Abstract
General scene knowledge (our "scene grammar") and episodic memory play an important role in guiding search and navigation. In two virtual reality experiments we investigated the influence of scene syntax on participants' interactions with objects and the detail of spatial memory representations formed during these interactions. In Experiment 1 (N = 9), participants had to arrange virtual objects consecutively in sixteen rooms. In half of the rooms participants were instructed to arrange objects in a syntactically meaningful way (e. g. placing a pot onto a stove), whereas in the other rooms the objects had to be arranged chaotically. In a subsequent, unannounced free recall, task location memory was assessed by asking participants to rebuild these rooms. Explicit location memory was better for syntactically consistent compared to inconsistently placed objects. The instruction to place objects inconsistently lead to a longer interaction time with objects – measured as object grabbing time. This was especially prominent for small (local) objects. Large (global) objects were placed earlier in the trial than local objects. In Experiment 2 (N = 8), participants had to build eight rooms in the same fashion as in Experiment 1, yet this time a surprise search task followed. Participants either searched for objects within the rooms they had built, or within rooms arranged by participants from Experiment 1. Search was speeded for consistently placed objects, especially for objects placed by participants themselves. Additionally, more distance was covered to find objects in inconsistent rooms, most prominently when arranged by somebody else. Together, these results suggest that contextual violations, even when self-inflicted, lead to differential object-interaction behavior, as well as a decrease in memory performance. Real-world search profits from a meaningful arrangement of objects both in self-generated as well as unknown contexts by reducing the effort to move the body to find the target.
Meeting abstract presented at VSS 2017