Learning about object locations in space usually involves the summation of information from different experiences of that space and requires various cognitive operations to make this possible. These processes are poorly understood and, in the extreme, may not occur--leading to mutual exclusivity of memories (Baguley, Lansdale, Lines, & Parkin, 2006). In this article, we investigate how the precision of location memory--evident in near-miss errors in recall--is related to different transformational processes in spatial cognition. Analyzing errors and latencies in a sequential comparative judgment task, 4 experiments show that the precision with which location is represented in memory is specifically degraded by a subset of transformations in which an object location encoded in reference to 1 anchor point is recalibrated in relation to another. We discuss the general implications of this finding for spatial learning and demonstrate that, rather than being a special case, exclusivity in memory is the extreme expression of a rational trade-off between the benefit of combining spatial information from more than 1 memory and the reduced precision that follows from the transformations required.