What is Barr X Inception CNN?

Barr X Inception CNN is an experimental project whose aim is finding out the conditions that will define our hyper technological future and the available spaces for propositional actions that will help avoid the collapse of human beings and guide us in the development of new forms of subjectivity

Specifically, Barr X Inception CNN investigates the new otherness that confronts the human being of the 21st century with a non-human “other” – the machine – by invoking traditional narratives, such as the man-machine relationship, or archetypal stories, such as the myth of Prometheus.

In order to carry out this reflection, we decided to take the famous 1936 diagram by Alfred H. Barr (the first director of MoMA in New York) as a starting point. In his chart, Barr established one of the most influential paradigms for the interpretation of the artistic evolution during the first third of the 20th century, by establishing relations of mutual influence between certain authors, styles and movements. This diagram, recently updated by the Genealogies of Modern Art exhibition , co-organized by the Juan March Foundation and the Picasso Museum in Malaga, prompted us to ask the following question: If a considerable number of the art pieces that are “representative” of the movements, styles and authors included in Barr’s diagram was analysed by Artificial Intelligence devices -such as convolutional neural networks (CNN)- which can detect visual (formal) similarities between digital images, what would the result be? Would it be similar to Barr’s approach in his diagram or would it be completely different? What can an artificial vision device “see” that a human being cannot “see” and vice versa? Is it possible to establish spaces of negotiation which allow for the expansion of cultural analysis interpretative horizons, or is this dialogue not possible?

This experimental project wants to avoid binary confrontational or simple comparative narratives and explore the multiple relationships that arise in various dimensions. The X in the title that links the two players -Barr and Inception CNN- expresses this multiplicity of connections and possible dimensions. It is a small tribute to the insightful reflection made in 2012 by  W. J. T Mitchell in his essay “Text X Image”, which he later presented at the closing conference of the congress Theory and Artistic Literature held at the Picasso Museum in Malaga in March 2013. 

The now legendary catalogue dust jacket of the Cubism and Abstract Art exhibition (1936) held at MoMA in New York, visually reflected the thesis developed by its curator, Alfred H. Barr Jr. (1982-1991). It established a genealogy of modern art that started in 1890 and concluded in 1935, covering a total of forty-five years of artistic production.  Barr’s daring task was selecting those productions of the last decade of the nineteenth century that were considered to be the precursors of the first avant-garde movements and use them to develop a network of relationships that advanced chronologically and was based mainly on influences and formal relations. The dense network of arrows in the diagram describes which artistic movements were influenced by previous artistic experiences, as well as the connections between the artistic trends of that time.

The graphic eloquence of the diagram provides an easy reading of the conclusions of the historiographical discourse developed by Alfred H. Barr, summarised in the title of the 1936 exhibition (Cubism and Abstract Art): cubism, due to its power of influence, was the avant-garde movement that represented a turning point in the evolution of modern art, which would lead to the domination of abstract art. Faced with the challenge of developing a coherent narrative for the first museum of modern art, its young curator and director elaborated a formalist, evolutionary and teleological discourse that placed pure abstraction as the paradigm of the art of the first half of the twentieth century. Alfred H. Barr Jr. built the most powerful historiographic account of modern art, which would be duly questioned by the critical discourses of the second half of the century, and some of its criteria are still operative.

If a considerable number of the artworks “representative” of the styles and movements included in Barr’s diagram was analysed by a computer vision system -such as convolutional neural networks (CNN)- which can detect similarities and visual-formal contiguity between digital images-what would the result be? What relationship would emerge from the computer analysis? Would other possible relationships between the analysed images become visible?  Could an artificial neural network identify formal analogies not perceived by the human eye that are potentially significant for historical-artistic interpretations? Should this be the case, which new narratives could be developed starting from this point?

aturally, additional questions  have emerged during the four months it took for the project to be developed, and they are directly related to the possible critical revision of Alfred Barr’s diagram: If we were to make sense of the results obtained, would it be possible to create an alternative diagram to the one outlined by Alfred H. Barr? Would it still be plausible to use the notion of genealogy in this new diagram? Would the paradigm of abstract art still be valid in this new scenario? Would there be another paradigm that works alongside it? Would the established hierarchies between the different artistic movements change? Would cubism keep its position as the most influential movement? There would also be more far-reaching questions about the limits of human perception-cognition and the development of a cultural gaze, a hybrid between the human and the non-human.

As a result, this experiment provides two lines of reflection, which are not restricted to the interpretations formulated in this project (always limited to a specific time and space) but remain open for further reflection. 

1.  The Modern Prometheus, again

Given that a CNN Inception is an Artificial Intelligence device that operates independently from human beings in its analytical and classificatory task, the question is: what new spaces of man-machine negotiation are opening up in the artistic field for the interpretative making and cultural analysis of human visual production? This question falls within the framework of an ancestral narrative, which is updated from time to time (see Hayles’ magnificent review), and which, somehow, forces the subject to adopt a position – be it a conscious or unconscious, explicit or implicit one. This positioning usually oscillates between the poles of the techno-utopic and techno-phobic, with a whole series of intermediate degrees between them,  moving between optimistic enthusiasm, critical caution, incredulous scepticism, indolent indifference, uncertainty, ancestral fear of the unknown or the exoticism of a new “other” to be confronted with.

In a scenario defined by the progressive decentralization of the human being -who now, when creating and producing knowledge, has to coexist with a whole plethora of non-human devices, the consequences of which are, among others, the displacement of the subject as the centre of the cognition-vision and unique referent of the decisional processes- the reflection on the relationship between humans and machines becomes especially relevant. Barr X Inception CNN is an attempt to explore the avenues of research and experimentation that can take place in History of Art with the aim of contributing to this debate.

The Myth of Modern Prometheus]

2Other Spatial Stories

Given that the computational analysis carried out by a CNN on a set of digital images transformed into vectors of numerical information is projected into a high-dimensional vector space giving rise to a visual field, which is, in fact, a metric space of a topological nature, the question is, in what way can this type of space contribute to the creation of new interpretative categories for cultural analysis? What is its potential as a space, alternative to the physical-Euclidean space and to the geographical space and, consequently, what other types of spatial situations can be developed??

An investigation in terms of spatial interpretation could not be more appropriate for the case at hand, since, due to its diagrammatic nature, Alfred H. Barr’s diagram is also a visual device based on a spatial ordering logic; in other words, its decoding and interpretation is given by the arrangement of the elements on a two-dimensional level.   Likewise, the Genealogies of Art exhibition (2019-2020) -which was the inspiration for the Barr X Inception CNN experiment- transforms, in a risky exercise of great museographic creativity, Alfred H. Barr’s diagrammatic chart into a three-dimensional, physical space, modelled by the visual relations that the artworks that are materially and physically present, establish with each other.


Possible scenarios

1. Equivalence: the result of the analysis with a CNN Inception is similar and/or equivalent to Alfred H. Barr’s proposal in his diagram. This equivalence scenario entails two possible interpretations (not exclusive). 

_ AI, with its human-like abilities, can be a useful tool in an equivalence scenario, which will therefore help optimize and streamline specific processes.

_ An equivalence scenario, where the decisions of human beings and those of machines are similar, can generate restlessness and uneasiness because of the possibility of the two becoming indistinguishable, as demonstrated by the Turing test 

2. DivergenceThe result of the CNN Inception analysis is essentially different from Alfred H. Barr’s diagram.

_ A scenario of divergence can be highly effective in raising the possibility of new interpretations and narratives, alongside the existing ones.

_ A scenario of divergence can also provide an interesting basis to reflect on the perceptive-cognitive limits of AI and humans, and also on the existing difficulties of the interconnection spaces. It also leads to reflect on the AI’s limitations to carry out analyses on cultural material in a way that is significant to humans.

3. Unreasonable: The result of the analysis carried out by a CNN Inception is not only divergent but absolutely unintelligible according to the way human logic and rationality work. It it not possible to interpret it in any way or make any sense of it.

_ A scenario of unreasonableness or unintelligibility makes us think about the gap that still exists between machine versus human logic, and on the future possibility that this will ever disappear.