Deceptive message production in high and low self-monitors : the impact of directed versus undirected rehearsal time