This project seeks to examine the mechanisms involved in the averaging of emotional facial expressions. Based on the findings of Hubert, Wallander & Boynton (2015) we aimed to investigate the role of working memory capacity in averaging facial expressions. To do so, (i) we first varied the number of stimuli to average and created sequences of 3, 5 and 8 stimuli, presented in random order. We then aimed to look at the individual differences, (ii) using the task of Digit Span, with the aim to examine probable correlations between the weights attributed to the last stimuli in the sequences of 3, 5 and 8 and the performance in Digit Span. The results showed that recency effects were more pronounced in sequences with a smaller number of stimuli (series of 3) and there were no significant correlations between the weights assigned to the latter stimuli and the performance in Digit Span. However, there were two numerically strong correlations that emerged in opposite directions. It seems that the pattern of the correlations is not consistent which may be related to the small number of participants.