The present study is a mixed design that examines instruction followin
g, with the accuracy of instructions varied across phases within subje
cts and the schedule of reinforcement varied across six groups. A choi
ce experiment was conducted in which subjects were provided with instr
uctions that varied systematically in their accuracy. Across phases, t
he accuracy of instructions (expressed in percentages) was either 0-50
-100-50-0 or 100-50-0-50-100. Across groups, the schedule of reinforce
ment was one of the following: continuous, Fixed Ratio 2, or Fixed Rat
io 3. With continuous reinforcement, subjects showed collateral conseq
uence control and either followed or disobeyed instructions, in keepin
g with the predictions of reinforcement theory. When the schedule of r
einforcement was leaned, however, insensitivity was observed. The impl
ications of the results for the controversies regarding the concept of
rule-governed behavior are discussed.