Visual search rate was used to assess attentional resources required for de
tection of opposing motions defined either by luminance or by modulations o
f texture contrast, flicker, or size. Though luminance-based targets were d
etected quickly, search through second-order motion was slow. Control exper
iments ruled out stimuli visibility, complexity, eccentricity sensitivity,
and attributes of the carrier as possible accounts. Results suggest separat
e processing of the two types of stimuli: Luminance-based motion is detecte
d by spatiotemporal filters, whereas second-order motion is likely processe
d by a capacity-limited, later stage. Rate-reducing effects of increased co
ntrast and speed mirrored previous research suggesting that effortful featu
re tracking may be the mechanism. (C) 2001 Optical Society of America.