This study generalizes the definition of the sentinel function, which was introduced by J.L. Lions to study identification problems, to a more insensitive kind of sentinel, which is applied to identify pollution terms of the polynomial form w. r. t. the real λ. The main idea is to reconstruct the sentinel function to be nth-order independent of the incomplete data. Contrary to the original definition, information about the pollution term with an error of order n + 1 is given by the sentinel of nth-order insensitivity.