Towards a Method for Discerning Sources of Supply within the Human Remains Trade via Patterns of Visual Dissimilarity and Computer Vision

The human remains trade, once an obscure market, has become increasingly common due to the Internet and social media platforms. The legality of this trade varies at the local and state level; however, Indigenous North American remains are federally protected under the Native American Graves Protection and Repatriation Act.

As part of the sales process, vendors will make claims about the remains’ archaeological origin and ownership history, but are very careful to say that they would never knowingly trade in skulls from Indigenous groups. Without physically examining the remains, though, these claims cannot be verified. A single photograph is the only evidence of a life lived.

In this report, Huffer and Graham have developed a novel image analysis using convolutional neural networks to predict the broad geographic ancestry of a skull on Instagram, in order to better understand which peoples’ remains are being bought and sold.

According to the results, over two thirds of the ancestries claimed by vendors were dubious. Not only that, it seems that skulls with Indigenous North American ancestry are circulating in this market far more than vendors either know or let on. This mismatch between what the vendors say about the skulls and what the model predicts lends weight to the idea that at least some collectors are misrepresenting the origins and life histories of human remains. While no predictive model will be as accurate as physically measuring a skull, the results of this initial experiment are promising.