I blindfolded seven Slate staffers and conducted a repeat-trial, forced-choice discrimination test. For each trial, the subjects compared the taste of pure, filtered water from the office cooler with that of a test liquid, which may or may not have contained eight drops from a vial of McCormick's NEON! food dye. (Test liquids were either unadulterated or colored pink, green, or blue.) I asked subjects to state whether they believed the test liquid to be colored or uncolored and then to rate how confident they were on a scale of one to five.
Given the results from 84 total trials, subjects were able to identify the presence or absence of artificial color 56 percent of the time. Correct answers were associated with average confidence ratings of 2.7; incorrect answers corresponded to an average confidence of 2.3. Subjects detected the presence of color in 79 percent of the cups containing artificial pink color, as opposed 57 percent each for green and blue.
The results of the taste test suggest that there may be a very slight but detectable difference in taste or mouth-feel between pure and artificially colored water. It's also possible that this difference is more pronounced for the pink dye. Would the subtle "taste" of artificial color be detectable in a slice of rich, chocolate cake? I'm guessing no.