Aura

It was 2022, and autumn was just beginning. DV had woken up early to take the train from Rotterdam to Amsterdam: her usual commute from home to the prestigious artist residency. That day, she had to present her progress after a year of work, embodying that attitude so typical of the contemporary art world where you pretend to be casual, as if you couldn't care less about the audience of specialists and your own peers.

The presentations were held in an attic on the top floor of the building, former 19th-century horse stables that are now government-protected heritage. The place was dark, the air barely circulated, and the forty ambitious artists crammed into that attic tried to breathe while also appearing confident when it was their turn to present.

DV talked about her trip from the past summer; she had gone to Las Vegas with her assistant Harry after an invitation to DEF CON, the world's largest hacker convention. They took the opportunity to film a video piece contrasting the desert landscapes with the crazy intensity of the city. As if she were just talking about her vacation, as if such trips were no big deal, DV showed photos of the red sports car she had rented for the road trip, videos of the highway surrounded by imposing barren landscapes, and selfies of her fooling around at the Venetian casino. She exuded a lot of charisma and cracked a few jokes about American society in decline. Cynicism reigned. She returned to her seat, mission accomplished, and continued watching the parade of her peers in the stifling air. Her last memory of that morning was a colleague's presentation showing horrific images of the war in Ukraine.

Immediately after, she found herself in the building's courtyard surrounded by her colleagues. It was sunny, her friend Fiona was massaging her back and DV felt grateful to be in such an affectionate environment. Later, they would explain to her that she had ruined the event because, from her seat, DV had loudly asked her friend Benjamin, who was sitting next to her:

To figure out what had happened, the artists blamed the heat, the professional pressure inherent to such events and speculated about a possible panic attack. It was Karel who, somewhat reluctantly, suggested that DV had experienced an epileptic seizure — that it had very similar characteristics to the episodes his father, who lives with that chronic condition, suffers from.

Months later, the medical diagnosis was conclusive: focal epilepsy. They were able to detect it after a brain scan and two electroencephalograms (one with sleep deprivation, for which DV had to spend the whole night at the Rotterdam casino losing money with her friend Katarina). Talking to those close to her, she gradually learned that the episode at the prestigious artist residency had not been the first, that on other occasions she had followed the same script: a distant gaze fixed on the horizon, agitated breathing, loud swallowing, and finally asking whoever was next to her:

Friends, colleagues, and acquaintances simply dismissed the episodes, attributing them to her eccentric personality, which leaned toward philosophy and mysticism. However, they were epileptic episodes in which she lost consciousness and which involved retrograde amnesia — meaning she had no idea they had occurred. By detailing the characteristics of her case, she was able to establish that each seizure is preceded by an aura. According to Jerome Engel, world authority on epilepsy, auras are neurological episodes with sensory, autonomic, or psychic symptoms that precede the loss of consciousness.

The elements that most capture her attention during the auras are, in a way, those that most evidence her existence in a civilization dependent on technical structures: the train she takes to travel across the country, her daily breakfast composed of organic foods thanks to food engineering, the Wolf Alice song in her headphones about urban heartbreak. The aura is frightening but also sparks curiosity; in those moments, DV always tries to push mentally further and understand what is happening.

Francisco Varela and Humberto Maturana were two highly influential Chilean biologists in global academia. Their theory of autopoiesis has had such an impact that it has not only been adopted in biology but has also given rise to a systems theory applied in computing and sociology. It is the work of Maturana and Varela that constitutes the basis of this essay, which connects the neurological phenomenon of the epileptic aura with a proposal for artificial intelligence, where AI serves as a tool to help explore what exists in this altered state of consciousness prior to its fading.

The scientific community tends to omit the more spiritual vein that the Chileans gradually adopted over time. Varela, in fact, insisted on incorporating Buddhist matters into his work on cognition. To address the notion of consciousness, for example, he used the model from the Abhidharma school, which defines a progression of five phases that make up the cognitive process:

• Contact (rupa): the moment of identifying what is presented externally, identifying its forms.
• Feeling (vedana): perceiving what type of emotion the identified thing produces.
• Discernment (samjña): recognizing what is in front of us to activate an impulse toward it.
• Intention (samskara): activating a network of perception patterns to apply to what we are identifying.
• Attention (vijnaña): deployment of consciousness, the mental experience that involves being in relation to the object.

In DV's particular case of epilepsy, there is an interruption at the third step — the step of discernment. When she loses consciousness, DV is capable of behaving in a seemingly conscious manner: she could distinguish her interlocutor and determine which language to speak to them in, she is able to walk down the stairs with her friends to the courtyard of the prestigious artist residency building. It could be said, according to the Abhidharma scheme of consciousness, that during the epileptic attack, she is capable of identifying forms and experiencing a feeling toward them — at least a basic feeling about whether they are good or bad. However, the subsequent stages become blurry in terms of memory.

In more intimate conversation, DV is able to admit that over time she has felt somewhat special having this condition, that she feels a bit like Joan of Arc having visions. She even channels a certain eroticism about it — erotic in the Georges Bataille sense when, in Erotism, the author describes the feeling that dying doesn't matter because erotic desire (sexual, romantic, and religious) tends toward the negation of individual duration. DV has a genuine curiosity about that state; she desires to understand what she barely manages to capture during the auras, to consider the possibility of putting the moment of discernment in slow motion and contemplating a detour onto other avenues of consciousness.

Should we start our search for ecstasy and head to the smartshop on the corner to buy hallucinogenic mushrooms? The truth is, DV no longer wants to keep doing drugs. As a textbook Western millennial, that is an avenue already traveled, and today her friends are too depressed to be suitable company.

So how could we map these detours? How can we recreate that pathway of consciousness that DV cannot return to with her own memory? Could a computer help?

By the mid-eighties, Francisco Varela was raising a criticism about the design of artificial intelligence systems, arguing that the discipline privileged a symbol-processing model oriented toward problem-solving logic, as opposed to a connectionist model that promoted the expansion of cognitive capacities.

For example, in terms of technical design, it is not difficult to ask a computer to deliver an answer to a specific task as long as the range of possible outcomes is discrete. This is the case of a computer playing chess — an instance in which all aspects of the game's rules can be covered because, although the possibilities are multiple, they are limited. Having technological equipment that plays chess as well as the most capable chess player is a great achievement of computer science; however, this does not constitute cognition. The counterpart is the case of activities or processes with infinite variables that a computer will never be capable of executing. Varela mentions driving a car as an example: while we can incorporate a high number of variables such as traffic laws, reactions to weather conditions, and even anticipate potential accidents, it is an activity where the list of possible events becomes infinite and, in the long run, inefficient compared to the work a human driver with common sense would do.

The approach in which artificial intelligence systems operate toward problem-solving is not beneficial for the arts. An illustrative case is the performance of image generators that resort to excessive stylization. Today it is relatively easy to recognize when a portrait of a human being has been created with generative AI software: the skin is so smooth it seems almost plastic, the noses are upturned, the bodies are of indeterminate races.

If artificial intelligence cannot exercise creativity in art, it is because we are asking it to solve the wrong problem. The problem is not generating an image according to parameters, but embodying the perceptual experience. Varela is clear in explaining that it is feasible for a computational system to perform complex tasks requiring advanced training; however, it is in simple cognition where this system fails. He uses the example of a baby employing a "fundamental" intelligence when learning to speak from scattered expressions and understanding what seems to be a "sea of lights." This fundamental intelligence is what is closest to the creative act and is what LLM developers should aspire to.

Gustave Flaubert had focal epilepsy; this diagnosis has been reached through the study of his letters and the medical evaluations he received during his life. In one of his letters, the French author writes how his friends told him that during his attacks he always repeated the same phrase:

Flaubert had annoyed his friends with his insisting pseudo-mystical interests; once they even told him to please stop his obsession with the devil, as he was boring everyone. In that context, his declarations about seeing golden light were considered nothing out of character. It was years later that doctors associated his condition with epilepsy; in that era, today's medications did not exist, so Flaubert suffered from unpleasant side effects and, to make matters worse, the epileptic episodes didn't even disappear completely. In one of his letters, he expressed his desire to investigate these altered states that possessed him; however, he also admitted the profound fear he felt about pursuing that — the fear of not returning.

From the philosophical avenue of deconstruction, Jean-Paul Sartre attempted a psychoanalytic analysis of Flaubert in The Family Idiot, attributing his episodes to generic family traumas. Mario Vargas Llosa, in The Perpetual Orgy, identifies the repetitive nature of this type of analysis, explaining how this decoding erases what is singular about Flaubert — speaking of a tiresome enterprise of deciphering the mind and its replicas: a skein that has a beginning, but no end. This failure is no different from the yearning for a self-driving car: the impossible desire to access cognition solely through the symbolic.

Varela said that to advance in the field of artificial intelligence, it is necessary to invert the expert and the child on the performance scale. To study the roots of cognition, one must look at the child and the altered state of consciousness in which the human being is outlined as a self-organizing organism, despite being on a different plane of understanding compared to the functional adult.

In this spirit of privileging the simple, Varela proposes focusing on the definition of problems and not on problem-solving. Is it really so strange to ask if someone is real? To delineate what your eyes are seeing? To notice that we are with the good ones and the bad ones? They are almost religious questions that current AI services cannot address beyond the symbolic, the representational.

One summer afternoon, DV was driving on the highway in Chile with her friend Andrés; they were going to Limache to see Coni. At the wheel, DV experienced an epileptic attack and began asking her friend if he was real. It is possible to consider that if they didn't die, it was due to the evolutionarily strategic nature of cognition, an enactive brain state so powerful that, in this case, it was capable of prevailing over the epileptic glitch. Andrés was scared to death, but DV kept driving just fine, and later they returned as if nothing had happened and listened to Los Tres:

Varela considered that the seemingly infinite amounts of knowledge an artificial intelligence system needs to perform simple cognitive actions must be delivered to a computer "by the spoonful." This Sisyphean process partly explains the mediocrity of current services in producing valuable artistic works, as they offer us clichés and worn-out metaphors. When delving into these problematic aspects, ChatGPT collapses, repeating patterns and being incoherent by affirming that the image of the melting clock is powerful but then admitting its lack of originality. This problem separates AI services from living systems by failing to properly proceed with the articulation of an operational closure.

In Maturana and Varela's theory of autopoiesis, they use the concept of operational closure as one of the keys to defining an autonomous entity. The word closure is used in its sense of a sustained operation within a space of transformations and serves to characterize forms of interaction mediated by a system's autonomy. From this position, input and output logics do not correspond to the definition of an autopoietic system. In the case of poor ChatGPT producing a poem and then complicating itself trying to explain why it wrote what it wrote, it becomes evident that it cannot go beyond the reactive output — that it is using symbols and grouping them into generic schemas. ChatGPT's poem may even be of better quality than one written by a human, but even the worst of poets is embodying a process of inquiry and not merely simulating results.

In one of my latest exhibitions, I presented the series Nocturnal, which consisted of images obtained through artificial intelligence as output after sharing my dreams with it. My goal was to observe how synthetic visual production dealt with the dreamlike. Just like in the poem written by ChatGPT, the representations were initially literal and clichéd. My solution was to incorporate new elements, both algorithmic and material, that would leave me satisfied. After this experience, I learned that when dealing with the creative, with that which is not of this world, it is not advisable to explain so many things to AI services — because they tend toward symbol deciphering, grounding everything and stripping the final product of what made it captivating in the first place. It is therefore recommended to be brief and a little confusing at the moment of input, to disregard the decoding process executed by the system, to turn it back into a black box.

Regarding the aura as an altered state of consciousness, a similar approach can be proposed. The aura has a preventive function — an epileptic feels it to pay attention to the coming attack and eventually understand that it is necessary to stop what they are doing and stay safe. But beyond this precautionary medical sense, benign auras can also leave the patient with a resonance in which resides the potential to extend consciousness. There are medical reports that speak of ecstatic auras that lead to a state of peace, serenity, and improved self-knowledge.

DV's auras are brief; it is difficult for her to anticipate when they will occur. However, it will always be possible for her to retain at least one word from the sensation they have produced. This word, this simplified sensation, vague in literal terms but rich in personal associations, could be the starting point for exploring this interruption in cognition using AI services — thus establishing an inter-system dialogue that contains the characteristics of a connectionist model: one that advances from specific tasks toward creativity, from sequential and hierarchical instruction toward parallel productions.

With this perspective, we would not be infinitely delivering all the information in the world to the AI system spoonful by spoonful for it to decipher. We would be slapping it, and it would return something unexpected, almost in the form of an immune response. Through a series of ecstatic jolts, we would be promoting a human-machine alliance that constitutes an evolutionary strategy of consciousness.

In Hippocratic Greece, epilepsy was called "The Sacred Disease." There was a widespread belief that epileptics were under the influence of demonic forces and should therefore be treated by invoking religious powers. It was precisely in that era, around 400 BC, that a researcher from Hippocrates' own school determined that epilepsy is a condition that originates in the brain and, in most cases, is inherited or manifests after a blow to the head.

Since then, neurology has made great strides in the study of epilepsy and, at the same time, from within neurology itself, arguments are raised suggesting that the patient's cultural aspects must be considered when initiating treatment. Engel, for example, speaks of the problem of "arrogance" among medical professionals when dismissing elements like traditions and local philosophies — not out of a desire for liberal progressivism but because of medical evidence that for centuries has demonstrated the key influence of external stimuli, often emotional, at the moment of epileptic episodes.

Oliver Sacks, a neurologist and science communicator, complements the arguments about the benefits of framing epilepsy beyond the merely empirical. He discusses the rich tradition of popular narrative related to medical conditions, a relatively widespread practice that reached its peak in the nineteenth century and then declined with the arrival of impersonal neurological science. Sacks talks about positioning neurological patients as figures similar to those in classical fables, travelers who go to unimaginable lands — this would benefit the treatment of conditions that are rarely identical from person to person.

I consider such an invitation, which comes from the medical discipline itself, to be consistent with Francisco Varela's proposals for technological development that allows expanding the limits of consciousness. Varela explains that technology, among other things, must act as an amplifier since cognitive sciences and cognitive theory are inseparable; they are vitally and complementarily united.

Currently, with AI services becoming increasingly accessible, we might be facing an opportunity to resume the scientific-aesthetic linkage that Sacks longed for. Our first synthetic poems, our first images, may turn out ugly. They will probably be digital babblings. However, as we engage in distributed contextual conversation with the machine, we will eventually reach a creative instance that satisfies us. Varela asks how to know if a cognitive system functions properly, then answers himself that it is when we see that the emergent properties correspond to a specific cognitive aptitude: a happy solution for the required task. A work that we like, that produced an expansion in our understanding of things.

A computational artificial intelligence system is not an autopoietic machine according to Maturana and Varela's definitions. The Chileans do not admit any synthetic system as living. They are also clear in considering that living beings do not depend on interaction with an observer and that we have a complex sense of purpose that leads us to embody a history of adaptive transformations with which we can even set goals and coordinate to achieve them. From this stance, we should dismiss speculations about machine singularity but not refrain from incorporating machines into our own autopoietic systems.

Maturana and Varela themselves affirm that living systems disintegrate whenever they lose their autopoietic organization; in this sense, a prosthesis that allows a body to survive would maintain autopoietic organization. On the other hand, the Chileans also affirm that an autopoietic machine (the human body, for example) has no inputs or outputs; therefore, when it is perturbed by an external factor, we perceive this influence from the context of observation and not from our functioning as machines. This would mean that computational outputs, like the responses of a chatbot advising us on our romantic dilemmas, are not truly brain prostheses.

Fig. 2 — Explanatory outline of advanced technology in relation to the autopoietic nature of a living system.

Epilepsy is tested by measuring electricity in the brain, but throughout this essay, we have reviewed that it is evidently something more than just electricity. If we understand consciousness as lived experience, then the aura is the multisensory map that gives us the clues to reach the altered state that later cannot be accessed through memory. The aura is an instance with the necessary ambiguity for this task, being considered a neurological phenomenon that is not "all or nothing" — variable and difficult to categorize; it is not a discrete problem that would be solved with the appropriate prosthesis.

In the arts and humanities, it suits us to adopt a favorable stance regarding computational models that can recreate synthetic forms of consciousness. It is tempting to cling to the opposite side, to be conservative and insist that a machine will never do research or create works of art as well as we can. That is why DV's refuge for the next natural disaster will be her studio full of photographs and strange poems produced during her interactions with artificial intelligence during the epileptic aura. These services will not simulate ecstasy; rather, they will be this strange companion, almost a mirror that reflects in its own way and amplifies the human being's own cognitive capacities to expand consciousness.