That test is pretty misleading, it's similar to the famous French study which used white wine dyed red and counted whether people used "red" words or "white" words. One obvious conclusion consistent with other studies is that people believe what they're told.

I hesitate to call this type of uncontrolled testing junk science, but it definitely doesn't demonstrate what it claims to, imho. Super popular with the "well actually" crowd, however.

Addendum: I have personally, on several occasions, failed to replicate this result in a study where the subjects are not tainted with false knowledge. Blindfolded drinkers who are given a glass of red and told nothing about it usually know it's red.