Power analysis for logistic regression with interactions

I recently consulted on a study which aimed to examine whether people are better able to solve a logic puzzle, called the “Wason task” (see Cosmides, Barrett, & Tooby, 2010), when it is framed in terms of a moral dilemma (experimental condition) than when it is framed in a neutral way (control condition). My collaborator wanted to explore whether individual differences in morality moderated performance on this Wason task. Perhaps people who score higher in morality would be better able to solve the logic puzzle when it was framed in terms of a moral dilemma?