It’s always a little exciting when your research area shows up in the news, especially when your work inclines to the obscure. This is even the case when the thing you study is about something better left in the past – no one should want to bring back phrenology into the present day. And yet, every few months, a scientific study hits the news and the cycle begins anew as the world “rediscovers” or “reinvents” phrenology, the nineteenth-century science of judging character and potential based on the shape of the skull.
The most recent case of the reinvention of phrenology appeared just weeks ago in Nature Communications, in an article entitled “Tracking historical changes in trustworthiness using machine learning analyses of facial cues in paintings.” The authors, Lou Safra, Coralie Chevallier, Julie Grèzes, and Nicolas Baumard, a group of neuroscientists and social scientists, examined European portraits produced between 1500–2000, applying an algorithm they developed that “estimates trustworthiness based on a pre-identified set of facial characteristics.” Through the application of this algorithm to thousands of Western European portraits, they found that “trustworthiness displays in portraits increased throughout history,” and further that “trustworthiness displays increased with affluence,” with affluence preceding changes in trustworthiness. The authors tie this supposed rise in trustworthiness to the “decline of interpersonal violence and the rise of democratic values observed in Western Europe,” as well as to rising standards of living.
Being neither a European historian, an art historian, or an expert in machine-learning, much of this study is beyond my ability to critique. But as a historian who studies phrenology, the findings are eyebrow-raising, particularly the notion of developing an algorithm that can determine such a quality as “trustworthiness” based on works of art – i.e. reducing the assessment of a subjective quality, based on evidence in an inherently subjective format, to something quantifiable and identifiable by algorithm.
What drew my attention – and what, arguably, drew the attention of many commentators to this essay – were the tweets issued by one of the authors, Nicolas Baumard, to promote this publication. One image and tweet in particular, since removed, received the bulk of the attention. This image prompted many scholars, social commentators, and the broader public to retweet this essay, framing it in particular as evidence of racial (and racist) science, physiognomy, and phrenology and tying its imagery to the work of Johann Kaspar Lavater (an eighteenth-century Swiss physiognomist), Cesare Lombroso (a late-nineteenth-century Italian criminal anthropologist), and others. Many tweets used images from an infamous phrenological work, L. A. Vaught’s Practical Character Reader (1902), known for its rich and humorous images depicting such aspects of the head and character as “an unreliable husband” or “a deceitful mouth.”
While on the one hand I was pleased to see so many commentators connecting the conclusions of this study to phrenology, physiognomy, and related sciences, this leads to its own problems. Most modern-day commentators (and many historians) continue to frame phrenology primarily as a “pseudo-science,” a term often used to dismiss “bad” science. But in so doing, they neglect the long-term influence of phrenology and related “failed” sciences. By dismissing this study as yet another example of phrenology, without simultaneously recognizing the historic scientific validity, utility, and pervasive nature of the science, commentators highlight its superficial aspects while enabling the continued influence of the underlying principles. This study demonstrates the continued influence of phrenological thinking, language, and imagery in modern culture – both popular and scientific.
In particular, as I often remind my students, images are arguments. Whether these co-authors intended it or not, the centrality of the above image to the argument in this paper, its promotion on social media, and the ease with which this image can be connected with phrenology is no accident. This image, and the argument it conveys about “good” and “bad” heads, replicates phrenological assessments from the nineteenth century.
One common refrain in the Twittersphere was that this study was evidence of physiognomy, the practice of judging character based on facial features dating to the Ancient world, not of phrenology. In point of fact, it was both. First, phrenology and physiognomy often overlapped in the nineteenth century; phrenologists included physiognomic ideas and imagery in their text. One of the founders of the field, Johann Spurzheim, even entitled one of his first English-language texts, The Physiognomical System of Drs. Gall and Spurzheim (1815). Phrenological texts adopted and adapted many physiognomic tenets and also borrowed concepts from other related fields, particularly racial science. One example is the adoption of the “facial angle,” a line sketched through the profile of the face, from forehead to chin. Originally introduced in the eighteenth century by Petrus Camper, a Dutch naturalist, the angle of this line came to be used as shorthand for racialized hierarchies of intelligence, civilization, beauty, and other qualities in nineteenth-century racial science.
But further, phrenology often spoke not just to the nature of individual organs, but about groups of organs as they affected the shape of the head and hence the character of the individual. Single organs thus sometimes coded for an entire group of associated organs, and the over- or under-development of this organ (and its associated region) therefore assumed outsized significance. One example of this dynamic is how the organ of Destructiveness, which was intended to indicate desire to destroy in general, came to signify more specifically murderous and other violent propensities, and second, how its size stood in for the entire region of the so-called animal organs, which broadly circle the ear. Over the course of the nineteenth century, this single organ and its effect on the character was used to signify “bad heads,” particularly those of criminals. But the enlargement of this organ and region was also visually apparent, particularly in images like the one below, which suggested how this single organ could shape the entire head and character.
Large and small Destructiveness, in images like these, helped average Americans to construct a visual language of good and bad heads, a language I argue is still with us. Not only does the Nature Communications essay replicate this imagery and its association with good and bad heads, but it is hardly the first to do so. Wide heads, particularly those narrow in the crown and wide over the region of the temporal bone, have been pointed to in studies over the last decade as predicting a variety of bad behaviors. These studies, focusing on facial height-to-width ratios (fWHR), have often included images similar to that found in Nature Communications, with a chipper, narrow “good” face compared with an angry, wide “bad” face. The Nature Communications essay, therefore, is only the most recent manifestation of a long tradition of constructing good and bad heads, whether through physiognomy or phrenology or these most recent findings in psychology and other fields.
The lingering question is, of course, what is the origin of such assumptions of the relative meanings of wide and narrow heads? If phrenology (and physiognomy) promoted an idea of “good” and “bad” heads focused in particular on the width of the head, and head width is linked in the present to undesirable traits – lack of trustworthiness, aggression, etc. – perhaps such studies in fact demonstrate not the correlation between head width and “badness,” but the influence of phrenological thinking itself. Did phrenology create such a successful composite image of the “bad head” that it is still structuring how we judge faces and fates today? If so, then phrenology and allied nineteenth-century sciences of the self and skull are still shaping how we make social – and scientific – judgments of character and potential.
In my forthcoming book, I close with a discussion of “phrenological futures,” which touches on studies like those I’ve discussed above. I conclude by observing that we have never truly left behind phrenology, that its concepts and constructs continue to shape our ways of looking at the world and knowing others. Perhaps, more than simply a question of futurity, I ought to have considered the present as more firmly grounded in phrenological assumptions. If scientists, theorists, and “tech-bros” keep reinventing phrenology, intentionally or not, is it possible that we have never abandoned phrenology? Is abandoning phrenology even possible? As long as these assumptions about “good” and “bad” head shapes keep being reinvented and rediscovered, the cycle will continue – leaving us in an unending phrenological now.
Carla Bittel, “Testing the Truth of Phrenology: Knowledge Experiments in Antebellum American Cultures of Science and Health,” Medical History 63, no. 3 (2019): 352–374.
Susan Branson, “Phrenology and the Science of Race in Antebellum America,” Early American Studies 15, no. 1 (2017): 164–193.
Sharrona Pearl, About Faces: Physiognomy in Nineteenth-Century Britain (Harvard University Press, 2010).
James Poskett, Materials of the Mind: Phrenology, Race, and the Global History of Science, 1815–1920 (University of Chicago Press, 2019).
Courtney E. Thompson, An Organ of Murder: Crime, Violence, and Phrenology (Rutgers University Press, 2021).
Daniel Patrick Thurs, Science Talk: Changing Notions of Science in American Popular Culture (Rutgers University Press, 2007).
- See, for example, studies that link wide heads and faces to aggression and other undesirable or unethical behaviors: Michael P. Haselhuhn and Elaine M. Wong, “Bad to the Bone: Facial Structure Predicts Unethical Behavior,” Proceedings of the Royal Society B 279, no. 1728 (2012): 571–576; Justin M. Carré, Cheryl M. McCormick, and Catherine J. Mondloch, “Facial Structure Is a Reliable Cue for Aggressive Behavior,” Psychological Science 20, no. 10 (2009): 1194–1198; Justin M. Carré and Cheryl M. McCormick, “In Your Face: Facial Metrics Predict Aggressive Behaviour in the Laboratory and in Varsity and Professional Hockey Players,” Proceedings of the Royal Society B: Biological Sciences 275, no. 1651 (2008): 2651–2656; and Michael P. Haselhuhn, Margaret E. Ormiston, and Elaine M. Wong, “Men’s Facial Width-to-Height Ratio Predicts Aggression: A Meta-Analysis,” PLOS One 10, no. 4 (2015). ↑