Context:
To ensure that the results were due to imagined consumption of M&M'S rather than the control task, the next experiment
manipulated the experience imagined (inserting quarters or eating M&M'S) and the number of times it was imagined. Again, the participants who imagined eating 30 M&M'S subsequently consumed fewer M&M'S than did the participants in the other groups.
More:
http://www.sciencedaily.com/releases/2010/12/101209141134.htm