The mirror test is miserable. A large portion of HUMANITY fails the mirror test, uncoached - turns out it's cultural!
It is one of those tests where a positive result means something (but not much) and a negative result means nothing, and yet people pretend that a negative result has some meaning.
It rests on several assumptions that have nothing to do with intelligence:
a) That an animal can recognize itself in a mirror. (This is the only part related to intelligence)
b) The animal cares about a dot on its head.
c) The animal wants to remove the dot.
d) The animal thinks it should remove the dot.
e) The animal thinks its okay to remove the dot.
f) A host of other unfounded assumptions having nothing to do with intelligence
A much better use of the mirror test is the COACHED mirror test - can an animal be trained to recognize itself in the mirror and point out a dot somewhere on its body, that it can't otherwise see? This is the only way to consistently get humans to pass the mirror test, after all. And, unsurprisingly enough, a great number of animals pass the test once they realize what it is we want them to do.
I'm sorry, the "mirror test", as stands, is about as meaningful as the rat-based "heroin test" that proved people would destroy their lives due to heroin given the chance. Or the LSD tests where they stuck someone in a bare room in a hostile environment and determined the large number of negative responses meant all responses always were negative. Or the numerous studies that assumed that every person in the world is exactly like a USA psychology undergrad.
It's bullshit, and people using bullshit extrapolations from crappy tests and studies infuriates me.
All that said:
Dolphin's seem to be pretty smart. It's a shame there's no concerted effort to really find out how smart or in what way, that isn't plagued by morons pretending to know what they are doing. It's like the founding of psychology all over again up in here...