Average can be good (on average)
The Most Average Facial Appearance Effect (MAFA effect for short) is the tendency to prefer faces in which the eyes, nose, lips and other features are close to the average for a ‘population’, and that faces which deviate significantly from that average are considered less attractive. Population generally refers to a person’s frame of reference, which is usually based on the group in which they were born and brought up, while average refers to the arithmetic mean of form, size and position of specific features.
It has been frequently demonstrated that when many pictures from a population are combined to form a single composite image then that composite image is considered relatively attractive by the majority of people and is usually similar to the facial configuration of professional models from the same population. [However, for individuals within the population it is often true that more idiosyncratic features may be considered even more attractive, although such individual preferences are never consistent.]
Symmetry and nature
This effect is most likely the result of evolution, symmetry and cognitive prototypes. In particular, evolution tends to select out population extremes over time, and therefore ‘averageness’ may be an indicator of general fitness. Also, average faces are symmetrical and this has long been known to indicate health and fitness, as those with asymmetrical features in any species tend to have fewer offspring and live shorter lives (asymmetry is often the result of disease, malnutrition or bad genes). Finally, as we see the faces of many other people, we build mental representations which are based on composite pictures (average patterns), which would make average facial features closer to our mental representation of a ‘face’.
These effects are demonstrated in the ways in which racial preferences develop among isolated groups (and how they are changing in modern societies). In isolation, groups form cognitive prototypes based solely on their own group members and these drive their preferences. When different groups first encounter each other, other group members will seem cognitively strange and less attractive (different to their existing mental representations), but as the differences become more familiar and members of the other group are assimilated into mental representations, the prototypes are updated and definitions of beauty change to reflect this.
In summary, the most average facial appearance is a good benchmark for ‘beauty’ in any population, although other elements will contribute to this too (for example, smiles vs scowls). Composite images of faces created from randomly sampled faces of a target population are a good proxy of local perceptions of beauty (as long as the ‘population’ is correctly defined and framed). These are important considerations for any marketing or advertising campaign which seeks to target a specific population, especially when models and budgets are unavailable or constrained.
Principles of Design, Revised and Updated by William Lidwell, Kritina Holden and Jill Butler (2010)