Back to the Fathers: Looking at Wölfflin and Riegl After Ai

15-05-2023 14:00 - 16:00 GMT+1 || Join us on Zoom

Style as Ideal Type and Weltansicht: A Reassessment of Heinrich Wölfflin’s Principles of Art History

Author:

Rémi Mermet - École normale supérieure Paris, Paris, France

Abstract

The Swiss-German art historian Heinrich Wölfflin (1864-1945) is usually considered, along with Alois Riegl, Aby Warburg and Erwin Panofsky, to be one of the founding fathers of scientific art history (or Kunstwissenschaft) in the early 20th century. His Principles of Art History, released in 1915, has been one of the most influential books in the history of the discipline. Yet, despite the recent celebration of the centenary of its publication, and its retranslation into English by the Getty Research Institute, this work still triggers strong opposition within the art-historical community. For a century, Wölfflin’s book has regularly been criticized for his so-called “formalist” approach to artistic styles— an approach that is generally deemed to be blind to the social, political and ideological context of the creation of the works of art (if not unconsciously racist).

Against such interpretations, I will seek to show that Wölfflin’s formalism is merely “tactical”. In the Principles of Art History, he turns away from the analysis of the content of the works of art to better grasp the inherent meaning of their form. His definition of style as Sehform (“form of seeing”), which owes much to the legacy Goethe’s morphology, is indeed very close to Max Weber’s notion of ideal type, as well as to Wilhelm von Humboldt’s notion of Weltansicht. For Wölfflin, as for Weber and Humboldt, form and meaning are not separate entities. On the contrary, form is the very manifestation of meaning, the expression of the spiritual in the sensible world—an expression that, according to Wölfflin, has been objectified in the various styles of the artworks of the past.

Bio

Rémi Mermet received a joint PhD in philosophy and art history from Paris 1 Panthéon-Sorbonne University and the University of Geneva in 2019. He is currently a "EUR Translitteræ" postdoctoral fellow at the École normale supérieure in Paris. His research focuses on German aesthetics (in particular early twentieth-century Kunstwissenschaft) and Ernst Cassirer’s philosophy of symbolic forms. He is also interested in the interactions between aesthetics and the natural sciences. He recently coedited (with Danièle Cohn) L’Histoire de l’art et ses concepts. Autour de Heinrich Wölfflin (Éd. Rue d’Ulm, 2020) and the new translation into French (by Sacha Zilberfarb) of Heinrich Wölfflin's Principles of Art History (L'Écarquillé, forthcoming).

Understanding the place of computer vision in the historiography of formalism

Author:

Tristan Dot — University of Cambridge, Cambridge, United Kingdom

Abstract

I would like to talk –in this presentation– about the link between the historiography of formalism, in art history, and the way computers ‘see’ images and artworks. Indeed, I think that such a work of comparison, between art-historical formalist theories, and computer vision algorithms, can be extremely fruitful, for both fields.
In my presentation, I will first do a literature review on the links between formalism and computer vision. If some authors have studied the correlation between the main factors explaining computational stylistic classifications and formalist criteria (especially, the criteria defined by Heinrich Wölfflin in his book Principles of Art History (1915), others have tried to meticulously re-analyse, and re-activate some formalist theories in light of computational methods. In a second part, I will present the first results of my research concerning Alois Riegl’s Stilfragen (1893) analysis in light of AI visual similarities. Is a computer vision algorithm able to automatically re-create the visual, formal connections between images (mainly images of ornaments, in the case of Stilfragen) exposed by Riegl in his book? This part will also be the occasion to discuss the impact of Alois Riegl’s works concerning the notion of artistic style, and the way computer vision can help us question Riegl’s conceptions of style. In a last part, I will focus more on the question of style classification, in the fine arts sense, and talk about the way some huge computer vision models –often called ‘foundation models’– understand the notion of style (these foundation models are trained on hundreds of millions of images, mostly from the web, but also from art historical collections – they are now more and more used to automatically generate images from textual prompts). What is called a ‘style’ in the AI world, do these new models perform more accurate, convincing ‘artistic style transfers’ than previous computer vision models? Centered around the historiography of formalism, and its links with current computer vision methods, my presentation will essentially question the way computers analyse –especially by computing ‘visual similarities’, and classifying into styles– images.

Bio

Tristan Dot is a first year PhD student in Digital Art History, at the University of Cambridge. Coming from a double background in machine learning and in art history, he has worked as a research engineer in computer vision, and has written his master's thesis on the new nature of digital images in the AI era. During his PhD, he is trying to determine the place of computer vision in the historiography of art-historical formalism. What are the possible links between some seminal formalist theories (such as the works of Aloïs Riegl, or of Heinrich Wölfflin), and the way neural networks embed images? Could computer vision re-activate the notion of style –for instance, by quantifying it–, and help us shape new, fruitful methodologies in art history? These are some of the key questions Tristan is working on.