I found consistently higher memory for unattractive over both attractive and medium-attractive faces

The influence of facial attractiveness on recognition memory: Behavioural findings and electrophysiological evidence. Carolin S. Altmann, Dissertation zur Erlangung des akademischen Grades doctor philosophiae (Dr. phil.), Friedrich-Schiller-Universität, Jena.

This thesis is based on a series of experiments that aimed to better understand the influence attractiveness has on memory while always controlling for perceived distinctiveness. First, I created a large stimulus pool of 1100 faces and obtained ratings on a number of relevant dimensions (see sections 2.1 and 3). In the first study, I investigated if memory for faces increased linearly with increasing attractiveness or whether this relationship was more complex (see section 3). In the second study, I investigated the combined influence of attractiveness and gender on recognition memory to competitively test predictions of perceptual expertise, social cognition, and alternative accounts (see section 4). In the third study, l, I investigated encoding-related neural correlates of the attractiveness effect on memory at retrieval (see section 5) whereas the first two experiments focused on ERP memory effects during retrieval.

Taken together, I found consistently higher memory for unattractive over both attractive and medium-attractive faces. Further, medium-attractive faces were significantly less well remembered than attractive faces in studies 1 and 2, and numerically in study 3. This difference disappeared when emotional relevance, i.e. valence and arousal, was taken into account. Inspection of ERPs showed increased P2 amplitudes for medium-attractive faces at retrieval in studies 1 and 2, and a pronounced Dm effect in this component in study 3. Thus, the attractiveness effect on face recognition memory seems already rooted in evolved, i.e. more refined and higher-level, perceptual processing of faces reflected in the P2. Overarchingly, these findings argue in favour of perceptual accounts, i.e. representational clustering (see sections 1.6.1 and 6.2.4), as both attractive and medium-attractive faces are supposedly more densely clustered in participants’ mental storage. The current data further indicate some contribution of emotional relevance. As no significant influence of face or participant gender was observed, there was also no compelling evidence for accounts of social cognition.