It would indeed not make any sense to have a NEGATIVE sample that is the same as the positive one.
Well,
actually it could occur (with low probability) if another word appears
elsewhere in the corpus with the same context, for instance here if we
had something like "... use teaspoon when... ". But it was NOT the case
in the quiz question.