Abstract
Because serial search involves focal attention, components of a visual search array that are not actively searched may be poorly remembered and learned. However, statistical learning mechanisms may still encode aspects of an array that observers do not actively search. Here, using contextual cueing and a preview search paradigm (Hodsoll & Humphreys, 2005), we investigated the degree to which implicit learning of repeated visual search context depends on active search. On each trial of a visual search task, half of the distractors appeared 800 ms before the addition of the other distractors and the target. Previewing a subset of distractors that never contained the target should limit search largely among the newly added set that always contained the target. Unbeknownst to the participants, across multiple blocks, we repeated either the locations of the previewed distractors, the new locations of the newly added items, both, or neither. Unlike a previous finding (Hodsoll & Humphreys, 2005), repeating the newly added set facilitated search, reaffirming the presence of contextual cueing in attended and actively searched contexts. In addition and in line with those previous findings, repeating the previewed context also facilitated search, even though the previewed context was not actively searched. In fact, the effect size of contextual cueing was greater when the previewed set repeated than when the actively searched set repeated. These data demonstrate that although active search can induce learning, it is not the only factor driving contextual cueing. Temporally leading context may also be learned even when not actively searched. These findings are inconsistent with theories of contextual cueing that place learning at the level of oculomotor or attentional scan path. Furthermore, the finding shows that long-term selection history effects, exemplified by contextual cueing, interact with short-term selection history effects.