It's absolutely an interesting paper, but just to note: Their finding is not that the spike activity is a language. They just propose it might be a language and then derive a number of statistics by treating it like a language for the sake of the argument.

... at least that was my understanding from the abstract and introduction.