Recent work has shown that attention can be selectively directed at one com
ponent of transparent motion (Lankheet and Verstraten, 1995) and that it ca
nnot be directed efficiently at two components at the same time (Valdes-Sos
a, Bobes, Rodriguez, & Pinilla, 1998). Here we estimate the time taken to s
witch attention between two surfaces defined by transparent motion. Two set
s of differently colored dots that rotated in opposite directions around fi
xation were used to create the surfaces. Discrimination of a brief change i
n direction of motion of one set was used to probe the distribution of atte
ntion. Two probes were presented on different surfaces with a variable SOA
between them. Discrimination of the first probe was prioritized, which was
reflected in an accurate performance for all SOAs. The second probe was acc
urately discriminated only for SOAs above 450 Ins. These results are not du
e to spatial constraints on the division of attention, since the two surfac
es were superimposed and in close subjective proximity. The results are con
sistent with object-based attention. The estimated dwell time of about 450
ms is close to previous assessments based on identification of letters plac
ed at different locations (Ward, Duncan, & Shapiro, 1996), (C) 1999 Academi
c Press.