Behavioral and computational research has proposed that a scene (e.g., a kitchen) can be represented by two independent, yet complementary descriptors: i) its spatial boundary (i.e., the external shape, size, and scope of the space the scene represents) and ii) its content (i.e., the internal elements encompassing objects, textures, colors, and materials). But how does the brain represent these descriptors? A central hypothesis is that one scene-selective cortical region (parahippocampal place area – PPA) represents both spatial boundary and content information, while a second region (retrosplenial complex – RSC) represents spatial boundary only. Such representation for a third scene-selective region (occipital place area – OPA) has never been tested. To test spatial boundary representation, we compared responses to images of intact rooms with images of these same rooms when their walls, floors, and ceilings had been fractured and rearranged, such that they no longer defined a coherent space. We found OPA, unlike PPA and RSC, responded similarly to both the intact and fractured rooms, suggesting OPA does not represent spatial boundary per se, but rather the local elements (i.e., walls, floors, ceilings) composing the space, independent of their spatial arrangement. To test content representation, we compared responses to images of furniture with non-furniture objects. We found OPA, like PPA, responded more to furniture than non-furniture objects. Interestingly, however, while both OPA and PPA represent content information, they do so differently; in another test, we found only OPA was sensitive to the number of pieces of furniture, suggesting OPA represents the local elements of scene content, while PPA represents the global aspects of scene content, independent of the number of objects present. Taken together, our results suggest OPA analyzes local scene elements – both in spatial boundary and content representation – while PPA and RSC represent global scene properties.