This BBC article by Jonathan Amos tells us that new evidence appears to support the ‘dogs as scavengers’ hypothesis (i.e., that the human-dog relationship evolved through a process of assimilation; dogs hung around the food generated by human settlements, so hung around human settlements and moved from a commensal to a mutualistic relationship with humans by gradually involving themselves in human existence. The most docile and cooperative would be more successful and may therefore have benefited from additional human benefaction.
The ‘opposing’ hypothesis is that we humans were the agents in the interaction; some humans adopted wolves (as puppies, it is often suggested, which allowed them to be more easily tamed) and began to use them for hunting or possibly security, selecting for traits that served the humans best.
If I had to pick one, I would go for the scavenger hypothesis; not just because of this recent evidence, which is interesting, but also because in some cultures dogs still are, essentially, commensals. They hang around villages, not owned and not wanted, and scavenge. Genetically, these dogs are as closely related to wolves as they are domestic dogs; the researchers think this indicates that they have been more ‘naturally selected’ than other domestic dogs, with less human input into their evolutionary development. (This suggests, though, that human involvement in an association automatically makes the result ‘unnatural’, a line of thought I don’t tend to agree with – more on that another time).
The human preferences in question have led, it is thought, to ‘neoteny’ in dogs. This means they appear to have evolved into permanent wolf-pups, with many youthful characteristics, (that would in wolves be ‘grown out of’) preserved into adulthood. Examples are obedience, docility, flatter faces and floppy ears, though this obviously varies between breeds and individuals.
Personally, I don’t think the two theories are necessarily mutually-exclusive; if I remember rightly, there is also evidence to suggest that wolves were domesticated almost concurrently in multiple cultures; this being the case, there’s no reason why different processes could have prevailed in different areas. There’s also nothing to say that humans chose to feed, raise and employ certain individuals from the local scavenger pack, which could have accelerated a symbiotic process.
Acknowledging, then, that the dog is not a chosen domesticate or companion everywhere, for those cultures (like mine) which have an enduring and generally positive relationship with them, I prefer to think in terms of ‘co-evolution’: we didn’t pin them down and force them to associate with us and they didn’t coerce us into it…. I like to think we chose one another.