followers of Aboriginal people which used to accompany them on hunts, and were helpful in locating prey."
"I have had exactly this relationship with a very unfriendly kelpie sheepdog (kelpies are believed to have been cross-bred from dingoes) – on a ‘personal’ level she had no interest in being friendly, she showed no affection or signs of being friendly at all (farmers discourage developing affectionate attachment to humans in sheepdogs they train because they think it distracts the dogs from their work), but she was a great working dog herding sheep, and whenever I went hunting kangaroos in the rapidly darkening twilight, she was right there, she knew what I was going out to do, and she would find the feeding kangaroo mobs for me, and then lead me to the kangaroo I had shot, even repeatedly looking back at me to make sure I was following her as she located the dead animal. That dog knew exactly what she was doing – it was a partnership, an equal partnership based on mutual benefit, and mutual affection had no part in it, it was strictly business."
"The contract was that she helped me find the prey and then locate the animal I had killed, and she got a portion of the kill – I got the hind legs and tail, she and her pups got the rest."
"So I go for Theory 2. The wolves chose humans as companions, not the other way round – but it wasn’t just a case of wait and get the scraps, it was teamwork in hunting. There was no affection permitted, if I had tried to pat her she would have taken my hand off, but it seemed to me there was companionship and some kind of mutual respect. I could be anthropomorphizing that last part, but whenever I picked up my rifle and walked out the door after sunset, she would look at me like “OK, here we go, I’m in on this, right?”, when the rest of the time she avoided any kind of eye contact with me or looking at my face."