Running Head: Memory for emotional faces Memory for emotional faces in naturally occurring dysphoria and induced negative mood

The majority of mood-congruent memory research has confirmed the existence of a memory bias for affectively toned words or phrases. However, the current study investigated a memory bias for emotional facial expressions, in induced and naturally occurring mood states. In experiment 1 twenty-five dysphoric participants and twenty non-dysphoric participants were presented with a set of emotional facial expressions and asked to identify the emotions portrayed by each face. The participants were then tested on recognition memory for these facial expressions. It was found that dysphoric participants correctly identified more happy facial expressions compared with sad and neutral expressions. During the recognition memory phase the dysphoric participants demonstrated superior memory for sad facial expressions and inferior memory for happy expressions relative to nondysphoric participants. In experiment 2 twenty-four healthy undergraduate students underwent a positive mood induction (group 1), and another twenty-four underwent a negative mood induction (group 2). These participants also took part in the same emotion identification trial and recognition memory phase, same as participants in experiment 1. All participants identified happy facial expressions more accurately than neutral or sad facial expressions. In the recognition phase, group 1 recognised more happy facial expressions, and group 2 recognised more sad facial expressions, supporting the recognition hypotheses. Findings were discussed in terms of Bower’s (1981; 1987) Network Theory of Affect.