Facial recognition, values, and the human brain
|Image courtesy of Pixabay.
Research is not an isolated activity. It takes place in a social context, sometimes influenced by value assumptions and sometimes accompanied by social and ethical implications. A recent example of this complex interplay is an article, “Deep neural networks can detect sexual orientation from faces” by Yilun Wang and Michal Kosinski, accepted in 2017 for publication in the Journal of Personality and Social Psychology.
The abstract of the article ends with the sentences (p.2): “Those findings advance our understanding of the origins of sexual orientation and the limits of human perception. Additionally, given that companies and governments are increasingly using computer vision algorithms to detect people’s intimate traits, our findings expose a threat to the privacy and safety of gay men and women.”
|Image courtesy of Flickr.
Apart from the above-mentioned research ethics issues, ethical aspects matter in two respects: first with regard to the value assumptions implicit in the study design and second with regard to possible ethical implications of the research. Physiognomy, the broader context in which the study is located, is a controversial field that many consider to be a pseudoscience (Emspak 2017).
Physiognomy assumes that a person’s facial features give indications of his/her personality traits. It is not by chance that the study reminds me of the pseudo-scientific phrenological approaches of the 19th century that attempted to draw conclusions about individuals’ personality traits based on the shape of their skulls (Holtzman 2015). What unites these two is that their approach is influenced by social value assumptions and the motivation to be able to identify individuals with behavior or with characteristics considered socially deviant.
Other brain-related research fields are not immune to social value assumptions either. An example is craniometry and the highly questionable claim made by Samuel George Morton in the 19th century that differences in cranial capacity between different ethnic groups are indicators of the intellectual capacity of these ethnic groups. The same applies for discussions on the relevance of brain size for intelligence (Fausto-Sterling 1993). These examples show us more about the underlying social assumptions of the researchers than about the actual relevance of their measurements.
One of the central conclusions of the study is that human faces “contain more information about sexual orientation than can be perceived or interpreted by the human brain” (Wang & Kosinski, p. 29). Deep neural networks are reported to provide more accurate results in the described study setting because they take features into consideration that are not accessible or not relevant for humans and the human brain when it comes to distinguishing between heterosexual and homosexual individuals based on their faces. Nevertheless, it is obvious that the resulting data needs human interpretation, especially in view of the intimacy of the trait under investigation. For example, the authors explain the higher probability of seeing a shadow on the forefront of heterosexual men and lesbian women in the study by the tendency of both groups to wear baseball caps and “the association between baseball caps and masculinity in American culture” (p. 20). In other cultural contexts, different influencing factors may be expected. But, it remains unexplained as to why there is a higher probability of gay people wearing glasses in the study (Emspak 2017).
|Image courtesy of Pixabay.
The underlying question is: how can we ever adequately interpret the resulting data in a situation in which not only a considerable number of the elements used by the system, but also their relevance escape our understanding? There is a clear risk for discrimination against homosexual men and women based on opaque algorithms (O’Neil 2016; Agüera y Arcas et al. 2018). This leads to the question of possible negative consequences for homosexual men and women.
Want to cite this post?
Hildt, E. (2018). Facial recognition, values, and the human brain. The Neuroethics Blog. Retrieved on , from http://www.theneuroethicsblog.com/2018/06/facial-recognition-values-and-human.html