Models of Humans
The Surprising Similarities Between AI, Spirituality, and Mental Illness
"A spirit is something like a multiply realizable generative function." John Vervaeke
Have you ever been at a family gathering and found yourself inexplicably drawn to a particular person? It's like they have some magnetic pull, an unspoken charisma that urges you to listen, learn, and emulate them. You can't quite put your finger on it, but it's there.
So, what's really going on here? It's not just that you're interested in what the person has to say. You're studying their expressions, their gestures, movements, and sentence structure. You might be doing this so intently that you find yourself missing what they actually said. You're training a homunculus model in your mind.
"Spinoza is hard... It's the most logical, rigorous thing... and you're trying to remember all the predicates and then what happens though is you get what he talks about. You see all of the whole in each premise and each premise in all of the whole, and you see all of reality like that. You see spinosistically. You take on Spinoza's perspective, rather than remembering his propositions."
"Yes, that's something like ancestor worship."
John Vervaeke x Jordan Peterson
Diving into a philosopher's ideas or listening to a meditation guru can have the same effect. In these situations, we're learning on two parallel levels: the ideas themselves and, in the simplest terms, their way of being. On the thinking level, this encompasses their mental models, core beliefs, argument construction, prioritization, and openness to new information. But it goes beyond just their mode of thinking. It also includes their speech, what catches their attention, their driving force, and how their thinking connects to their emotions and focus. It's "them": the way their mind works, their connection to the world, and key opinions. We're trying, often subconsciously, to step into the perspective from which they view the world. We're training a model of them, which we can compare to others, learn from, and, in rare cases, embody. That is, shift from training mode to generative mode.
Thanks for reading 10-Year Horizon! Subscribe for free to receive new posts.
For those unfamiliar with machine learning terms, the similarity between our brains and AI models runs deep. Imagine our brains and AI learning like buddies in school. Both absorb data, experiences, and behaviors from others—our brains create mental models, while AI uses machine learning (ML) models. They both go through a "training" phase, soaking up data to recognize patterns and abstractions. After training, they enter a "generative" mode, where they apply the learned patterns to new situations (by way of predicting). It's like studying for a test and then using that knowledge to ace it. So, both our brain models and ML models work in a predictive manner, learning and adapting to understand the world around us.1
Consider the following Zen story.
Matajuro wanted to become a great swordsman, so he sought out the renowned master Banzo. Banzo initially refused to teach him, stating that Matajuro lacked the necessary dedication. Matajuro insisted and finally convinced Banzo to take him as a student.
Banzo told Matajuro to perform menial tasks, such as cooking, cleaning, and working in the garden. No mention of swordsmanship was made for months. Desperate, Matajuro asked Banzo when the real training would begin.
Banzo replied, "You cannot see it, but you have been learning swordsmanship since the day you arrived. It is as you cook, sweep, and work in the garden that you are training."
The story is not about sword fighting. It is a parable of training an inner model of Buddha in mundane situations. That is the point of the Buddhist teaching that Buddha nature is within each of us: simply train a mini-Buddha as you go through your day, his capacity for compassion and equanimity, and you will gain the most direct access to his teachings.
In some of the most extreme examples of this ability, we might consider cases such as mediums, spiritualists, or individuals who enter trance states. These individuals demonstrate an uncanny ability to tap into the essence of another person, often transcending time and space. One such intriguing case is that of Jane Roberts, an American author who not only claimed to channel an entity named Seth, but also managed to co-write numerous books with this mysterious being. This in itself is a remarkable feat, but it doesn't stop there. What's even more astonishing is that Jane Roberts embodied prominent historical figures, such as philosopher William James and renowned painters Paul Cezanne and Rembrandt. She wrote entire books from their inner perspectives. This begs the question: how literally should we take this? Can a person truly access the memories, experiences, and inner workings of another individual, especially those from the distant past?
We don't know for certain, and the answer to this question is heavily influenced by one's fundamental understanding of reality. For those who adhere to a materialist model of reality, such experiences might simply be dismissed as impossible, thereby blocking access to these potentially transformative moments.2 To the materialist, the rationalist, and the coder, a metaphor of a deep learning model being trained in one's brain may be a lot more accessible. In any case, the ability to summon a helpful figure, whether it arrives as a spirit or a model, is clearly extraordinarily valuable.
Self-portrait as a woman, Rembrandt van Rijn, 1668
In the Harry Potter series, we have the most stereotypical example. Young wizard Harry Potter frequently communicates with the spirit of Albus Dumbledore, his late mentor, through various magical means. These interactions guide Harry through his most challenging moments, providing wisdom, insight, and encouragement that help him overcome obstacles and ultimately save the wizarding world. On the other hand, the TV series "Hannibal" features the character Will Graham, a highly empathetic profiler for the FBI. His unique ability to mentally inhabit the minds of serial killers enables him to solve complex cases. Although this ability takes a toll on his mental health, it is also an indispensable tool in understanding and capturing criminals. There are as many uses as the narratives we choose to live.
Okay, so how does one learn this trick? How can we channel an inner Spinoza, Buddha, or a serial killer on a whim? The good news is, we do this naturally. It's a completely natural process for our brain. From the moment we're born, we unconsciously sculpt a model of each of our parents, absorbing their behaviors, beliefs, and ways of being like sponges. Eventually, they become an inner voice—an inner parent.
In the movie "The Lion King," there's a scene that beautifully illustrates this innate power. Young Simba, the lion cub, is faced with a crisis of identity and responsibility after the death of his father, Mufasa. During a pivotal moment in the film, Simba encounters the wise mandrill Rafiki.
Rafiki: You're Mufasa's boy!
Simba: You knew my father?
Rafiki: Correction—I know your father.
Simba: I hate to tell you this, but… he died. A long time ago.
Rafiki: Nope. Wrong again! He's alive—and I'll show him to you. You follow old Rafiki; I know the way. Come on!
They run through the bushes and rocks until they arrive at the edge of a lake.
Rafiki: "Look down there." He touches the still water of a pool with his finger. "Tell me what you see."
Simba: "I see... my reflection."
Rafiki: "Now look harder." He stirs the water with his finger, and Simba's reflection is replaced with Mufasa's. "You see? He lives in you."
But though this skill comes naturally, it is possible that it also has a dark side. Spirit possessions as much as schizophrenia could be explain through the same model. Take the movie "A Beautiful Mind," where John Nash, a brilliant mathematician, struggles with schizophrenia. He experiences hallucinations of imaginary characters, who both help and hinder him. Throughout the film, he learns to cope with these figures and use them to his advantage, ultimately leading to groundbreaking work and personal growth. That's a well-known example of thoroughly transformative self-therapy.
Training inner humans, recognizing them, and managing their relationships means consciously engaging with the natural process we've used since birth. One way to go deeper is by exploring a set of techniques called Internal Family Systems (IFS). IFS is a powerful therapeutic model that dives deep into our psyche, identifying and understanding the various "parts" or sub-personalities within us. It offers to view one's mind as an assembly of beings. They come to life to act together in harmony or, let's be honest, sometimes discord. Each persona represents a part of the self, with its unique role, voice, and contribution to one's life. Each persona is a model meticulously trained for a certain purpose or reason. The inner child is a model. The policeman you are still arguing with in your mind one hour after he gave you a ticket is a model. That amazing person you can't stop thinking about is a model, as is the buffoon who had the nerve to dump you – and you just can't forget him.
Learning the skill of conscious modelling is one of the most efficient ways to transcend ourselves. Not into an infinite non-duality, but into other modes of being (other beings). The process of embodying 'spirits' is fascinating in its own right. However, as is the case with other abstractions, when we master multiple examples (such as deeply internalizing the models of Buddha, Spinoza, William James, and one's parents) we may perhaps unlock a higher level: an understanding of how humans are construed.
This comparison certainly has its limitations, but there is more to learn from the similarities than not - consider the Bayesian brain theory, predictive coding, and the free energy principle.