Last year I attended the Tinbergen Lecture in Leiden, where prof Frans de Waal told us about his research on animal behaviour. He is well known for his work with chimps, but he also reported how they were finally able to demonstrate Mirror Self Recognition behaviour with Elephants. Previous attemps had failed, apparently because testers had been using small mirrors, which would not have large enough for an elephant to even see a significant portion of itself.
Now the same test has been performed with birds - magpies to be precise - not by de Waal, but he comments on it at PloS: doi:10.1371/journal.pbio.0060201.
He explains the significance of the test: It's not about intelligence or about the concept of self-identity as such, but a side-effect of the species developing stronger social skills. Primates, elephants and magpies all have complex social lives and that requires them to be be able to distinguish each other and put oneself in another's position to understand them.
And who knows what they think about each other...
And that's what Dave Kellet thought as well.