Three horses were trained with a discrimination task in which the color (bl
ue or yellow) of a center panel signaled the correct (left or right) respon
se (lever press). Reinforcing outcomes for the two correct color-position c
ombinations (blue-left and yellow-right) were varied across phases. Discrim
ination performance was better when the combinations were differentially re
inforced by two types of food (chopped carrot pieces and a solid food pelle
t) than when the combinations were randomly reinforced by these outcomes or
when there was a common reinforcer for each of the correct combinations. H
owever, the discrimination performance established by the differential outc
ome procedure was still 80% to 90% correct, and an analysis of two-trial se
quences revealed that the stimulus color of the preceding trial interfered
with discrimination performance on a given trial. Our demonstration of the
differential outcome effect in the horse and its further analysis might con
tribute to more efficient control of equine behavior in the laboratory as w
ell as in horse sports.