In three experiments with 78 3-month-olds, we asked what determines wh
ether or not a stimulus will pop out and cue retrieval from long-term
memory. All infants were trained with mobiles displaying either Qs (fe
ature-present stimuli) or Os (feature-absent stimuli) and were tested
24 h later. When the diagonal line of the Q bisected its rim, feature-
absent stimuli controlled retrieval in tests with homogeneous displays
, and stimulus novelty controlled retrieval in tests with pop-out disp
lays. A follow-up experiment revealed that the similarity between Q an
d O determined whether or not Q popped out: When its tail projected ex
ternally from the rim, Q popped out and cued retrieval, but O did not
(search asymmetry). When its tail projected internally from the rim, h
owever, 3-month-olds failed to discriminate Q from O (the externality
effect). These data reveal that target-distracter similarity constrain
s whether or not a feature-present stimulus will pop out and cue retri
eval. (C) 1998 Academic Press.