Yes, dogs are the worst possible place to look for ‘anthropomorphic’ behaviour — they've had thirty or forty thousand years of evolutionary pressure to please humans.

Well, depending on whether you are looking for naturally occurring anthropomorphic behavior, they are either the best or worst place to look. :)

That is, they are probably good evidence for some emotions in animals, but fairly useless in their current incarnation as to whether it was naturally occurring. Interestingly, it's entirely possible (and likely, IMO) that they exhibited some of this behavior originally due to their social nature, and this is what causes humans to start the process of domestication in the first place (not that it was likely a decision like that).

What's really interesting is how quickly changes, both behavioral and physical happen during domestication (and how quickly domestication works, possibly on the order of a decade or so)[1].

1: http://www.radiolab.org/story/91696-new-nice/

‘Behaviour’ was the wrong word for me to use. I should have said, dogs are the worst place to look for human-like internal mental state. If your dog looks contrite when they've done something wrong, do they feel contrite, or have they just mechanically evolved to make you think they do?

They don't have to have done something wrong.

Years ago I had a lurcher, raised from a pup. One day I noticed that if you spoke to him in a stern manner - even if he had done absolutely nothing wrong he would hang his head low and look up at you, and put his tail between his legs as if he was guilty of something bad and wanted to look contrite.

It seems to me that it may be a long ago learned, now instinctive, behaviour to react that way to a particular tone of human voice.

It might have a simpler explanation. Maybe what you are seeing isn't guilt, but submission, and you are mistaking it for guilt. When spoken to a certain way, submission is the correct social response. This is true in people as well. If you have a work superior approach you in an angry manner, your best option is likely not to respond angrily or indifferently.

A good question, and very complicated. For example, what if dogs have always been capable of that emotion, but expressed it differently? What if the evolved response is not the emotion, but to match a specific behavior to the emotion? What if the evolved response is not attached to the emotion and just for our benefit, but the emotion does exist in other situations? It would be so much easier if we could accurately communicate with our dogs.

On that note, I'll leave you with this, which is wonderful: https://www.reddit.com/r/HFY/comments/4rkdit/text_we_didnt_s...