Preference for Facial Self-Similarity (Part II)
Facial resemblance can be an important cue to kinship, and one that alters our perceptions of attractiveness. In previous studies, facial self-resemblance was found to increase attractiveness judgments of same-sex faces more than other sex-faces. We plan to test this hypothesis with homosexual and heterosexual individuals to further understand the evolutionary pressures that create this effect.
Preference for Facial Self-Similarity (Part II)
Facial resemblance can be an important cue to kinship, and one that alters our perceptions of attractiveness. In previous studies, facial self-resemblance was found to increase attractiveness judgments of same-sex faces more than other-sex faces. We plan to test this hypothesis with homosexual and heterosexual individuals to further understand the evolutionary pressures that create this effect.
Average composite faces were made using a publicly available face database from the University of Texas, Dallas. White male and female composites were created using 20 individuals from the face database of the same sex and phenotypic category. 172 points were placed on each face to delineate facial features. Image manipulation software, JPsychomorph, was used to combine shape, color, and texture information from each of the photographs to create the average composite face:
Participants will have their faces photographed a week before the experiment for an unrelated study about facial masculinity and facial memory. Doing so in advance was important for two reasons: first, this allowed us to match the Debruine procedure; second, it decreased the chance that participants would recognize their face in the transforms and thus guess the nature of the experiment. After photographs are taken, faces will be delineated using the same 172 points as before. For each participant face, we will create a total of four transforms: similar same-sex, similar other-sex, dissimilar same-sex, and dissimilar other-sex. The facially-similar transforms will be made in JPsychomorph by applying 50% of the shape difference between the participant’s face and composite face. Facially-dissimilar transforms will be created by applying 50% of the shape difference in the opposite direction to create, in essence, their "anti-face."
According to the JPsychomorph documentation,  the shape transformation process can be conceptualized using the following equation:
New = Original + Percent_Transform * (Original – Source)
In a simple two-dimensional representation of this morphing process, imagine that the center of the participant’s right eye is located at point x. The center of the composite face’s right eye is located at point x + 4. The center of the 50% facially-similar transform’s right eye would be located at point x + 2; however the right eye of the facially-dissimilar transform would be centered at point x – 2.
JPsychomorph performs this operation in three-dimensional space using the 172 points delimited on each face. After similar and dissimilar transforms are created, we will use a separate piece of morphing software, Fantamorph, to create a 61-frame morph sequence from the similar to dissimilar face for both same-sex and other-sex transforms.
Participants will be presented with a series of face morph sequences and asked to choose the face that they found most attractive using the circular slider. As the participant moves the mouse around the circular slider to morph the face, located in the center of the circle.
Participants will make a total of 20 face judgments of participant face morphs, 10 female and 10 male. The sequences will be presented in random order.
C) Statistical Methods
Each face morph sequence consists of 61 faces. Face 1 would be the self-similar face and face 61 would be the self-dissimilar face. Using the circular slider, participants select the face from the morph sequence that they find most attractive. This face is associated with an index in the sequence. The participant’s facial preference index will be recorded on each trial. Since each participant performs multiple trials, we will compute the mean index for each category (male and female participant morphs), which will serve as the participant’s facial similarity preference scores.
A factorial ANOVA will be used to determine the effects of the participants’ sexual preference and face sex on their facial similarity preference scores. An interaction between sexual preference, face sex, and facial similarity preference scores would support the prediction that the there exists a difference in resemblance preferences between heterosexual and homosexual males. Independent-samples t-tests will also be performed to investigate the nature of the relationships, and one-sample t-tests will be used to compare facial similarity preference scores with the similarity midpoint index, 31.
 The application download and documentation for JPsychomorph can be found at the following website: http://cherry.dcs.aber.ac.uk:8080/wiki/jJPsychomorph