Zanna, M. P., & Cooper J., Dissonance and the Pill: An attribution approach to studying the arousal properties of dissonance. JPSP, 1974, 29, 703-709

This study tested the idea of whether dissonance had arousal properties.

Cognitive dissonance is often experienced as psychological discomfort or tension. Yet prior research hasn't proven evidence of dissonance-produced attitude change.

Previous studies by Schachter showed that subjects unknowingly under the influence of epinephrine could be made to label their arousal as either angry or happy based on external stimuli. Another study with insomniacs showed that if they could get subjects to attribute their nocturnal arousal to a pill, they found it easier to fall asleep (and if they were told it would make them more relaxed and it didn't work they were even more aroused).

Theoretically, this should also work with attitude change. People who are put into a cognitive dissonance situation but can attribute their arousal to another factor (a pill) should be less likely to change their attitudes than people who don't attribute their arousal to the pill.

Procedure
Subjects were told to write an essay counter to their beliefs (about free speech on campus). One third were given a pill and told it would make them tense, one third told it would make them relaxed, and one-third were not given a pill at all. In this 2X3 experiment, the other variable was giving the subjects either high-choice in writing the counteressay (it's up to you...) or low choice (do this...).

After writing the essay they completed an attitude questionnaire that asked them about campus speech and their present feelings.

Results
People who were told the drug would make them feel more tense did indicated they were more tense, and those in the "relaxed" condition felt more relaxed (vs control). Interestingly, the people in the high-choice control group reported more tension than either the "tense" or "relaxed" groups (which is expected by dissonance theory).

The overall results were as expected. For people in the control group, those in the high choice condition had a bigger attitude change (agree with the ban on speech) than the low-choice people. They had nothing to attribute their action on the essay to.

In the "tension" condiition, subjects were able to attribute their tenseness to the pill and not the essay, so the dissonance effect was realized and their attitudes didn't change (agreement with the ban was low). On the contrary, in the "relaxed" condition there was increased cognitive dissonance (they felt tense rather than relaxed) and their shift in favor of the ban was more pronounced than with the control group.