Analyzing Historical Paintings as Data for Political Insights

Ongoing advancements in machine learning are creating significant opportunities to analyze extensive visual data, particularly within the realm of historical political economy. Research led by Valentine Figuroa from the Massachusetts Institute of Technology (MIT) explores how paintings from museums and private collections can serve as a valuable, yet underutilized, source of information regarding political and societal transformations throughout history.

Before applying computational methods to paintings, Figuroa emphasizes the necessity of establishing a robust framework to assess the information encoded within these artworks. This framework must clarify the assumptions under which the paintings can be interpreted, drawing on the enduring themes prevalent in traditional humanities scholarship.

Framework for Analysis of Paintings

The article presents a comprehensive framework, utilizing a database that encompasses 25,000 European paintings dated from 1000 CE to the First World War. Three distinct applications are highlighted, each examining different types of information conveyed in the artworks: depicted content, communicative intent, and incidental information. These applications also relate to a significant cultural transformation during the early-modern period.

The first application revisits the concept of the European “civilizing process.” This process refers to the internalization of stricter behavioral norms that coincided with the expansion of state power. By analyzing paintings of meals, the study investigates whether there is an observable trend toward increasingly complex etiquette depicted in these artworks.

The second application focuses on portraits to understand how political elites crafted their public image. It reveals a long-term evolution in representation, transitioning from chivalric portrayals to more rational-bureaucratic depictions of men in positions of power. This shift highlights the changing dynamics of political authority and societal expectations over centuries.

The final application documents a long-term process of secularization, which is measured by examining the proportion of religious paintings over time. The research indicates that this shift began prior to the Reformation and gained momentum afterward, reflecting broader societal changes in attitudes toward religion and spirituality.

Implications for Historical Understanding

Figuroa’s work underscores the potential of treating art as data in political history. By employing machine learning techniques to analyze visual representations, researchers can uncover deeper insights into societal norms, political identities, and cultural transformations. This innovative approach not only enriches our understanding of historical contexts but also encourages interdisciplinary collaboration between the humanities and data science.

This exploration into the intersection of art and political history invites further investigation into how visual culture shapes and reflects the dynamics of power, identity, and societal change. As machine learning continues to evolve, the integration of artistic analysis could redefine traditional methodologies in historical research, offering fresh perspectives and interpretations of the past.