Abstract
Most theories of visual processing assume that a target will pop-out from a field of distractors if targets and distractors can be discriminated preattentively. The distinction between pop-out and serial search is classically inferred from set size effects in visual search tasks, while the attentional requirements of a visual discrimination task are often assessed using dual-task paradigms, by measuring interference from a concurrently performed attentionally-demanding task.
Here we show that there is no such equivalence: while simple feature discrimination tasks that can be performed preattentively lead to pop-out in visual search (e.g. color or orientation discrimination), and many tasks that require attention need serial examination in visual search (e.g. rotated L vs. T or red-green vs. green-red patches), other tasks do not trigger pop-out even though they can be performed preattentively (natural scene categorization, color-orientation conjunction discrimination). Furthermore, certain targets that pop-out among distractors need attention to be effectively discriminated when presented in isolation (rotated L vs. +, cubes with different illumination directions). In other words, the distinctions “pop-out vs. serial search” and “pre-attentive vs. attentive processing” can be independent.
We suggest that (i) attentional requirements depend on the existence of specific neuronal populations selective to the target and distractor categories, independent of the level of processing involved (from V1 to IT), while (ii) set size effects in visual search are mainly determined by the level of complexity and the locus of the discrimination; pop-out occurs for targets that are discriminated in early areas (V1–V2) and not in higher areas (V4, IT) due to a combination of 2 related factors: receptive field size (leading to competitive interactions in higher-level visual areas) and non-classical receptive field interactions (e.g. grouping in lower-level areas).