Electronic Supplementary Material
Imprinting and flexibility in human face cognition
Urszula M. Marcinkowska1*, Julien Terraube2,3 & Gwenaël Kaminski4, 5.
1-Supplementary Methods
(a) Participants
In total, 2517 men completed the survey. The 76 participants that did not finish the survey were excluded a-priori. We excluded another 322 participants whose birth order, number, sex or age of siblings were not available. To increase standardisation of the sample as much as possible, we overlooked male participants that were not exclusively heterosexual (N = 375), that have either more of two sibling (N = 258) or more of four children (N = 83), as well as those that have a nationality that gathered less of 14 participants (N = 98).
Mean age (± SD) of the 1304 final participants was 26 (± 5.8) years, with the youngest and the oldest being 18 and 45, respectively (Table S1). 75% of the respondents had at least one sibling (Table S2 and Table S3). 50% indicated they were father and half of them stated that they had only one child (Table S4 and Table S5). 34% had signalled they had only other-sex sibling, 37% only same-sex sibling and, 29% both other-and same-sex siblings (Table S6). 51% of the participants were from European countries (Table S7).
Table S1. Age distribution of participants
Age / N / % /18 / 35 / 2.9
19 / 48 / 3.7
20 / 108 / 8.3
21 / 103 / 7.9
22 / 116 / 8.9
23 / 104 / 8.0
24 / 106 / 8.1
25 / 90 / 6.9
26 / 74 / 5.7
27 / 73 / 5.6
28 / 75 / 5.8
29 / 68 / 5.2
30 / 55 / 4.2
31 / 27 / 2.1
32 / 37 / 2.8
33 / 36 / 2.8
34 / 17 / 1.3
35 / 19 / 1.5
36 / 21 / 1.6
37 / 11 / 0.8
38 / 16 / 1.2
39 / 11 / 0.8
40 / 10 / 0.8
41 / 11 / 0.8
42 / 10 / 0.8
43 / 10 / 0.8
44 / 9 / 0.7
45 / 4 / 0.3
Table S2. Birth-order of participants
Birth-order / N / % /No-siblings / 320 / 24.5
First-born / 432 / 33.1
Middle-born / 438 / 33.6
Later-born / 114 / 8.7
Table S3. Number of siblings’ participants
Number of siblings / N / % /0 / 320 / 24.5
1 / 626 / 48.0
2 / 358 / 27.5
Table S4. Parenthood status of participants
Parenthood status / N / % /No-father / 658 / 50.5
Father / 646 / 49.5
Table S5. Number of children’s participants
Table S6. Sibling composition
Sibling composition / Age of sibling / N / % /Only other-sex sibling / Older / 163 / 16.7
Younger / 171 / 17.4
Only same-sex sibling / Older / 167 / 17.0
Younger / 194 / 19.7
Both same- and other-sex sibling / Older / 78 / 7.9
Younger / 67 / 6.8
Both / 144 / 14.6
Table S7. Nationalities of all participants
Nationality / N / % /Nepal / 187 / 14.3
France / 122 / 9.4
Estonia / 106 / 8.1
Slovakia / 105 / 8.1
Finland / 86 / 6.6
Iran / 80 / 6.1
Mexico / 69 / 5.3
Russian Federation / 58 / 4.5
Singapore / 47 / 3.6
China / 44 / 3.4
Poland / 41 / 3.1
Australia / 40 / 3.1
Latvia / 37 / 2.8
United Kingdom / 36 / 2.8
New Zealand / 34 / 2.6
Colombia / 32 / 2.5
Brazil / 26 / 2.0
Sweden / 26 / 2.0
Portugal / 24 / 1.8
Romania / 24 / 1.8
United States / 18 / 1.4
Germany / 17 / 1.3
Spain / 16 / 1.2
Czech Republic / 15 / 1.2
Croatia / 14 / 1.1
(b) Sibling effects
To assess the relevance of sibling effects on the preference for femininity, we mixed three factors in order to create a single variable, called Family Composition, with 15 modalities (Table S8): the participants’ birth-order (4 modalities, Table S2), their number of siblings (3 modalities, Table S3) and the sibling composition (3 modalities, Table S4).
Table S8. Description of the Family Composition variable with the absolute (N) and relative (%) number of participants in each of the 15 modalities. OO: Older Other-sex; OS: Older Same-sex; YS: Younger Same-sex, YO: Younger Other-Sex.
Family Composition / Birth-order / Sibling composition / Number of Siblings / N / % /1 / Later-born / OO / 2 / 16 / 1.2
2 / Later-born / OS / 2 / 20 / 1.5
3 / Later-born / OO and OS / 2 / 78 / 6.0
4 / Middle-born / OO and YS / 2 / 48 / 3.7
5 / Middle-born / OO and YO / 2 / 33 / 2.5
9 / Middle-born / OS and YS / 2 / 27 / 2.1
10 / Middle-born / OS and YO / 2 / 36 / 2.8
12 / Later-born / OO / 1 / 147 / 11.3
14 / Later-born / OS / 1 / 147 / 11.3
6 / First-born / YO and YS / 2 / 67 / 5.1
7 / First-born / YO / 2 / 15 / 1.2
8 / First-born / YS / 2 / 18 / 1.4
11 / First-born / YO / 1 / 156 / 12.0
13 / First-born / YS / 1 / 176 / 13.5
15 / No-siblings / - / 0 / 320 / 24.5
(c) Stimuli
The stimuli used in this experiment consisted of pairs composed of a masculinized and feminized version of the same woman face. Twenty pairs were generated (for example, see Figure S1). Facial pictures of women were manipulated with PSYCHOMORPH Programme (Tiddeman, Burt and Perrett 2001) on a femininity–masculinity spectrum. Pictures were created by subtracting or adding 50% of the linear difference between an average male and female face (prepared beforehand from 40 aged-matched individual pictures).
Figure S1: Example of feminized (right) and masculinized (left) female faces used as stimuli. The individual whose face appears here gave signed consent for this photo to be published in this article; she is not among the items used in the experiment.
(d) Data analysis
Before performing our multiple logistic regression analyses, we checked for potential problems of collinearity. We found that most associations among independent variables were all weakly significant. For the Family Composition variable, the number of siblings’ participant and birth order were the two strongest association (Cramer’s V = 1); all the others were less than 0.23. For the parenthood status, the strongest association was obviously with children (Cramer’s V = 1), followed by age (Odds-ratio =1.11 [1.11-1.12], i.e., the odds of becoming parent (in our study’s participants) is 1.11 times for each year of life increase); all others were less than 0.18. In our final model, the two independent variables, Family Composition and parenthood status, were weakly associated (Cramer’s V = 0.11).
References
Tiddeman B, Burt DM, Perrett DI. 2001 Computer graphics in facial perception research. IEEE Comp. Graph. Appl. 21, 42 – 50. (doi:10.1109/38.946630)
2-Supplementary Results
The probability of preferences for feminine faces at a typical man from our sample was 0.75 [0.67-0.83]. 23% of the variance in preference for feminine faces could be explained by participant identification. Stimuli identification explained a significant 18% of the variation in preference, and nationality explained also a small (2%) but significant part of this variation.
(a) Family composition effect
There was a significant variation among sibgroup modalities in preferences for feminine faces (F14, 25725=2.07; P=0.011), with the weakest and the greatest mean being 0.72 and 0.81, respectively (Table S9).
Table S9: Family Composition effect in preferences for feminine faces (mean and 95% CI)
As Family Composition variable was created by mixed three factors, we showed mean for each factor (Tables S10, S11 and S12)
Table S10. Birth-order effect in preferences for feminine faces (mean and 95% CI)
Birth-order / Mean / 95% CL /No-siblings (15) / 0.75 / 0.67 to 0.82
First-born (6,7,8,11,13) / 0.75 / 0.67 to 0.82
Middle-born (4,5,9,10) / 0.78 / 0.70 to 0.83
Later-born (1,2,3,12,14) / 0.75 / 0.67 to 0.82
Note: The different modalities of the ‘Family composition’ variable (see Table S8), that are included for each comparison, are indicated between brackets
Table S11. Number of siblings’ participants effect in preferences for feminine faces (mean and 95% CI)
Number of siblings / Mean / 95% CL /0 (15) / 0.75 / 0.67 to 0.82
1 (11,12,13,14) / 0.75 / 0.67 to 0.81
2 (1,2,3,4,5,6,7,8,9,10) / 0.76 / 0.68 to 0.83
Note: The different modalities of the ‘Family composition’ variable (see Table S8) are indicated between brackets
Table S12. Sibling composition effect in preferences for feminine faces (mean and 95% CI)
Sibling composition / Age of sibling / Mean / 95% CL /Only other-sex sibling / All (1,5,7,11,12) / 0.74 / 0.66 to 0.81
Older (1,12) / 0.74 / 0.65 to 0.81
Younger (7,11) / 0.72 / 0.63 to 0.80
Only same-sex sibling / All (2,8,9,13,14) / 0.78 / 0.70 to 0.84
Older (2,14) / 0.77 / 0.69 to 0.83
Younger (8,13) / 0.78 / 0.70 to 0.84
Both same- and other-sex sibling / All (3,4,6,10) / 0.75 / 0.67 to 0.82
Older (3) / 0.74 / 0.65 to 0.81
Younger (6) / 0.75 / 0.65 to 0.82
Note: The different modalities of the ‘Family composition’ (see Table S8) are indicated between brackets
Table S13. Effect of participants’ siblings’ sex and age on feminine faces (mean and 95% CI)
Age of sibling / Sister / Brother / Chi2 df=1 / P // Mean / 95% CL / Mean / 95% CL /
All
(1,5,7,11,12 vs 2,8,9,13,14) / 0.74 / 0.66 to 0.81 / 0.78 / 0.70 to 0.84 / 8.05 / 0.005
Younger (exclusively)
(7,11 vs 8,13) / 0.72 / 0.64 to 0.80 / 0.77 / 0.69 to 0.84 / 6.01 / 0.014
Younger (at least one)
(5,7,10,11 vs 4,8,9,13) / 0.75 / 0.67 to 0.82 / 0.78 / 0.70 to 0.84 / 4.72 / 0.029
Older (exclusively)
(1,12 vs 2,14) / 0.74 / 0.65 to 0.82 / 0.77 / 0.69 to 0.83 / 1.35 / 0.24
Older (at least one)
(1,4,5,12 vs 2,9,10,14) / 0.75 / 0.67 to 0.82 / 0.78 / 0.70 to 0.84 / 3.8 / 0.051
Note: The different modalities of the ‘Family composition’ variable (see Table S8) are indicated between brackets.
(b) Parenthood status effect
There was a marginal significant effect of parenthood status (F1, 25725=3.57; P=0.059, Table S5) on the preference for femininity with father that had 1.1 times more preference for feminine faces than childless’ men (OR =1.11 [0.99-1.24]).
Table S14: Parenthood status effect on preferences for feminine faces (means and 95% CI)
Parenthood status / Mean / 95% CL /Father / 0.77 / 0.69 to 0.83
No-Father / 0.75 / 0.66 to 0.81
(c) Effects on the interaction between family composition and Parenthood status
There was a robust interaction effect of the two factors, parenthood status and family composition (F14, 25725=3.11; P<0.0001, Table S6-S8 and Figure S1-S2).
Table S15: Means and odds-ratio (with 95% CI) of femininity preference for interaction between family composition and parenthood status.
Family composition / Childless men / Men with child / Chi2 df=1 / P / OR / 95% CL /1 / 0.70 / 0.77 / 0.80 / 0.37 / 0.70 / 0.32 to 1.53
2 / 0.78 / 0.78 / 0.01 / 0.97 / 0.99 / 0.60 to 1.64
3 / 0.73 / 0.74 / 0.47 / 0.49 / 0.92 / 0.73 to 1.16
4 / 0.73 / 0.77 / 2.12 / 0.15 / 0.80 / 0.59 to 1.08
5 / 0.74 / 0.81 / 3.78 / 0.052 / 0.69 / 0.47 to 1.00
6 / 0.77 / 0.73 / 2.72 / 0.09 / 1.24 / 0.96 to 1.61
7 / 0.63 / 0.79 / 8.02 / 0.005 / 0.44 / 0.25 to 0.78
8 / 0.77 / 0.81 / 0.64 / 0.42 / 0.80 / 0.46 to 1.39
9 / 0.78 / 0.84 / 2.61 / 0.11 / 0.68 / 0.43 to 1.08
10 / 0.80 / 0.73 / 4.29 / 0.04 / 1.45 / 1.02 to 2.06
11 / 0.71 / 0.75 / 4.01 / 0.04 / 0.84 / 0.70 to 0.99
12 / 0.75 / 0.74 / 0.17 / 0.68 / 1.04 / 0.87 to 1.24
13 / 0.77 / 0.74 / 4.04 / 0.04 / 1.18 / 1.00 to 1.40
14 / 0.77 / 0.76 / 0.61 / 0.43 / 1.07 / 0.89 to 1.29
15 / 0.77 / 0.73 / 9.81 / 0.002 / 1.11 / 1.04 to 1.17
Note: In family composition 5, 7 and 11, participants had only one or two sisters and at least one younger sister; in family composition 10, participants had an older brother and a younger sister and in family composition 15, participants had not siblings
Table S16. Means and odds-ratio (with 95% CI) of femininity preference for interaction between birth-order and parenthood status.
Birth-order / Childless’ men / Men with child / Chi2 df=1 / P / OR / 95% CL /No-siblings / 0.77 / 0.73 / 9.81 / 0.002 / 1.11 / 1.04 to 1.17
First-born / 0.73 / 0.76 / 3.42 / 0.06 / 0.92 / 0.84 to 1.01
Middle-born / 0.76 / 0.79 / 2.36 / 0.12 / 0.92 / 0.84 to 1.02
Later-born / 0.74 / 0.76 / 0.44 / 0.51 / 0.96 / 0.87 to 1.07
Note: The different modalities of the Family composition those are included for each comparison, are indicated in the table S10
Table S17: Means and odds-ratio (with 95% CI) of femininity preference for interaction between sibling characteristics (age and sex) and parenthood status.
Sibling / Others siblings / Childless’ men / Men with child / Chi2df=1 / P / OR / 95% CL /
Other-sex
Older / no / 0.72 / 0.76 / 0.6 / 0.43 / 0.92 / 0.75 to 1.13
yes / 0.73 / 0.77 / 3.92 / 0.048 / 0.82 / 0.67 to 0.99
Younger / no / 0.67 / 0.77 / 10.76 / 0.001 / 0.78 / 0.67 to 0.90
yes / 0.73 / 0.76 / 3.21 / 0.07 / 0.92 / 0.85 to 1.01
Same-sex
Older / no / 0.77 / 0.77 / 0.05 / 0.82 / 1.02 / 0.89 to 1.16
yes / 0.77 / 0.77 / 0.01 / 0.95 / 0.99 / 0.92 to 1.09
Younger / no / 0.77 / 0.78 / 0.04 / 0.85 / 0.98 / 0.85 to 1.14
yes / 0.76 / 0.78 / 1.02 / 0.31 / 0.96 / 0.88 to 1.04
Note: Older other-sex (no: 1,12 ; yes: 1,3,4,5,12) ; Younger other-sex (no: 7,11 ; yes: 5,6,7,10,11); Older same-sex (no: 2,14 ; yes: 2,3,9,10,14); Younger same-sexe (no: 8,3 ; yes: 4,6,8,9,13). In blue the different modalities of the Family composition (see Table S8) that are included for each comparison.