The problem of closure for the traditional unstructured possible worlds model of attitudinal content is that it treats belief and other cognitive states as closed under entailment, despite apparent counterexamples showing that this is not a necessary property of such states. One solution to this problem, which has been proposed recently by several authors (Schaffer in: Hawthorne and Gendler (eds) Oxford studies in epistemology, Oxford University Press, Oxford, pp 235–271, 2005; Yalcin in Philos Phenomenol Res 97(1):23–47, 2018; Hoek in: Kindermann (ed) Unstructured content, Oxford University Press, Oxford, forthcoming), is to restrict closure in an unstructured setting by treating propositional attitudes as question-sensitive. Here I argue that this line of response is unsatisfying as it stands because the problem of closure is more general than is typically discussed. A version of the problem recurs for attitudes like wondering, entertaining, considering, and so on, which are directed at questions rather than propositions. For such questioning attitudes, the appeal to question-sensitivity is much less convincing as a solution to the problem of closure.