Spatial attention was measured in visual search tasks using a spatial
probe. Both speed and accuracy measures showed that in a conjunction t
ask, spatial attention was allocated to locations according to the pre
sence of target features. Also, contrary to some predictions, spatial
attention was used when a clearly distinguishable feature defined the
target. The results raise questions about any account that assumes sep
arate mechanisms for feature and conjunction search. The probe method
demonstrated here allows a very direct measurement of attentional allo
cation, and may uncover aspects of selection not revealed by visual se
arch.