An initial analysis was used to test whether the order of sessions had an influence on the results. Difference-scores were subject to a three-way mixed measures ANOVA with two withinsubjects factors: ‘identity condition’ (same-identity, different-identity) and ‘target expression’ (Neutral–Angry, Happy–Angry), and one between-subjects factor: ‘session order’
(Neutral–Angry then Happy–Angry, Happy–Angry then Neutral–Angry).
There was no main effect of session order and the two- and three-way interactions involving this variable did not reach significance (all p > .05). The data were therefore collapsed across the counterbalanced groups (see Fig. 2).