Attentional misguidance from contextual learning after target location changes in natural scenes

Abstract

Attentional orienting in complex visual environments is supported by statistical learning of regularities. For instance, visual search for a target is faster when a distractor layout is repeatedly encountered, illustrating that learned contextual invariances improve attentional guidance (contextual cueing). Although contextual learning is usually relatively efficient, relocating the target (within an otherwise unchanged layout) typically abolishes contextual cueing, while revealing only a slow recovery of learning. However, such a “lack-of-adaptation” was usually only shown with artificial displays with target/distractor letters. The current study in turn used more realistic natural scene images to determine whether a comparable cost would also be evident in real-life contexts. Two experiments compared initial contextual cueing and the subsequent updating after a change in displays that either presented artificial letters, or natural scenes as contexts. With letter displays, an initial cueing effect was found that was associated with non-explicit, incidental learning, which vanished after the change. Natural scene displays either revealed a rather large cueing effect that was related to explicit memory (Experiment 1), or cueing was less strong and based on incidental learning (Experiment 2), with the size of cueing and the explicitness of the memory representation depending on the variability of the presented scene images. However, these variable initial benefits in scene displays always led to a substantial reduction after the change, comparable to the pattern in letter displays. Together, these findings show that the “richness” of natural scene contexts does not facilitate flexible contextual updating.