Figure 3.

Schematic of “looking at nothing” behavior, whereby participants reinstate encoding-related eye movements during retrieval in the absence of visual input, across three task types. Row 1 depicts tasks in which participants are required to remember the relative locations of presented objects (left). During maintenance (whereby a representation is held in an active state in memory) or retrieval (right), participants’ eye movements reinstate the locations and spatial relations among encoded objects, e.g., [36,43]. Row 2 depicts tasks in which participants are required to remember a complex scene that was presented either visually or auditorily (left). During retrieval (right), participants’ eye movements return to regions that were inspected during encoding, e.g., [48,56]. Row 3 depicts tasks in which participants are required to answer questions or make judgments about previously presented items. During retrieval (right), participants look in the region of the scene that previously contained the target item, even when successful task performance does not require the retrieval of the previously observed spatial locations [42,45]. For within-item effects, see [40]; for words, see [46]; such effects persist even after a week-long delay [61].