Content deleted Content added
General clean up from beginning to "Scientific Investigation" Moved section of "critics of Levater" up to appear directly below the section describing his work. |
Added substantive section on recent research and section on 3 theoretical pathways linking facial appearance with psychological traits |
||
Line 120:
== Scientific investigation ==
Due to its legacy of racism and junk science masquerading as criminology, the scientific study or even discussing the relationship between facial features and character have become taboo. However, science has uncovered many links between character and facial appearance. For example, there is evidence that character can influence facial appearance<ref>"Dress for success: human facial expressions are important signals of emotions" Lomus, et al. Annales Zoologici Fennici 46(1): 79-80. 2009</ref>. Also, facial characteristics influences first impressions of others, which influences our expectations and behavior, which in turn influences character<ref>"The relationship between appearance and personality across life span" Zebrowitz et al. 1998</ref>. Lastly, there are several biological factors that influence both facial appearance and character traits, such pre- and post-natal hormone levels<ref>"Stability of women's facial shape throughout the menstrual cycle" Marcinkowska and Holzleitner Proceedings. Biological Sciences. 2020 </ref> and gene expression<ref>"Diagnostically relevant facial gestalt information from ordinary photos" Ferry, et al. 2014</ref>.
Research in the 1990s indicated that three elements of personality in particular – power, warmth and honesty – can be reliably inferred.{{clarify|reason=From what features?|date=May 2022}}<ref name="psych">Chamorro-Premuzic, Tomas and Adrian Furnham. ''The Psychology of Personnel Selection''. 1st ed. Cambridge: Cambridge University Press, 2010. Cambridge Books Online. Web. 07 April 2016. {{doi|10.1017/CBO9780511819308}} pages 13-14</ref>▼
Recent progress in AI and computer vision has been largely driven by the widespread adoption of deep neural networks (DNN), which mimics mimic the neocortex by simulating large, multi-level networks of interconnected neurons. DNNs excel at recognizing patterns in large unstructured data such as digital images, sound, or text, and analyzing such patterns to make predictions<ref>"Deep neural networks are more accurate than humans at detecting sexual orientation from facial images" Wang and Kosinski Journal of Personality and Social Psychology. Vol 114: 246-257. 2018</ref>. The superior performance of DNNs offers an opportunity to identify links between characteristics and facial features that might be missed or misinterpreted by the human brain.<ref>"Deep neural networks are more accurate than humans at detecting sexual orientation from facial images" Wang and Kosinski Journal of Personality and Social Psychology. Vol 114: 246-257. 2018</ref>
The relationship between facial features and character traits such as political or sexual orientation is complex, but involves the fact that facial features can shape social behavior, partially as a result of the self-fulfilling prophecy effect<ref>Robert Merton. "The Unanticipated Consequences of Purposive Social Action" American Sociological Review. 1(6): 894-904. 1936</ref>. The self-fulfilling prophesy effect asserts that people perceived to have a certain attribute will be treated accordingly, and over time may engage in behaviors consistent with others’ expectations of them<ref>"Internalized Impressions: The Link Between Apparent Facial Trustworthiness and Deceptive Behavior Is Mediated by Targets' Expectations of How They Will Be Judged" Slepian and Ames. Psychological Science. 27(2): 282-288. 2016</ref>. Conversely, social behavior such as addictions to drugs or alcohol, can shape facial features.
▲Research in the 1990s indicated that three elements of personality in particular – power, warmth and honesty – can be reliably inferred
Some evidence indicated that the pattern of whorls in the scalp had some correlation to male homosexuality,<ref>{{Citation
Line 133 ⟶ 139:
A February 2009 article in ''[[New Scientist]]'' magazine reported that physiognomy is living a small revival, with research papers trying to find links between personality traits and facial traits.<ref name="newscientist2"/> A study of 90 ice hockey players found a statistically significant correlation between a wider face—a greater than average cheekbone-to-cheekbone distance relative to the distance between brow and upper lip—and the number of penalty minutes a player received for violent acts like slashing, elbowing, checking from behind, and fighting.<ref>{{Cite journal|last1=Carré|first1=Justin M.|last2=McCormick|first2=Cheryl M.|date=2008-11-22|title=In your face: facial metrics predict aggressive behaviour in the laboratory and in varsity and professional hockey players|journal=Proceedings of the Royal Society of London B: Biological Sciences|language=en|volume=275|issue=1651|pages=2651–2656|doi=10.1098/rspb.2008.0873|issn=0962-8452|pmc=2570531|pmid=18713717}}</ref>
This revival has
In 2017, a controversial study claimed that an AI algorithm could detect sexual orientation 'more accurately than humans' (in 81% of the tested cases for men and 71% for women)<ref>"Deep neural networks are more accurate than humans at detecting sexual orientation from facial images" Wang and Kosinski. Journal of Personality and Social Psychology. 114:246-257. 2018</ref>. A director of research of the [[Human Rights Campaign]] (HRC) accused the study of being "junk science" to the [[BBC]].<ref>{{cite news|title=Row over AI that 'identifies gay faces'|url=https://www.bbc.co.uk/news/amp/technology-41188560|access-date=14 January 2018|work=BBC News|date=11 September 2017}}</ref> The director, an 'equity and inclusion strategist' with no scientific background, was criticized by the researchers for "premature judgement"
In 2020, a study on the use of consumer facial images for marketing research purposes concluded that deep learning on facial images can extract a variety of personal information relevant to marketers and so users' facial images could become a basis for [[Targeted advertising|ad targeting]] on [[Tinder (app)|Tinder]] and [[Facebook]].<ref>{{cite journal|last1=Tkachenko|first1=Yegor|last2=Jedidi|first2=Kamel|title=What personal information can a consumer facial image reveal? Implications for marketing ROI and consumer privacy|url=https://ssrn.com/abstract=3616470|website=SSRN|ssrn=3616470|access-date=17 June 2020|date=1 June 2020}}</ref> According to the study, while most of facial images' predictive power is attributable to basic demographics (age, gender, race) extracted from the face, [[Visual artifact|image artifacts]], observable facial characteristics, and other image features extracted by [[deep learning]] all contribute to prediction quality beyond demographics.
== Three Theoretical Pathways linking Facial Appearance with Psychological Traits ==
Kosinski (2023)<ref>"Theory of Mind May Have Spontaneously Emerged in Large Language Models" Michal Konsinski. preprint.</ref> argues that there are three theoretical pathways that could link facial appearance with psychological traits. The three pathways are:
▲In 2017, a controversial study claimed that an algorithm could detect sexual orientation 'more accurately than humans' (in 81% of the tested cases for men and 71% for women). A director of research of the [[Human Rights Campaign]] (HRC) accused the study of being "junk science" to the [[BBC]].<ref>{{cite news|title=Row over AI that 'identifies gay faces'|url=https://www.bbc.co.uk/news/amp/technology-41188560|access-date=14 January 2018|work=BBC News|date=11 September 2017}}</ref> The director, an 'equity and inclusion strategist' with no scientific background, was criticized by the researchers for "premature judgement". The researchers criticized [[GLAAD]] and HRC's press release for falsely stating the paper was not peer reviewed. In early 2018, researchers, among them two specialists of AI working at Google (one of the two on face recognition), issued a reportedly contradicting study based on a survey of 8,000 Americans using [[Amazon Mechanical Turk|Amazon's Mechanical Turk]] crowd-sourcing platform. The survey yielded many traits helping to discriminate between gay and straight respondents with a series of yes/no questions. These traits had actually less to do with morphology than with grooming, presentation, and lifestyle (makeup, facial hair, glasses, angle of pictures taken of self, etc.).<ref>{{cite web|last1=Agüera y Arcas|first1=Blaise|last2=Todorov|first2=Alexander|last3=Mitchell|first3=Margaret|title=Do algorithms reveal sexual orientation or just expose our stereotypes?|url=https://medium.com/@blaisea/do-algorithms-reveal-sexual-orientation-or-just-expose-our-stereotypes-d998fafdf477|website=Medium|access-date=14 January 2018|date=11 January 2018}}</ref> For more information of this sexual orientation issue in general, see [[gaydar]].
First, facial appearance can shape psychological traits (face→mind pathway). The self-fulfilling prophecy effect<ref>Robert Merton. "The Unanticipated Consequences of Purposive Social Action" American Sociological Review. 1(6): 894-904. 1936</ref> postulates that people perceived as having a particular attribute are treated accordingly; internalize such attributions; and, over time, may engage in behaviors consistent with others’ perceptions<ref>"Internalized Impressions: The Link Between Apparent Facial Trustworthiness and Deceptive Behavior Is Mediated by Targets' Expectations of How They Will Be Judged" Slepian and Ames. Psychological Science. 27(2): 282-288. 2016</ref>. For example, people with larger jaws are perceived as more socially dominant; and thus, with time, may become more so<ref>"Geometric morphometrics of male facial shape in relation to physical strength and perceived attractiveness, dominance, and masculinity" Windhager et al. American Journal of Human Biology. 23(6): 804-814. 2011</ref> 2011). Moreover, face-based perceptions influence consequential outcomes such as the length of prison sentences, occupational success, educational attainments, the chances of winning an election, income, and status<ref>"Predicting political elections from rapid and unreflective face judgments" Ballew and Todorov. Proceedings of the National Academy of Sciences. 104(46). 2007</ref<ref>"Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts Capital-Sentencing Outcomes" Eberhardt et al. Psychological Science. 17(5): 383-386. 2006</ref><ref>"Inferences of Competence from Faces Predict Election Outcomes" Todorov et al. Science. 2005 https://www.science.org/doi/10.1126/science.1110589</ref><ref>"Social Psychological Face Perception: Why Appearance Matters" Zebrowitz and Montpare. Social and Personality Psychology Compass. 2(3): 1497-1517. 2008.</ref>.
▲In 2020, a study on the use of consumer facial images for marketing research purposes concluded that deep learning on facial images can extract a variety of personal information relevant to marketers and so users' facial images could become a basis for [[Targeted advertising|ad targeting]] on [[Tinder (app)|Tinder]] and [[Facebook]].<ref>{{cite journal|last1=Tkachenko|first1=Yegor|last2=Jedidi|first2=Kamel|title=What personal information can a consumer facial image reveal? Implications for marketing ROI and consumer privacy|url=https://ssrn.com/abstract=3616470|website=SSRN|ssrn=3616470|access-date=17 June 2020|date=1 June 2020}}</ref> According to the study, while most of facial images' predictive power is attributable to basic demographics (age, gender, race) extracted from the face, [[Visual artifact|image artifacts]], observable facial characteristics, and other image features extracted by [[deep learning]] all contribute to prediction quality beyond demographics.{{clarify|reason=What does "beyond demographics mean, if anything?|date=May 2022}}
Second, there are latent factors shaping both psychological traits and facial appearance (face←factor→mind pathway). Those include socioeconomic status, environmental and developmental conditions, hormones, genes, etc. Twin studies, for example, have found that genes are responsible for over 50% of the variation in both facial features<ref>"Facial Genetics: A Brief Overview" Richmond et al. Frontiers in Genetics. Vol 9. 2018</ref> and political orientation<ref>"Are Political Orientations Genetically Transmitted?" Alford, Funk and Hibbing. American Political Science Review. 99(2): 153-167.2005.</ref>. Similarly, prenatal and postnatal testosterone affect jaw shape and are linked—to a small degree—with character traits closely related to political orientation, such as conscientiousness<ref>"Facial Width-to-Height Ratio Does Not Predict Self-Reported Behavioral Tendencies" Michal Kosinski. 2017</ref>. Furthermore, prenatal exposure to nicotine and alcohol affects facial morphology<ref>"Facial Genetics: A Brief Overview" Richmond et al. Frontiers in Genetics. Vol 9. 2018</ref> and cognitive ability, which is associated with political orientation<ref>"The Association of Cognitive Ability with Right-wing Ideological Attitudes and Prejudice: A Meta-analytic Review" Onraet et al. European Journal of Personality. 29(6): 599-621. 2015</ref>
Third, psychological traits can shape facial appearance (mind→face pathway). While we tend to think of facial features as relatively fixed, they are shaped by factors such as facial care, diet, substance use, physical health, injuries, exposure to sunlight, harsh environmental conditions, or emotional states<ref>>"Facial Genetics: A Brief Overview" Richmond et al. Frontiers in Genetics. Vol 9. 2018</ref>. Exposure to such face-altering factors, in turn, is associated with psychological traits. Liberals, for example, tend to smile more intensely and genuinely<ref>"Conservatives report, but liberals display, greater happiness" Wojcik et al. Science. 2015. https://www.science.org/doi/10.1126/science.1260817</ref>—which, over time, leaves traces in wrinkle patterns<ref>"From skin microrelief to wrinkles. An area ripe for investigation" Piérard, Uhoda, and Claudine Pié. Journal of Cosmetic Dermatology. 2(1): 21-28. 2003</ref>Conservatives tend to be more self-disciplined and are thus healthier, consume less alcohol and tobacco, and have a better diet<ref>"Are republicans healthier than democrats?" Subramanian and Perkins. International Journal of Epidemiology. 39(3): 930-931. 2010.</ref>, altering their facial fat distribution and skin health<ref>"Facial Genetics: A Brief Overview" Richmond et al. Frontiers in Genetics. Vol 9. 2018</ref>
== In media ==
| |||