My wife made an account for our dog, with an appropriate headshot as the profile photo, about 9 years ago. He posts regularly, and we tag him in photos.
My guess is that they only block fake accounts if they are manually reported by someone.
You would think this, but as a counter example Xbox (back in the 360 days) had a "report gamertag as offensive" (aka offensive username).
Users could pay $ to change their gamertag (~$10).
It was fairly common to see people publically asking "please report my name as offensive" as it was known that if you got enough reports, you would be "forced" to change your name for free. People who wanted to change their name anyway would solicit reports against themselves to avoid a $10 fee.
Clearly no human was reviewing these reports... and that was a paid service!
That's true. I know a lot of people with a lot of fake accounts for entirely legitimate reasons, as well as some trolls who abuse the ease with which they can be created. A good number of people in the first group maintain multiple sock puppets because their subject to frequent attacks by trolls who report accounts to under false pretences, knowing that FB's response is as likely to be wrong as right.
Really, what's needed is a decentralized approach where no central authority exists, only mutual reinforcements. I have found a marvellous solution for the next generation of social and informational networking but alas this comment field is too small to contain it.
My guess is that they only block fake accounts if they are manually reported by someone.