The design of any experiment involving more than one operant (all studies of choice, for example) requires the operants used to be both functionally equivalent and neutral for the participants prior to the experiment. In a series of three studies on learning history variables, persistent systematic biases were observed; these were associated both with the hand motions involved in executing each operant and with the operant’s visual aspects. This type of finding proves to be general and has implications for any behavior research that assumes equivalence among operants.