Previous research has demonstrated global dominance for attended and unatte
nded stimuli. In this paper, this phenomenon is shown to be restricted to s
mall compound stimuli. As a first step, local dominance was obtained with l
arge (8 deg in height) attended stimuli when a single stimulus was displaye
d. Next, dominance in attended and unattended stimuli was investigated by d
isplaying two large compound stimuli, one surrounded by a square (attended
compound stimulus), the other one enclosed in a circle (unattended compound
stimulus). The way attention was directed to the attended stimulus was var
ied. No dominance was observed when subjects were instructed to process the
stimulus appearing in the square (experiment 2). However, when a rapid-ons
et cue pre-directed attention to the attended stimulus, local dominance eme
rged for attended, but not for unattended stimuli (experiment 3). This latt
er result was obtained whether or not subjects were more experienced at loc
al than global processing (experiment 4). The implications of the results f
or the locus of processing dominance are discussed.