The Turing test is the famous method for determining whether computers can think--can the computer's conversation with a person be so good it can't be distinguished from that of a human?
There was a piece I read today discussing other tests for distinguishing computers and humans. But I want to discuss going the other way--distinguishing humans. I'd suggest the only way to distinguish humans from other entities, whether they be computers or chimpanzees, is the genetic one. By that I mean that a human is born of another human and contains DNA from one or more humans.
When you expand your mental image of "human" from a mature adult to include infants and the mentally and physically challenge I don't think there's a reliable performance test. The reverse Turing test doesn't work--many humans cannot converse, a few have no language at all. So I think, rather than performance, the only test of humanity is the genetic history.