Fig 1.
Top row: Random phase patterns with different slopes (as indicated) of the radially averaged Fourier power spectrum, and a mid-grey control image. Bottom row: Stimuli used in Study 2A. The composite stimuli consists of the original face image (FACES database [59]) with the respective image of the top row superimposed at an opacity of 15% and a black oval window. Note that differences between the conditions are subtle and might be invisible due to the small size and low resolution of the images here. Images of higher resolution are provided as supplemental material (S1 Fig).
Fig 2.
Log-log plot of radially averaged spectral power versus spatial frequency.
Average curves are given for male and female face images from the dataset used in Study 1A (FACES database [59]).
Table 1.
Pearson correlation coefficients of Study 1A (Nfaces = 110; Nraters = 40).
Table 2.
Pearson correlation coefficients of Study 1B (Nfaces = 479; Nraters = 24).
Table 3.
Results of second-level regression models of Study 1B using rated age to control for age effect (Nfaces = 479; Nraters = 24).
Table 4.
Chronological age of depicted individuals used in Study 2A.
Table 5.
Results of Fourier transformation of stimuli used in Studies 2A, 2B, 2C and 3A.
Fig 3.
Results of attractiveness rating (A; N = 20) and age estimation (B; N = 21) in Study 2A.
The different symbols and lines represent mean attractiveness or age ratings, respectively, for each stimulus category, as indicated, and the overall mean (thick solid line). Error bars represent standard error. For a statistical analysis, see text.
Fig 4.
Example of the stimuli used in Studies 2B and 2C with corresponding Fourier slope in parentheses.
Face images (from the FACES database [59]) with manipulated slopes of the radially averaged log-log Fourier power spectrum. A: Stimuli used in Study 2B. B: Stimuli used in Study 2C. Note that the subjective impression of each image changes with size and viewing distance, and that the images with shallower slopes become more blurry at this resolution due to the reduction of power in the low spatial frequencies. Images of original size are provided as supplemental material (S2 Fig and S3 Fig).
Table 6.
Summary of results of Study 2B.
Fig 5.
Faces from FACES database [59] in front of random phase patterns with different slopes in the radially averaged log-log Fourier power spectrum.
Fig 6.
Different symbols and lines represent different stimulus categories and the overall mean (thick solid line). Error bars represent standard-error. For a statistical analysis, see text.
Table 7.
Summary of results of Study 3B and pairwise comparisons (Wilcoxon tests).