DescriptionThe only a priori constraint on the kinds of entities that can be the members of a to-be-counted collection is that they can be treated as if they are separate from each other. Otherwise, the result would not be a natural number. But, there is nothing about counting itself that dictates the nature of the entities for a given count. As Gelman and Gallistel (1978) noted, any items that can be treated as discrete can be collected together for a count. The entities need not be objects; they can be spaces between telephone poles, the number of great Presidents, or even the number of good ideas one had in a given time period. This degree of permissiveness introduces a fundamental question: What sets the boundary conditions on what is actually contained in a to-be-counted collection? The studies presented here demonstrate that people use a variety of constraints. These constraints include some that are highly conceptual, some that involve simple perceptual groupings, and some that are invoked by verbal labels or conversational pragmatics. The common characteristic of these options is that they provide contexts that place items at a common level of interpretation or perception. It is shown that the same rectangle can serve different framing effects depending on whether it is dubbed a mirror, a window, or a picture frame. It is also shown that the numerosity of a given number of circles is interpreted differently, depending on whether the circles are arranged concentrically, whether some of the circles are positioned inside others, or whether none of the circles are so bounded.