Egocentric coding of space for incidentally learned attention: effects of scene context and task instructions

Visuospatial attention prioritizes regions of space for perceptual processing. Knowing how attended locations are represented is critical for understanding the architecture of attention. We examined the spatial reference frame of incidentally learned attention and asked how it is influenced by explicit, top-down knowledge. Participants performed a visual search task in which a target was more likely to appear in one, "rich," quadrant of the screen than in the others. The spatial relationship between the display and the viewer's perspective changed partway through the experiment. Because incidentally learned attention is persistent, the spatial bias that developed during training was present following the change in viewer perspective. Despite the presence of multiple environmental landmarks including a background scene, participants prioritized rich regions relative to their perspective, rather than relative to the environment. Remarkably, the egocentric attentional bias was unaffected by explicit knowledge of where the target was likely to appear. Although participants used this knowledge to prioritize the region of space they were told was likely to contain a target, a strong egocentric bias to a region that was unlikely to contain a target persisted. These data indicate that incidental attention differs fundamentally from attention driven by explicit knowledge. We propose that attention takes 2 forms. One is declarative, based on maps that explicitly prioritize some regions of space over others. The other is procedural, influenced by implicit knowledge that modulates how attention is moved through space.