A robust implicit measure of facial attractiveness discrimination

Abstract

Decisions of attractiveness from the human face are made instantly and spontaneously, but robust implicit neural measures of facial attractiveness discrimination are currently lacking. Here we applied fast periodic visual stimulation coupled with electroencephalography (EEG) to objectively measure the neural coding of facial attractiveness. We presented different pictures of faces at 6 Hz, i.e. six faces/second, for a minute while participants attended to a central fixation cross and indicated whether the cross shortly changed color. Every other face in the stimulation was attractive and was replaced by a relatively less attractive face. This resulted in alternating more/less attractive faces at a 3 Hz rate, eliciting a significant increase in occipito-temporal EEG amplitude at 3 Hz both at the group and the individual participant level. This response was absent in two control conditions where either only attractive or only less attractive faces were presented. These observations support the view that face-sensitive visual areas discriminate attractiveness implicitly and rapidly from the human face.

Keywords: FPVS; facial attractiveness perception.

Fig. 1

Experimental design for the CONTRAST condition. Faces rated as more attractive (A) and less attractive (B) are alternating at 6 Hz presentation rate, reaching full contrast halfway through the image presentation cycle. More and less attractive faces are repeated every 3 Hz. Faces are randomly chosen from the respective set and have the same sex within a sequence. Sequence duration is 64 s including 2 s of fade in and fade out.

Fig. 2

Image characteristics. Results for female (A) and male (B) faces. Left: mean femininity/masculinity rating for each of the 12 facial images presented in the FPVS experiment. Middle: facial expression categorization for each image rated as more attractive. Right: facial expression categorization for each image rated as less attractive.

Fig. 3

Spectral representation and scalp distribution of EEG signal during FPVS. (A) Symmetrical response. Grand-averaged SNR spectra at 6 Hz at electrode Oz for CONTRAST (red), MORE ATTRACTIVE ONLY (yellow) and LESS ATTRACTIVE ONLY (brown) conditions. Topographical maps illustrate the distribution of the 6 Hz responses. Bar graph (Mean ± SEM) displaying summed baseline-corrected amplitudes (μV) for the 6 Hz response and its harmonics for each condition across the three ROIs, MO, LOT and ROT, indicates that the symmetrical response is largest over MO sites with no significant differences across conditions. (B) Asymmetrical response. Grand-averaged SNR spectra at 3 Hz at P10 for the CONTRAST (red), MORE ATTRACTIVE ONLY (yellow) and LESS ATTRACTIVE ONLY (brown) conditions. Topographical maps illustrate the distribution of the 3 Hz responses. Bar graph (Mean ± SEM), displaying baseline-corrected amplitudes (μV) for the 3 Hz response for each condition across the three ROIs, MO, LOT and ROT, indicates that the asymmetrical response is observed only for the CONTRAST condition, particularly over the ROT region.

Fig. 4

Individual asymmetrical response. (A) Individual asymmetrical response at 3 Hz for the CONTRAST condition. Individual displays are centered at 3 Hz and show data from 2.83 Hz to 3.18 Hz corresponding to the 10 surrounding bins with resolution = 0.0167. Channels are indicated above the displays. (B) Individual scalp maps showing the topography of the asymmetrical response at 3 Hz. SNR responses are displayed and the individual range indicated above the topographies (1 = noise level).