This research furthers understanding of the factors that induce individuals to choose deception. Its goals were to expand and test a recent account of the cognition of deception, Activation–Decision–Construction–Action Theory (ADCAT), in a mock job interview inviting impromptu deception. Decisions to lie are hypothesized to depend on the truth and other information activated from long‐term memory by the social context. Activated information then guides evaluation of the likely costs of truth telling and benefits of deception; 166 college students participated in the job interview, who learned about the position and then adopted the role of job applicants. Afterward, participants shared their thoughts when responding to five questions from the interview. The most important findings are that the negative expectations of sharing truths and the positive expectations of sharing deceptions each account for unique variance in deciding to lie. Implications for lie reduction and detection are considered. Copyright © 2016 John Wiley & Sons, Ltd.