In the present study, we reported the findings of three experiments aimed at replicating Kessler and McKenna’s Overlay Study15. Unlike Kessler and McKenna, who administered stylized drawings of the human body (Supplementary Table S16), we administered the Sex/Gender Attribution Test for Adult13, created by digitally morphing bodies of two human male and female models into realistic images (Supplementary Table S16). We assumed that, because evolved cognitive mechanisms are triggered on very specialized inputs22, more ecological and lifelike stimuli could have returned more reliable information about evolved cognitive processes. We expected, based on previous studies13,15, to find that primary sexual characteristics (genitals) would determine sex attribution (male/female) more than secondary/gender-linked sexual characteristics (short/long hair, male/female face, flat chest, breasts, narrow/wide hips, and body/no body hair), and that male sexual characteristics would determine sex attribution more than female sexual characteristics, with a significantly stronger effect of the penis compared to the vulva, ceteris paribus. The results have substantially reconfirmed the results obtained in the previous studies13,15. To facilitate the reading of the data, already discussed experiment by experiment above, we have reported the main results, in synoptic form, in Table 1.
Table 1
Synoptic table of the main results obtained in the three studies (Experiments 1, 2, and 3)
|
Experiment 1
|
Experiment 2
|
Experiment 3
|
Participants
(N = 1,706)
|
n = 897
Mean age = 21
Females = 54.7%
|
n = 30
Mean age = 21
Females = 70%
|
n = 779
Mean age = 22
Females = 57.9%
|
Kinsey scale
|
Heterosexual = 83.7%
|
Heterosexual = 50%
|
Heterosexual = 88.7%
|
Political opinions effect on sex attribution
|
Left-wing: lower percentage of female sex attribution
|
No difference
|
No difference
|
Cronbach’s alpha
|
α = .90 sex attribution
α = .99 confidence
α = .99 pleasantness
|
α = .70 sex attribution
α = .99 confidence
|
α = .86 sex attribution
|
Male attribution (vs. female attribution)
|
60% all stimuli
84.4% male stimuli
35.3% female stimuli
60.1% neutral stimuli
83.3% neutral/male
63.2% neutral/female
87.7% penis exposed
69.6% vulva exposed
|
58.1% all stimuli
82.0% male stimuli
34.3% female stimuli
57.8% neutral stimuli
81.4% neutral/male
65.3% neutral/female
88.4% penis exposed
73.3% vulva exposed
|
58.8% all stimuli
82.7% male stimuli
35.1% female stimuli
58.6% neutral stimuli
80.1% neutral/male
62.6% neutral/female
84.3% penis exposed
68.5% vulva exposed
|
Confidence (certain = score 7)
|
27.1% penis exposed
20% vulva exposed
|
27.1% penis exposed
19.3% vulva exposed
|
Not applicable
|
Confidence (uncertain = scores 1–6)
|
78.6% on female attribution
73.2% on male attribution
|
80.6% on female attribution
74.5% on male attribution
|
Not applicable
|
Unpleasantness (score 1)
|
Neutral stimuli = 38.6%
Female stimuli = 31.3%
Male stimuli = 30.5%
|
Not applicable
|
Not applicable
|
When the penis was apparent in a picture, the participants attributed male sex significantly more often (84.3–88.4 percent) than female sex when the vulva was apparent (69.6–73.3 percent). In other words, when male external genitalia were exposed (not covered by clothes), the odds of male sex attribution are 5.688 compared to 1.823 female attribution when female genitalia were exposed (Experiment 1). In addition, the certainty in sex attribution was reported to be greater when participants had attributed male versus female sex (Experiments 1 and 2). Furthermore, the participants attributed male sex to neutral stimuli 2.34, 2.83, and 2.17 times more often when the penis was displayed than when the vulva was shown in the Experiments 1, 2, and 3, respectively.
All findings had a strong statistical significance (Experiments 1, 2, and 3, p = < .001), substantially confirming those found by Federici and colleagues13 with the same realistic stimuli (SGAT-A) and by Kessler and McKenna15 with stylized drawings of the human body (Overlay Study).
Although all the sexual characteristics significantly contribute to the attribution of sex, female attribution appears to be triggered only when every other male cue has been excluded (Experiments 1 and 3). In other words, the presence of female characteristics reduces the probability of assigning male sex. Therefore, all other things being constant, a female cue is recognized as such only in the absence of male cues. Whereas a male sex cue most likely equals male, a female cue equals female with much less probability, confirming Federici et al.’s13 and Kessler and McKenna’s15 findings.
The most salient variables were the penis followed by masculine face. The face assumed a prevalent role along with the flat chest and breast only when the external genitalia were covered by jeans (Experiments 1 and 3). The male face is an excellent predictor of male sex attribution23,24 and, if associated with the penis can overshadow all other female cues13,15. These findings have confirmed what other similar studies found13,15,25: The penis and male sexual characteristics make the difference in sex recognition.
The salience of male versus female sexual characteristics suggests that the psychological mechanism does not operate on a dichotomous concept and binary sex categorization, but rather to solve an adaptive problem of avoiding at all costs a false negative by detecting a female when it is male26. In fact, for all individuals (e.g., infants, females), the risk of socializing with a male is greater than with a female, because male individuals tend to be physically stronger and much more aggressive. From this perspective, to mistake a female when it is a male is potentially more dangerous than the opposite for human survival. Therefore, a subtraction is applied to the higher danger condition (male) to get to the lack-of-danger condition (non-male). In other words, to survive, it is much more convenient to make a wrong female than a wrong male sex attribution. These errors of judgment are determined by cognitive mechanisms evolved by natural selection that “occurred despite the fact that subjects were encouraged to be accurate and were rewarded for the correct answers”27.
In Experiment 3, we have also examined the time taken by participants to attribute sex after each stimulus was displayed. As expected, the shortest times were observed with masculine stimuli (i.e., with male variables > 3; Supplementary Tables S16 and S17), whereas the longest times were observed with neutral stimuli (i.e., with a balanced co-presence of 3 male and 3 female variables; Supplementary Tables S16 and S17). This result mirrors those by Simpkins24, who showed participants images of parts of real human faces, created by photographing 28 people who varied by sex, ethnicity, and age, on the grounds that the face is “usually the first source of information available about a person”23. The study concluded with a significant disproportion between the times in which the participants attributed male sex to the stimuli compared to times in which they attributed female sex to the stimuli. This was similarly demonstrated in the study by Wenzlaff and colleagues25, which replicated Kessler and McKenna’s study using eye tracking on digital reproductions of original stimuli. Participants gazed longer when they attributed female sex when a penis was present, than when they attributed male sex with a vulva shown. This is indicative of higher cognitive effort and more difficulty ignoring the penis as opposed to the vulva. Attributing a female sex when the individual might be a male requires a more careful and effortful attentional and decision-making process that also involves inhibiting the cognitive bias mechanism of male preference.
Experiment 2 also provided us with pilot data on cross-cultural differences on sex attribution. Comparing different cultures allows us to investigate whether the process of sex attribution can be ascribed more to human universals14,28—that is, to the effect of evolved psychological mechanisms—rather than to the influence of memes. Although the results were based on a small-scale sample, not generalizable to be representative of another culture (Chinese), nevertheless they encourage us to continue our cross-cultural research. The fact that the young Chinese participants behaved similarly to those of the two Italian samples in attributing sex to stimuli from Western human models allows us to infer that the psychological mechanism of sex recognition may be metacultural and precede any form of ethnic differentiation. In the Pleistocene, our ancestors must have already been able to reliably proceed, being adaptively successfully in recognizing the sex of conspecifics despite the individual phenotypic variation. We do not have a recognition mechanism that can differentiate a chicken from a hen, because it is not essential to our fitness. But we certainly needed to know how to discriminate the sex of a Neanderthal, Denisovans, or Homo sapiens29 before any ethnic (cultural) differentiation.
Overall, the study supports the assumption that thinking a person is a male rather than a female is more likely and quicker, ceteris paribus, because maximizing male sex attribution reduces the risk of a false negative. We read these results not by limiting ourselves to an ethnomethodological perspective, as Kessler and McKenna did in the 1970s, but by integrating this with assumptions from evolutionary psychology and cognitive science according to an Integrated Causal Model stating that “the distinction between the biologically determined and the nonbiologically determined can be seen to be a nondistinction”30.
The adaptive strategy, evolved in a psychological architecture of mind, neither excludes nor minimizes the cultural gender construction. Human minds and behavior, human artifacts and culture are
all biological phenomena—aspects of the phenotypes of humans and their relationships with one another. The rich complexity of each individual is produced by a cognitive architecture, embodied in a physiological system, which interacts with the social and nonsocial world that surrounds it.30
This suggests that what Kessler and McKenna15 argued from a constructionist point of view that “‘[sex] assignment’ and ‘gender construction’ may be synonymous”15 does not contradict the biological outlook according to psychological adaptive mechanisms evolved to respond to specific problems raised by the environment affecting human sex attribution. The fact that, in a phallocentric culture, a penis makes somebody more often a male person rather than a female one does not negate the fact that these cultural constructions were guided by an adapted mind1. There is no doubt that, in patriarchal cultures, the female role is derived from the space left free by the male role, though still under patriarchal control. So we can read biological cues as cultural: “the only sign of femaleness is an absence of male cues”15. However, this does not contradict that what culture has expressed, strengthened, sedimented, socially stratified, and handed down through cultural products and memes may have evolved from cognitive processes that have guaranteed human survival31–35. In case of ambiguity or complexity in the detection of sex cues, a cognitive bias has saved humans from a risky encounter with an aggressive male26,36,37. Cultural contents (e.g., phallocentrism, patriarchalism, androcentrism, etc.) cannot precede those psychological mechanisms that had produced them. This is not to say that culture only echoes mental contents. Alexander38 summarizes this point well as follows: “I have not suggested that culture precisely tracks the interests of the genes—obviously this is not true—but that, in historical terms, it does so much more closely than we might have imagined” (p. 142).
Once produced, the culture constitutes part of that environment within which the mechanisms of natural selection evolve. And since not only does the environment select the individual, but the individual modifies the environment, culture cannot completely introduce content that goes beyond the boundaries of those cognitive constraints within which all variations are possible and learnable. Otherwise, the content would be unlearnable, inexperienced and, therefore, without effect on the individual behavior and the phenotypic evolution.
How could we have evolved without developing a computational cognitive system specific enough to solve the problem of sex attribution in an infinite combinatorial variety of phenotypic and genetic variables? Just by combining six variables of primary and secondary male and female sexual characteristics and two clothes, we produced 120 stimuli. Combinatorial explosion refers to the fact that with each new degree of freedom added, or dimension, or choice added, the total number of alternative possibilities quickly explodes. If only we were to combine the SGAT-A variables with other well-known variables that influence sex recognition—such as tone of voice, body posture and gait, social status, and so on39—the alternatives would soon multiply endlessly. Yet, each of us solves this “frame problem” without great difficulty or cognitive effort, in a matter of milliseconds6. This makes us lean toward a domain-specific mechanism that cooperates and competes with other evolved mechanisms of our adapted mind that have evolved for adaptive success and ensure good fitness in Pleistocene.