Intraub posits the existence of two separate processes in scene memory: one in which we automatically extrapolate the visual information in a scene beyond its boundaries (scene construction), and one in which we normalize our memories to either a schema or an average (normalization). She claims that scene construction will lead to transformations exclusively in the direction of boundary extension (BE), while normalization will produce bidirectional transformations of both BE and the opposite effect of boundary contraction (BC). Thus, because we observed both BE and BC in our study [1], our paradigm must be tapping into additional cognitive processes than just scene construction. However, our paper [1] questions this premise — if BE and BC are equally common using large, representative stimulus sets in the same tasks used previously [2,3], then perhaps prior studies primarily found unidirectional BE effects due to limited stimulus sampling, and there was no privileged link between boundary extension and scene construction in memory to begin with.

Intraub假定场景内存中存在两个独立的过程：一个过程中，我们自动将场景中的视觉信息外推到场景的边界之外（场景构造），另一个过程中，我们将内存归一化为模式或平均值（归一化） 。她声称场景构建将仅导致边界扩展（BE）方向的变换，而规范化将产生BE的双向变换和边界收缩（BC）的相反作用。因此，由于我们在研究中同时观察到了BE和BC，因此我们的范例必定会利用除场景构建之外的其他认知过程。但是，我们的论文[1]对这一前提提出了质疑-如果在以前使用的相同任务中使用大型，有代表性的刺激集，那么BE和BC是否同样普遍，[2,3]