Subjects detected weak solutions of sucrose or citric acid under condi
tions in which attention was directed toward one of the tastants or th
e other. Detection thresholds were measured using an adaptive, forced-
choice procedure, with a three-down one-up rule, which computer simula
tions suggest should be more reliable than the popular two-down one-up
rule. The thresholds were modestly but systematically lower for atten
ded tastants than for unattended ones. Similar results have been repor
ted in other sense modalities, including vision (greater sensitivity t
o stimuli presented to attended versus unattended spatial locations) a
nd hearing (greater sensitivity to stimuli presented at attended versu
s unattended sound frequencies). Taken together, the findings are cons
istent with a general hypothesis regarding attention in sensory system
s: gains or losses in detectability occur when a central attentional m
echanism (or, conceivably, a preattentive mechanism) selectively and p
referentially monitors signals arising from particular subsets of peri
pheral neural inputs.