Political Visual Literacy Open Access

Abstract

Visual politics is a fast-growing field and much of it is focused on images that inspire criticism. This tendency results in a lack of attention to oppressive visual practices. A political visual literacy approaches all visual practices as being layered with different “visual truths” that were developed in response to political commitments over time. These “visual truths” inflected visual practices in ways that may outlive the political settings in which they were first articulated. Most important of these is the desire to capture human difference that framed the development of visual technologies and is still embedded in a range of visual practices. The methodology I develop here links a conjunctural analysis of visual tools and practices and the visual truths implicated in them with their operationalisation by actors as cultural tools through the framework of mediated action. I develop this approach by interrogating two layered and harmful visual practices: the White-centrism of visual technologies and the racialised origins of transphobic visualities in automatic gender recognition technology.

La culture visuelle politique envisage les visuels et visualités telles des couches d'objets et de pratiques qui contiennent différentes techniques pour atteindre des «vérités visuelles». Souvent, ces «vérités visuelles» servent des idéologies opprimantes; en premier lieu, la racialisation, pourtant largement absente de la politique visuelle. En employant une culture visuelle politique, nous reconnaissons les «vérités visuelles» néfastes qui encadrent ces pratiques et technologies quand elles se développent, mais aussi leur persistance dans les pratiques visuelles contemporaines. Cet article défend la nécessité d'une telle culture visuelle politique en politique visuelle et présente une méthodologie qui interroge deux aspects des implications de la race et de la pratique visuelle: l'ethnocentrisme blanc des technologies visuelles et l'origine racialisée des visualités transphobes dans la technologie de reconnaissance automatique du genre (RAG).

La alfabetización visual en el campo de la política considera las imágenes y las visualidades como objetos y prácticas formados por diversas capas, en las que se han utilizado diferentes técnicas con el fin de conseguir «verdades visuales». A menudo, estas «verdades visuales» están al servicio de ideologías de carácter opresivo, sobre todo de la racialización, las cuales están, en su mayoría, ausentes de la política visual. Reconoceremos, mediante el uso de la alfabetización visual en el campo de la política, las «verdades visuales» dañinas que contextualizaron estas prácticas y tecnologías a medida que se fueron desarrollando, así como su presencia duradera en la práctica visual contemporánea. Este artículo defiende la necesidad de esta alfabetización visual política dentro de la política visual y desarrolla una metodología que cuestiona dos aspectos de la implicación de la raza y de la práctica visual: el centrismo blanco de las tecnologías visuales y el origen racializado de las visualidades tránsfobas en la tecnología de reconocimiento automático de género (AGR, por sus siglas en inglés).

“Photography was developed as a tool to capture visually and classify human difference.”

Ruha Benjamin

The idea for a political visual literacy started in the seminar rooms of a final-year visual politics course that I taught over the past three years. The student class in our London-based department had changed dramatically. Five years ago, about three-quarters of our students were White, but our current cohorts are majority Black and global majority. In the seminars, we discovered that the traditional representational explication: essays inspired by a visual artifact that explore wider connotational associations, often bypass or rush through deeply engrained structures of oppression that many of our students wanted to focus on. We also found that insisting on the interrogation of visual practices and the political implications of visual methods and technologies tends to ground analysis in precisely these “missing” politics.

These pedagogical reflections carry to the field of visual politics, where visual analysis is often flattened to images that inspire criticism. At the same time, the discursive practices that frame the production of visuals are approached as contextual backgrounds or theoretical discoveries, rather than integral to these artifacts. Priya Dixit’s critique that the field ignores “the relationships between those who visualize and those who are visualized” (Dixit 2014, 338) still stands. For Dixit, the answer is to embrace a “postcolonial visuality” and to engage in “visual decolonization.” I fully endorse this approach, which echoes those of Sealy (2019) and Mirzoeff (2011) among others. Additionally, I argue that any engagement that takes seriously visual discourses, practices, technologies, and their interaction will quickly be asking questions about power and oppression.

This article develops a “political visual literacy,” an approach that interrogates the ways in which political “visual truths” develop and enfold into practices and technologies of visualization. To make the case for a political visual literacy, I first survey existing approaches and align visual politics with visual literacy to argue that it is the appropriate next step. Second, I employ a conjunctural analysis that builds on the concepts of “aesthetic technologies” (Dyer 2017) and demonstrate its utility for political visual literacy by applying it onto contemporary technologies and practices around photography and racialization. Third, I link the focus on visual discourses to their operationalization by practitioners with an additional framework of "mediated action" (Wertsch 1998). Lastly, I deploy the full methodology to explore a case study of a transphobic visuality manifest in automatic gender recognition (AGR) technology and demonstrate the long and ignominious trajectory of visual truths of visual essentialization.

Visual Politics and Visual Literacy

Visual politics is doing well. Grafting from security studies, the new inter-sub-discipline easily maps into existing structures of research and teaching in and around international relations and its “visual turn.” Visual Politics’ project of “open[ing] up debates as widely as possible” (Bleiker 2018, 4) makes it a vibrant field that is difficult to plot. Yet, as Dean Cooper-Cunningham (2020) noted, research tends to align with two approaches: the first focuses on the analysis of visual artifacts, and the second uses visual methods to analyze political life. In other words, “visual reading” and “visual writing.”

Visual reading is a response to the proliferation of visual artifacts that forms the "visual turn" (Bleiker 2018). Scholars have examined how images participate in what Stuart Hall (1997, 232) termed “regimes of representation” in different contexts and aligned such visual practices to other theoretical approaches to explain the political work that visual artifacts do. For example, Bleiker and Kay (2007) looked at different strategies of representation of HIV/AIDS, while both Hansen (2011, 2015) and Schlag and Heck (2013) extended representation toward securitization theory. Van Veeren (2011), Manor and Crilley (2018), and Constantinou (2018) wrote about representation as a diplomatic instrument, while Campbell (2007) drew on Deborah Poole to link representation to a “visual economy.”

Visual writing is a response to the proliferation of advanced visual capabilities that often hide behind simplified graphic interfaces (Drucker 2014). In recent years, visual politics scholars have used creative methods to “write visually.” Until recently, this approach was deemed less common (Vuori and Andersen 2018, 13), but this may no longer be the case. Visual politics scholars now compose photo-essays (Lisle and Johnson 2019; Hansen and Spanner 2021), engage in visual auto-ethnographies (Bleiker 2019), build collages (Särmä 2018), engage in participatory photography (Nyman 2021, 323), and research films (Callahan 2020; Weber 2011; Harman 2019). This is very much an active area of research that will greatly expand if we are to judge by conference papers and research grants.

However, Andersen et al. (2015) suggest that these two artifact-based approaches are only half the story. Visual reading and writing concern “visuals,” but then, there are also “visualities,” which are “techniques of knowledge” that determine how vision itself is constructed and the political implications of that construction. I refer here to what they term “visuality as a modality” or politically charged ways of seeing as a “discursive practice” (Mirzoeff 2011, 80) that denote a regime of visual truth (Mitchell 1992). These concerns echo Allen Feldman’s definition of “scopic regimes” as “political visualization” that asserts itself through “an ensemble of practices and discourses that establish the truth claims, typicality, and credibility of visual acts and objects and politically correct modes of seeing” (Feldman 1997, 30).

The three strategies of visual politics noted above (reading, writing, and visualities) are congruent with Kędra’s (2018) definition of “visual literacy,” which is composed of “visual reading” and “visual writing” as well as “visual thinking.” Visual literacy is a pedagogically centered field concerned with helping users to develop the competencies they need to engage fully in our visually saturated world. This project is legitimized by an axiom that would be familiar to students of visual politics, according to which images are fast becoming the “lingua franca of our daily communication” (Kędra and Žakevičiūtė 2019).

Political visual literacy suggests that the key to achieving “visual thinking” in political analysis is to move beyond representational analysis and to explore the political implications of the visualities that are enfolded into visual practices. To be sure, such work is underway. Visualities and their implication in practices, mostly in relation to security, have been approached as scopic regimes (Campbell 2007; Tidy 2017; Grayson and Mawdsley 2019) as well as different kinds of gazes, often in the context of legitimizing targeting and killing (Wilcox 2017; Stahl 2018; Galai 2019). However, visual politics is marked by the neglect of the most prominent visuality of modernity: race and racialization.

Robert Miles (1989, 75) defines racialization as instances where “social relations between people have been structured by the signification of human biological characteristics in such a way as to define and construct differentiated social collectivities” and these significations are often visual and always implicated with visualities. Kalpana Seshadri-Crooks (2000, 2) explained that “Race is fundamentally a regime of looking, although race cannot be reduced to the look,” while Ruha Benjamin (2019, 68) powerfully asserted that the most prominent visual technology (photography) was developed as “a tool to capture visually and classify human difference.”

The visual aspect of racialization is the ground for much scholarship outside of visual politics. For example, in visual culture studies, Nicholas Mirzoeff (2011) explores “visuality complexes” as oppressive techniques of control that often developed along racial lines and calls on scholars of visuality to explore counter-visualities to these prevailing ways of seeing. In surveillance studies, Simone Browne (2015, 11) extended the Foucauldian panoptic metaphor at the heart of the field and traced it back to transatlantic slavery and forward to anti-Black surveillance, while in photography studies, Amos Morris-Reich (2016) retraces the visual truths that support racialized world views. Yet, visual politics has refrained from any sustained engagement with this paradigmatic concern. This article argues that studying the project of the visual capture and classification of human difference (to paraphrase Ruha Benjamin) must be central to visual politics. Political visual literacy allows us to interrogate the ways in which visualities and their visual truths, which very often signify perceived markers of human difference across raced and gendered lines developed and interacted with practices and technologies of visualization.

This idea builds on existing approaches in visual politics that choose to engage with visual production as more than a straightforward matter of representation. Van Veeren’s (2019) reading of the collage work of Wangechi Mutu and Njideka Akunyili Crosby is based on a “wandering” that follows the layered logics of their construction and thinks through them. Relatedly, the idea of “collaging” has entered visual politics as a method of “visual writing” in the project led by Särmä (2018, as well as by Kangas et al. 2019). It thinks through the act of collaging as a uniquely queer epistemological approach that involves "cutting and ripping apart materials, and then repairing this damaged knowledge by gluing and pasting the materials back into a new kind of a composition” (Kangas et al., 9). The goal of such collaging is not to explicate different meanings but rather to engage in a form of parataxis and to present a renewed intellectual “landscape” to “wander” through both “composition and layering” (ibid., 5).

Drawing on these efforts, I argue that students of visual politics need to pay attention not only to regimes of representation, but also to the visual tools, practices, and visualities that are employed in the production of visual artifacts and visual practices. This work is made possible if we treat visuals and visualities as layered and sedimented with ideologically laden responses to the different political contexts in which they developed. Technological innovation or cultural changes in our perception are connected, gradual, and complex, and each change leaves the indelible marks of the visual truths it claimed to demonstrate or deliver.

Such a project may seem unwieldy, especially with instruments of visualization getting increasingly elaborate. However, “doing” political visual literacy should not be any more about a technical teardown than a theoretical work is. In fact, this project is congruent with existing approaches in the social sciences. First, to a genealogy that comprises “a history that highlights the contingent origins and often ignoble outcomes of critical responses to a question” (West 1990, 94–95) or a conjunctural analysis (Gilbert 2019) that interrogates how techniques and technologies that were developed in one political setting carried its conventions and visual truths to another. Second, it echoes, simply, a methodology, which Van Rythoven and Sucharov (2019, 3) define as “a reflection on the core wagers-the choices made prior to the conduct of empirical research that subsequently structure research design.” The “wagers” are the claims for visual truth that are implicated with political projects.

In summary, political visual literacy approaches visuals and visualities as layered artifacts and practices that carry different claims for visual truths, which require a mode of explication that aligns with their layered construction. Van Veeren’s method of “wandering” (2019) is one possibility for reading and collaging (Särmä 2018; Kangas et al. 2019) is a method for visual writing; however, I suggest an alternative that interrogates visual reading, visual writing, and visualities in relation to the visual truths that are enfolded into them. In the next three sections, I will demonstrate a methodology for political visual literacy by focusing on two visual truths that implicate visual technologies with racism and transphobia. First, I will deploy a conjunctural analysis to explore White-centrism in photographic technology. Second, I will elaborate the method further by introducing the level of negotiation between actor and tool. Lastly, I will employ it to interrogate a transphobic visuality in the employment of automatic gender recognition technology.

Political Visual Literacy Through Conjunctural Analysis and “Aesthetic Technologies”

What follows is a conjunctural analysis that focuses on the white gaze as a visual truth that is enfolded into contemporary photographic practices. This will demonstrate the potential of this approach to extend analysis beyond the representational. Gilbert (2019, 6) defines conjunctural analysis, after Stuart Hall, as “the analysis of convergent and divergent tendencies shaping the totality of power relations within a given social field during a particular period of time.” Crucially, a conjunctural analysis is “a practice, a process. . . a collective effort” (Grossberg 2019, 42). It seeks to capture the wider dynamism in which a historic moment is implicated, and it can only aspire to illuminate some of the complex factors that are at play.

While this approach often focuses on world orders and hegemonic structures (most recently Danewid 2022), Inch and Shepherd (2020) suggest that “conjunctural thinking” can be mobilized to analyze more limited and clearly defined sites. In the context of political visual literacy, a conjunctural analysis allows us to interrogate visual artifacts and practices and their visual truths, or the roles that social forces play in their articulation over time. Furthermore, this approach fits within International Political Sociology and its focus on intricate assemblages across different levels of analysis. More broadly, it addresses the ways in which visual practices employ oppressive visualities.

Richard Dyer demonstrated how the white face became the “touchstone” for the development of photographic technology (cameras, lenses, and lighting), and when commercial equipment became available, most importantly film stock, it was calibrated accordingly (Dyer 2017, 90). White-centrism was not always so embedded in the technology. Frederick Douglass used photography extensively. He strongly believed that the visual verisimilitude of the new technology would destroy a racist and dehumanizing anti-Black regime of representation (Wexler 2011, 148). However, once photography evolved from artisanal production to commercially available stock in the mid-nineteenth century, a white-centrism, and more so the prevalent “white gaze” (Yancy 2008, 2) that was “predicated upon fictive “race” categories that were believed to designate empirical differences” was enfolded into the technology.

Establishing the links between the political and the technological is a crucial move for a political visual literacy, and Dyer (2017, 83) establishes it by arguing that “all technologies are at once technical in the most limited sense (to do with their material properties and functioning) and also always social (economic, cultural, ideological).” He used the term “aesthetic technology” to demonstrate the implication of Whiteness in visual practice, both in the technical as well as habitual senses, and I adopt this concept as a lens that snaps conjunctural thinking into focus.

In the 1950s, this White-centric visual truth was extended through a technological transformation to commercially produced color stock. Lorna Roth (2009) explored the photographic “Shirley cards” that were distributed by Kodak to Film-labs for color calibration. These picture-cards featured an accurate color map as well as the face of a white model (nicknamed “Shirley”) for the purpose of lighting and processing calibration. Roth explained that “the global assumption of Whiteness” (ibid., 117) was engrained in the chemical recipe for film emulsions that were still manufactured to produce fidelity with white skin (ibid., 118).

If Black-and-White film consistently lacked in detail in non-white subjects, then color film introduced more pronounced chromatic variations. Syreeta McFadden (2014) wrote about her experiences of using color film:

By the 1990s, when I began taking pictures, I hated shooting brown skin on color film… it seemed the technology was stacked against me. I only knew, though I didn't understand why, that the lighter you were, the more likely it was that the camera — the film — got your likeness right.

Furthermore, the racialized aspect of photography extended beyond the film stock. It was transposed from practices (lighting styles) to technologies (film stock) back to practices (color calibration in printing) and later in digital rearticulations of the same processes (algorithms, filters, and artificial intelligence [AI] trained on White-centric data). The white-centrism of visual capture technologies extends much further. It is embedded in technologies like biometric machines, often used to adjudicate access, which routinely trigger a “failure to enrol” (FTE) for users with darker complexions (Pugliese 2005). White-centrism marks what Roth termed, following Franco Vaccari and quite similar to Dyer, a “technological unconscious” that is aligned to a naturalized whiteness (Roth 2009, 126).

The same “technological unconscious” or “aesthetic technology” is evident in Annie Leibovitz’s August 2022 portrait of US Supreme Court Justice Ketanji Brown Jackson, the first Black woman to serve in the US Supreme Court. Leibovitz, a highly celebrated portraitist, photographed Jackson in the Lincoln Memorial in Washington, DC using a fairly standard set-up, which combined ambient and artificial lighting. While Lincoln’s white statue in the background was perfectly lit by the ambient warm lighting, Jackson, the subject of the portrait, was darkly lit by an artificial light source with a much colder tone, while an additional close-up portrait with the reflecting pool in the background similarly underexposed and tinted Jackson’s face. This was a well-funded production assigned by a prestigious magazine (Vogue) that utilized the expertise of multiple professionals who handled the lighting and retouching of the digital frame. Yet, the result echoed the aesthetic technology of earlier analog technologies and their associated practices.

As might be expected, the response to these images was overwhelmingly negative. Leibovitz published a “sneak-peak” on Twitter and “critics—consisting mostly of Black women—quickly noted how poor the lighting was and how it wasn't properly suited for Brown's darker skin tone” (McDuffie 2022).¹ The “aesthetic technology” here, as Tayo Bero (2022) explained, projected a “white gaze.” Importantly, Twitter users responded to Leibovitz’ tweet with digital exposure and tone corrections that brightened and warmed up the tone of the images, prompting Vogue Magazine to do the same. The response on Twitter was quick and knowledgeable because the "recipe" to this high profile case was both culturally and technologically familiar. It was aligned to an existing regime of representation and it also shared its aesthetic technology with widely available practices, from film stock to visual retouching apps, where the prevalence of white-calibrated digital filters is well documented (Hunter 2019).

Then, in 2021 came the Google Pixel 6 camera, which promised to redress White-centrism by deploying a strategy of “photo equity.” This is how the new process was explained in the marketing campaign: “Historically, camera technology has excluded people of color, especially those with darker skin tones. . . We vastly improved our camera tuning models and algorithms to more accurately highlight the nuances of diverse skin tones” (Google 2021). The term “historically” does a lot of work here, denoting a break from the past that Google is seemingly spearheading. The 170 years of the photographic medium that was created as “a tool to capture visually and classify human difference” (Benjamin 2019, 68) and that persevered through technological shifts was treated as some bad “legacy code” that the tech giant had just "patched," echoing some colorblind societal discourses that frame racism as some outdated pathology that society is growing out of (Bonilla-Silva 2017, 103).

This technology and its claims for a new visual truth highlight another key concern: The interplay between technology and visuality is radically changing. The mobile devices that we use are complex technological assemblages that link visual acquisition and processing technologies with location and communication technologies, yet these capabilities are “flattened” onto a simplified graphic interface, as Johanna Drucker explains:

We look at interface as a thing, a representation of computational processes that make it convenient for us to interact with what is “really” happening. But the interface is a mediating structure … a space between human users and procedures that happen according to complicated protocols (Drucker 2014, 138–139).

The Google Pixel smartphone promises to bypass the White gaze with a tap and to reproduce the global majority more accurately, but a political visual literacy resists such flattening of visual practices and technologies, much like it resists the flattening of visual artefacts into mere representations. Instead, it insists on interrogating the layered aspects of visuals and visualities and the visual truths enfolded within them. Furthermore, I argue that we should take political visual literacy one step further and include within the analysis the relationship between practitioners and practices, or the negotiation of users with the constraints and affordances of aesthetic technologies.

Political Visual Literacy Through Mediated Action

In this section, I elaborate a methodology for Political visual literacy that links the conjunctural analysis of “aesthetic technologies” with their operationalization by actors. For this purpose, I introduce the theory of mediated action, which was developed by James Wertsch and based on the work of Lev Vygotsky (Norris and Jones 2012). Mediated action “conceptualizes how various culturally afforded tools extend our competence to perform actions” (Hilppö et al. 2017, 360). We know from scholars like Dyer, Roth, and Drucker that visual artifacts are “encoded by their technologies of production and embody the qualities of the media in which they exist” (Drucker 2014, 22). The mediated action approach centers the interaction between actors and the “aesthetic technologies” or “culturally afforded tools” they use.

Hilppö et al. (2017, 360) explain the concept of mediated action generally by using a relevant artifact:

When using a camera, for instance, we apply the culturally accumulated knowledge of various scientific and technical discoveries put into practice through the design process of the tool. Yet, neither the design process nor the cultural-historical trajectory of the tool alone determines its use… the way in which cultural tools mediate our actions, their mode of use, is flexible and open to the dynamics of the situation.

This “culturally accumulated knowledge,” which Wertsch refers to simply as “cultural tools,” is similar to the “aesthetic technology” (Dyer 2017). The “dynamics of the situation” involve what Wertsch terms “an inherent, irreducible tension between agent and ‘cultural tools’” (Wertsch 2002, 6). According to this ontological reshuffle, at the center of social life is human action, which is always mediated by cultural tools. At the most basic level, these would be language, custom, and convention, but cultural tools can also be physical devices or best practices that have sets of affordances and constraints. These may be material, like the limits of exposure of the film or the sensor, or to do with conventions and therefore be cultural, like determining what angle, depth of field, or exposure would be “correct” way to light images of people with different complexions.

Visual technologies, visual practices, and visualities are all “cultural tools” that co-determine visual action when employed by an actor. Importantly, Wertsch noted that the “study of the forces that give rise to cultural tools has not usually been the main focus of analyses of mediated action” (Wertsch 1998, 521). I argue that to understand the restrictions and affordances of visual practices, we must study these forces via a conjunctural analysis of visual practices, but also interrogate their operationalizations by actors and their negotiations with aesthetic technologies.

Wertsch (1998) suggested that there are two modes in which actors may use cultural tools: The first mode is “Knowing how,” which is the socio-technical knowledge of using a cultural tool. The other is “appropriation,” which is the “willingness to use the tool and. . .sense of ownership of it” (Wertsch 1998, 524). However, appropriation may be rejected in favor of “resistance,” which could occur in “explicit reflection or through other aspects of practice” (ibid., 524). “Appropriation” seems an apt label for Leibovitz’s portraits that deployed the established aesthetic technology that aligned to a White-centric visual truth. “Resistance” can apply to the retouching of that work that was shared on Twitter against the grain of white-centric technologies. Achieving a political visual literacy requires both a measure of “knowing how,” or at least a basic competence in the sets of constraints and affordances of the cultural tool/aesthetic technology being investigated beyond the ability to aesthetically contextualize it, as well as “resistance,” by interrogating the visual truths enfolded within it.

Of course, “Resistance” to visual cultural tools is nothing new. Much artistic practice engages with the subversion of visual practices that manifest harmful gazes. For example, Pushpamala N’s restaging of early anthropometric portraits in the Andaman Islands by Maurice Vidal Portman in her series “Native Women of South India (2000–2004)” demonstrates that mode of engagement. Katja Phutaraska Neef (2020, 132) analyzed this restaging as a resistance: “Pushpamala emulates Portman's colonial portrait, instantaneously dismantling his use of the camera, and challenging the authenticity of colonial, anthropological photography to shape colonised identities.” Pushpamala engaged with colonial the gaze, as “an unequally constituted right to scrutinize, to represent what is gazed at, and, if judged necessary, to intervene and alter the object of the gaze” (Ram 2018) and dismantled it from within by adhering to the basic scientistic aesthetic while performing different characters as the subject of the portrait.

We can also find “resistance” in visual politics. Särmä’s collaging is practiced as “cutting and ripping apart materials, and then repairing this damaged knowledge by gluing and pasting the materials back into a new kind of a composition” (Kangas et al. 2019, 9), while Van Veeren’s (2019) commitment to a layered reading of collages also gestures toward resistance to a straightforward representational explication. The approach that I advocate resembles these projects. It starts with the moment of “action” and considers the visual cultural tools used as layered “aesthetic technologies,” with implicit and explicit visual truths that, once “knowing how” has been achieved, may be appropriated or resisted by users. Next, I will operationalize this approach on a case study that traces back a visuality embedded in what is often framed as a new kind of visual practice: AI and facial recognition or more specifically: automatic gender recognition.

Transphobia, AGR, and the Operational Imaginary of “Infiltration”

Transphobia is “a hostile response to perceived violations of gender norms and/or to challenges to the gender binary” (Bettcher 2014). It is a moral panic that scapegoats a vulnerable group. Recent scholarship has noted the global proliferation of transphobia as an ingredient of extreme right politics (Butler 2021; Kray and Linke 2022, 17). It also explored its integration into popular right-wing beliefs (Borba 2022, 74; Pearce et al. 2020) and, in turn, into policies (Galman 2022, 215). This spread has been plotted by scholars like McLean, who found an “importation of culture war tropes from the US” in the United Kingdom (McLean 2021, 1; Borba 2022), who focused on the role of the Vatican in spreading transphobia globally; Kováts (2017), who wrote of so-called gender ideology as an “enemy image” in Europe; and Kuhar and Paternotte (2017), who noted how European anti-gender activist groups replicated each other’s protest methods. The global trajectory of this proliferation should make transphobia a central concern for International Relations.

Exploring visual politics in the context of trans people is a sensitive task. Colliver’s (2021) recent collection of testimonies of trans women victims of hate crimes demonstrates the centrality of visual perception to violent encounters, while Begun and Kattari (2016) explain how often “visual conformity” is crucial to the mental well-being of trans people. This case study does not interrogate the lives of trans people, but rather uses a political visual literacy to examine a transphobic practice that claims an ability to “expose” trans women.

The case study focuses on the use of a specific AI technology by “Giggle”: A social media application that has recently received some media attention. It was designed as a social network that provides a safe online space for women over the age of 16, and it uses automated facial analysis to inspect selfies uploaded by users to visually verify that the user is a woman.² This technology is termed “automatic gender recognition” and it is “a sub-field of facial recognition that seeks to algorithmically classify gender from photographs” (Keyes 2018). The reason for the media attention was that the algorithm seemed to do a bad job at identifying trans women as women and it was not clear how the company would approach this sensitive issue after it had initially declared that it welcomed trans women (Schiffer 2020).

Soon after, Giggle’s CEO announced that trans women are unwelcome and that the need to police access to the site stems from their attempts of "infiltration” (Scheuerman et al. 2021, 12). Such a step deployed against “infiltration” is clearly transphobic, and it had been labeled as such (Schiffer 2020; Scheuerman et al. 2021). It strives for a visual truth of sex/gender that is based on the perceived immutability of sex as an essential category. Importantly, it seeks to legitimize this stance through the technology, which is cast as an objective and automated judgment delegated to a “neutral” visual technology that can help to identify and exclude trans women (Scheuerman et al. 2021).

Scholars have been raising the alarm about the way technology is trusted to make "objective" judgments on race and gender, despite a horrible history of such attempts (Pugliese 2005; Browne 2015). Biometric technologies, in particular facial recognition, are deeply racialized and “infrastructurally calibrated to whiteness” (Pugliese 2005, 5; see also Buolamwini and Gebru 2018). Beyond white-centrism, Simone Browne wrote about biometric technologies in the context of anti-Black practices of surveillance, a “digital epidermalization” that follows a logic of surveillance as “branding” that could be traced back to chattel slavery and the physical branding of enslaved Black people (Browne 2015). Browne also found openly racist ideas in contemporary research in machine learning and recognition. For example, a call for machine learning to use anthropometry, a pseudoscience developed in the late 19th century by Alphonse Bertillon that was devoted to mapping physical differences to aid in identification but also in racialized typification (ibid., 112; see also Stevens and Keyes 2021, 847; Pugliese 2005, 7). In response to this challenge, Browne suggested that biometric technologies like facial recognition should be considered through a “critical biometric consciousness” that examines their “historical antecedents” (ibid., 118).

Scheuerman et al. (2021) identify different “historical antecedents” for these visual practices. They termed the underlying logic of the practice: “auto-essentialization.” This is “the use of automated technologies to re-inscribe essential notions of difference, with the face currently serving as the medium of choice” (ibid., 2). The “re” in re-inscription stems from the delegitimation of all such essentializing practices associated with scientific racism after the Second World War and their later re-emergence as technological innovations that bear a supposed objectivity and an ability to glean biological facts (Browne 2015, 114). They traced it back to its analog origins in facial measurement by Cesare Lombroso, a contemporary of Bertillon and founder of modern criminology, who measured faces and skulls to deduce theories of origin that were congruent with the race sciences of his day.³

In the nineteenth and early twentieth centuries, the scientific study of race and gender was intertwined, and “the scientist could use racial difference to explain gender difference, and vice versa” (Stepan 1986). Lombroso was no exception to this rule. He combined phrenology and physiognomy with outrageous speculation and just-so stories to comment on race and gender. In particular, he explored the visual resemblance between people and different animals (see Gould 1996, 124). He aimed to read “the biological “truths” of racial difference and hierarchy” (Scheuerman et al., 7), focusing on visual signs or “stigmata” that indicated a biological “atavism”: supposed primitive features that remain in some individuals and that manifest both physically and mentally.

For a genealogy of “ideological biologization” (Sekula 1986, 12), both Lombroso and Bertillon are relevant origins, and importantly, these pseudo-scientific essentializations form an integral part of contemporary social sciences. We often forget the degree to which the desire to mark and hierarchialize human difference is enfolded into our disciplines (Winant 2015). Scheuerman et al. (2021, 8) explain that while physiognomy and phrenology have been discredited as unscientific and racist endeavors, their core visual truth survived: gleaning identities from faces.

I agree with this conclusion, but I argue that a different historical antecedent best explains this specific use of AGR technology as a visual cultural tool: Francis Galton and his project of “pictorial statistics.” To be sure, this is not a forgotten history. Sekula’s influential piece on “The Body and the Archive” introduced Galton’s method as one of the major approaches alongside Bertillon and Lombroso. Moreover, Lila Lee-Morrison has identified his “pictorial statistics” as “an analogue antecedent to the application of statistical methods of facial recognition” (Lee-Morrison 2018, 115). However, Lee-Morrison had not considered the aspect that I find central in this context—the racist visuality enfolded into the practice, one that obsesses over Jewish faces and the “Jewish gaze” as a visual truth of an essential Jewishness. This was a vision-machine tasked with the “objective” identification of a racialized group, yet it carried within it a desire to exclude the group because of the danger it was believed to encompass should it be allowed to assimilate or “infiltrate.” In the next part, I will offer a conjunctural analysis of these aesthetic technologies and their claims for visual truths that relate to “infiltration” that echo the contemporary transphobic visuality of AGR.

New Skin for the Old Ceremony: Galton and His Face-Machine

Francis Galton was an English polymath. He was the inventor of the statistical regression and the founder of Eugenics and his “Pictorial Statistics” falls somewhere between the two. Galton developed the method of “composite photography” (which he later called pictorial statistics) as a means of teasing out “racial types” and to find what he called the “true physiognomy of a race” (Galton in Sekula 1986, 367). His method was to “to throw faint images of the several portraits, in succession, upon the same sensitised photographic plate” (Galton 1885, 243). In simple terms, this was the layered reproduction of similarly posed portraits through multiple exposures on the same photographic plate. Each photograph was reshot and exposed for a fraction of the total exposure time (say, 12 images, each at 1/12 of a correct exposure time) resulting in a “composite” image of the portraits that bore the marks of the different exposures.

Galton reported his surprise when he saw the result. He expected that the layering of different portraits would result in “mere smudges” (Galton 1879, 3, 5–6) but instead he found that:

There are then so many traits in common, to combine and to reinforce one another, that they prevail to the exclusion of the rest. All that is common remains, all that is individual tends to disappear… the process of composite portraiture is one of pictorial statistics…they are real generalisations, because they include the whole of the material under consideration (ibid.).

Here was an “ideal form of empiricism” (Ambrosio 2016, 55), a visual essentialization that revealed the supposed statistical true mean of a group. A visual cultural tool that could glean the essential facial qualities of a "race" and that derived its claim for visual truth from the universal rules of statistical distribution. This visually supercharged pseudoscience drew on the ocular-centrism of Victorian scientism (Jay 1988; Grayson and Mawdsley 2019), which privileged sight as a mode of verification and was racialised to begin with, as George Yancy (2008, 2) explains:

The European gaze was able to discern with “clarity” and “accuracy” the “truth” about certain human bodies vis-a-vis a white racist discursive regime of truth. The gaze reinforced the truth of the racist categories and the racist categories reinforced the gaze.

This visual truth was intensified by the authority of the camera, which WJT Mitchell explained was a “supereye. . .a perceptual prosthesis that can stop action better than the human eye, resolve finer detail, remorselessly attend to the subtlest distinctions of intensity, and not leave unregistered anything in the field of its gaze” (Mitchell 1992, 28). Here was a clean and scientifically cutting-edge method of visually extracting a hidden essential trait. And so, much like Muybridge’s still images of a galloping horse used a short exposure time to reveal that it lifts all four legs, or like Frank Gilberth’s motion studies used a long exposure time to track the supposed optimal movement of workers on the production line, Galton’s layering of portraits was purported to operationalize an “ideal empiricism” to reveal a new visual truth, a perfect form of racial visual essentialization.

This method was an aesthetic and technical conceit that answered a longstanding question, which was most recently re-asked by Keith Kahn-Harris (Kahn-Harris and Stothard 2022) in his book: “What does a Jew look like?” This more recent version of the question was asked as a response to the visual representational regime of Jews in Britain, who are often represented in the media through a particular image of two ultra-orthodox Jewish men. To answer his question, Kahn-Harris enlisted photographer Robert Stothard, who took that ubiquitous photograph, to produce a portfolio that showcases the diversity of Jewish people in Britain and challenges their stereotypical portrayal. Kahn-Harris challenged the still prevalent idea that “Jews that look ‘just like us’ or who signify in more ambivalent ways cannot stand for the whole. Only those who cannot be assimilated into ‘us’ can truly represent ‘them’” (ibid., 5).

This current representational regime is a diluted form of what was at Galton’s time a much more pronounced racist vigilance, which was set to the backdrop not only of the race science of the day and the racialized visualities of modernity, but also to a longstanding Christian European violent antisemitism, which was based in part on an inability to visually identify Jewish “infiltrators.” From the early thirteenth century until the late eighteenth century (and again in Nazi-occupied Europe), many groups of European Jews were required to display a visual identifier (Wallach 2017, 8).⁴ Composite photography was a new technology enlisted as a substitute, a machine for “clocking” Jews in the absence of decrees requiring us to wear a visual sign, a new skin for the old ceremony.

When we break down the method Galton used, it leaves no doubt as to the deep antisemitism that drove his project. Galton was after a particular kind of mark that he termed a “Jewish gaze” (Morris-Reich 2016). This was to him the key to “Jewishness” that he believed, emanated from Jewish eyes. In 1885, he traveled to the Jews’ Free School (JFS) in Spitalfields to photograph Jewish schoolchildren for his composites.⁵ It was the stated goal of the JFS “to wipe away all evidences of foreign birth and foreign proclivities, so that [your] children shall be identified with everything that is English in thought and deed” (school principal Abrahams in Kahn-Harris and Gidley 2010, 20). These boys, who were being assimilated into British society, represented a threat of “infiltration,” which to a eugenicist like Galton was a racial threat to the national “stock.” However, Galton’s reflections on encountering the faces of the Jewish boys suggest that beyond the “normal” pseudoscientific pretences of a eugenicist there was a familiar antisemitism:

There was no sign of diffidence in any of their looks, not of surprise at the unwonted intrusion. I felt, rightly or wrongly that every one of them was coolly appraising me at market value, without the slightest interest of any kind (Galton 1885, 243).

Here, we need to pause and consider Galton’s method of “pictorial statistics” as “mediated action.” He explained that the key to a successful composite is a perfect alignment of the subject’s eyes as the anchor for the layered composite (Galton 1879, 5). He later considered his portraits of Jewish children to be his best work, but the amount of labour required made him suggest using group portraits instead (Galton 1885, 244). His reflections on the boys’ gaze stem from a long and meticulous session in which he projected these old racist tropes, which he long held.⁶ In other words, the visual tool Galton developed for establishing the visual truth of the Jewish gaze was enfolded with an antisemitic visuality. Galton used this method to investigate other groups, but it was specifically his work on “The Jewish gaze” that lived on.

Galton’s composites of Jewish children appeared in Hans Günther’s visual albums of race science that were published in Weimar and Nazi Germany. Günther, who was the most prominent Nazi race scientist, shared Galton’s fascination with Jewish faces and was inspired by Galton’s work (Gilman 2002). Günther similarly considered “the Jewish gaze” or the Blick to be the key to the identity of the Jewish “anti-race” (Morris-Reich 2016, 138). However, he believed that visual racial markers are much more muted than Galton thought because “infiltration” was already well under way. Therefore, he did not think composite portraits reveal the “mean type” of a race. He still used Galton’s images, but instead of archetypical representations, they were integrated into a new visual cultural tool that was meant to produce a new type of racial vigilance.

Kerry Wallach (2017, 7) wrote about Jewish visual “passing” in the Weimar Republic, when “Jewishness emerged as a known quality that one could display and perceive, and which was desirable at certain times, often in Jewish-friendly spaces.” Much like Galton, Günther deployed a visual cultural tool to challenge Jewish ownership of their own visibility. It retained the same underlying antisemitic visuality and was remade for mass public use. Readers of his popular books “Racial Science of the German People” and “Racial Studies of the Jewish People” were invited to explore large clusters of portraits, including Galton’s composites, and to decide for themselves what might connect them. Günther promoted a new visuality in a way that is echoed in how computer vision functions, “training” users on a database of images (Stevens and Keyes 2021, 835).

Of course, this was a stone soup. There was no underlying logic to be discovered, no guiding line to be found. What the project did, was to mass-produce a Judeophobic vigilance and a cottage industry of home-made “stigmatas” with which readers would be on the lookout for “infiltrators,” helping to train a “racially conscious social subject” (Morris-Reich, 142). The composite portraits, as well as Günther’s albums, were both materializations of a wider visuality that focused on a “Jewish gaze.” This visuality desired to perfect visual facial readability in a way that would enable users to mark subjects. They were what Edkins called, after Deleuze, “abstract machine(s) of faciality” (Edkins 2015, 124).

However, in order to bring this “abstract machine” into focus, in true visual form, we need to project it through a “circle of confusion” and deploy a unique visual artifact as a theoretical aid. Lee-Morrison (2018) found an interesting link between composite photography and a type of facial recognition: the “eigenface portrait.” This is a widely used technology in facial recognition that is a visual mapping of a group of faces that looks similar to Galton’s composites. It is an integral part of the computational process that is not meant as a representation (Stevens and Keyes 2021, 840). However, as Lee-Morrison (2018, 128) argued, once displayed, it bears a striking resemblance to composite photography, and “it functions as both a representation as well as a construction of human faces.” Such a visual artifact can be labeled an “operative image” (Farocki 2004, 17), which is a visual representation “that do[es] not represent an object, but rather [is] part of an operation.”

Jenny Edkins explained that there is a “myth of the power of facial recognition technology” that far exceeds what the technology can actually do (Edkins 2015, 100). Writing in the context of policing, she argued that automatic recognition is more about threats than abilities. This is certainly the case with the technology used by “Giggle,” which effectivity is minimal. ⁷ What is at stake here is not any functioning machine of faciality, but the dangerous discourse that surrounds its (in)efficacy. As the Günther example shows, visualities can be integrated into technologies and practices in such a way that their basic function is shifted, so that the intended output is not to actually “clock” groups, but to feed the desire for their essentialization and exclusion by establishing the possibility of discoverability and the promotion of visual truths that target “infiltrators.” Therefore, if we think about this this through the representation in   figure 1, what we are dealing with here is not an “operative image,” but rather an “operative imaginary” of transphobia powered by a well-established mode of technological justification, or visual truth.

Political visual literacy interrogates visual practices so that we understand the visual truths enfolded in them and in turn the wider politics they are implicated in. The approach I promote here uses conjunctural analysis to explore “visual cultural tools” through their “aesthetic technology” as well as their use as mediated action. Using this approach, I demonstrated that the tributaries of AGR, as it is used here, are not only avenues of racial hierarchialization that harm trans and non-binary people, but its deployment as a filtering technology connects to a well-established didactic project of targeting, set to mark “infiltrators.”

Conclusion

The field of visual politics is booming, but it largely neglects the study of oppressive visualities, even when it engages with practices suffused with them. Critical scholars in the social sciences follow theories back to their past uses and related traditions of thought, resisting straightforward progressive narratives of intellectual development. If scholars of visual politics would adopt a similar method, we would find that the imprints of oppressive politics are at least as prevalent in the visual realm as they are anywhere else, and at times, these politics are articulated visually.

Representational analysis, which is prevalent in visual politics, is an important and fruitful avenue for research, but it should be complemented by some measure of a political visual literacy that interrogates the complex technocultural implications of projects of visualization. In this article, I demonstrated the utility of political visual literacy in two cases: the deep implication of the white gaze in visual technologies and the historical genealogy of a purportedly neutral method of visually gleaning human difference, which is in effect a transphobic visuality that follows an established script of visual targeting.

The approach to political visual literacy that I presented draws on cultural studies and contains two parts. First, it employs a conjunctural analysis of aesthetic technologies, into which visual truths are enfolded. This highlights the ways in which harmful gazes frame visual technologies and practices. Second, it aligns this analysis with an ontology of “action,” in which social actors use these aesthetic technologies and negotiate with their affordances and constraints to either appropriate or resist them. This is one possible application of a more general need in visual politics to extend our work to the wider political projects that shape visual thinking, visual reading, and visual writing.

Acknowledgments

This article is dedicated to the memory of my friend and mentor Rosy Hollis. The manuscript has greatly benefitted from generous and critical readers. I want to thank Laura Sjoberg, Amber Onver, Raz Weiner, Suki Finn, Amos Morris-Reich, Louise Bethlehem, Clive Gabbay, Federica Frabetti and Kat Gupta for their engagement and comments.

Footnotes

References