As this brilliant thread makes clear, that evopsych paper tracks "trustworthiness displays" in a database of portraits of WHITE Euro elites, using an algorithm engineered to detect _contemporary bias_ in the perception of character of WHITE people...
https://twitter.com/yet_so_far/status/1309475976376725504
In other words, the study is not only projecting presentist bias towards WHITE displays of "trustworthiness" (wtf that means) onto the past, it's doing so under the cover that these biases are natural & universal, a conclusion that studies the paper cites don't even support...
Thus, one might conclude from the study & graphs like this that greater GDP leads to more trustworthiness displays, but what we're really seeing is correlation btwn rise of colonizing states' GDP & presentist bias towards the appearance of white-coded trustworthiness/dominance... 
Among the other many problems with the study is that the authors seem unaware of the fact that museum collections are themselves biased and selective, and that the NPG--from which they draw one dataset--was founded in 1856, during the UK's rise as a global colonial power.
So the NPG dataset isn't a random selection of portraits produced in the UK btwn 1500-2000, it's a dataset filtered according to 19th-20th-c. Anglo-white standards and preferences. No wonder they found a correlation with contemporary biases towards Anglo-white faces!
Especially disturbing, as @yet_so_far points out, is that the study makes no attempt to distinguish btwn _bias_ towards "trustworthiness displays" & trustworthiness writ large. So one may be led to believe that (biased, acculturated) displays of trustworthiness = trustworthiness.
Yet Alexander Todorov et al., authors of one of the articles our study cite, express skepticism that the (again, biased) detection of facial cues may somehow predict distinct character traits. See, e.g., this passage from p536 of Todorov et al.: d1wqtxts1xzle7.cloudfront.net/40252345/Todor…
But we see no such discussion or acknowledgement in the present study. We should thus ask why this is the case, and why the study mined such a limited and highly filtered dataset of objects belonging to a colonial institution formed in the 19th century?
Why do other studies that it cites, like this gem, consistently look to Renaissance Europe as the birthplace for "cognitive superiority" (also note the reference to "cultural selection"!)?
What these studies _don't_ do is run the same algorithm on portraiture from places outside of Europe. I wonder why?
• • •
Missing some Tweet in this thread? You can try to
force a refresh
