Abstract
Spatial descriptions can specify the location of an object (located object) by spatially relating it to a reference object that is selected from surrounding objects. Previous research with currently viewed scenes has shown a preference for selecting reference objects on the basis of spatial features such as proximity and alignment. However, we often talk about scenes that are not in current view. Two experiments demonstrated that the spatial features defined by the perspective from which a scene was learned influenced reference object selection in later descriptions of the scene from memory. Specifically, participants learned an array of objects, and judgments of relative direction verified the perspective with which the object locations were encoded. Participants then moved to a new location from which the scene was no longer visible, and completed a reference object selection task by describing the scene from memory, using the sentence frame, “The is near the ____”. Results of JRD task showed that object locations were represented with respect to a specific intrinsic reference direction, as determined in Experiment 1 by learning direction and in Experiment 2 by the dimensions of a rectangular table on which the objects were placed. For the reference object selection task, participants were significantly more likely to choose reference objects along the intrinsic reference direction that corresponded to the way that the objects were represented as reflected in the JRD task. These findings were also supported at the individual level as a correlation between the strength of the organizational bias with respect to particular axes in memory and the likelihood of selecting reference objects along these axes in the spatial language task. These findings extend the prioritization of spatial features for selecting reference objects to memory representations, and are consistent with claims of correspondence between memory and linguistic representations.