Can a Model of the Brain Replace the Experience of Being Alive?

When Words Begin to Trap Us

BrainLatam Special Series — SBNeC 2026

Let us begin with a map.

In 1946, the Argentine writer Jorge Luis Borges, in On Exactitude in Science, imagined an empire whose cartographers became so obsessed with precision that they created a map exactly the same size as the territory.

The map coincided with the Empire point by point.

And precisely because of that, it became useless.

Perhaps there is an important question here for Neuroscience:

How much does a model need to leave out in order to remain a model?

A word reduces.

An EEG reduces.

A brain image reduces.

An equation reduces.

An artificial intelligence reduces.

That is not necessarily a failure.

It is what makes representation possible.

The problem begins when we forget that:

representation and experience are not the same thing.

The word “pain” does not hurt

We can describe pain.

Give it a score from 0 to 10.

Measure brain activity, heart rate, and autonomic responses.

We can even build models capable of recognizing patterns associated with painful states.

But where, exactly, is the feeling of pain?

The word does not hurt.

The number does not hurt.

The brain image does not hurt.

And an AI capable of writing a perfect description of suffering does not demonstrate, by that fact alone, that it has suffered.

The point is not to diminish the value of models.

It is to ask:

What can we explain about an experience without confusing the explanation with what was lived?

We are building increasingly sophisticated maps of consciousness

Latin American researchers such as Enzo Tagliazucchi, associated with Argentina, and Rodrigo Cofré, in Chile, have been using mathematics and computational modeling to investigate properties associated with different levels of consciousness.

Recent studies examine temporal irreversibility, entropy, brain dynamics, and personalized models in disorders of consciousness.

These studies show the enormous power of Computational Neuroscience.

We can compare states.

Find regularities.

Develop biomarkers.

Perhaps even predict characteristics of certain experiences.

But we are still performing one operation:

modeling.

So we can return to Borges:

If we could represent every neuron, every synapse, every memory, every bodily state, and every relationship with the environment, would we have reproduced an experience — or merely built an extraordinarily detailed map of it?

Today, Science has no definitive answer.

And perhaps it is important not to pretend that it does.

The brain does not arrive alone

The Argentine researcher Ezequiel Di Paolo, through enactive approaches, helps us widen the question.

Cognition and experience do not need to be understood only as the internal processing of representations.

Body, action, environment, and relationships also participate in the production of meaning.

This connects directly with Body-Territory.

If we remove:

body;

history;

relationships;

territory;

and keep only a brain model,

how much of the experience we wanted to explain is still there?

The cut may be necessary for study.

But that does not mean what we isolate exists in isolation while we live.

Archaeology has lived with this limit for a long time

Archaeologists work with fragments.

Ceramics.

Charcoal.

Seeds.

Modified soils.

Marks on the landscape.

From these traces, they construct models of worlds they can no longer directly observe.

The Brazilian archaeologist Eduardo Góes Neves, in Sob os tempos do equinócio (2022), brought together decades of evidence that help revise older representations of the Amazon as sparsely occupied and only lightly transformed by Indigenous societies.

Miguel Aparício, Claide Moraes, Anne Rapp Py-Daniel, and Neves have also shown how archaeology, ethnology, and ecology reveal much more complex histories of interaction between societies and forests.

The Amazon did not change when these studies were published.

Our map of it changed.

That is an important lesson.

A model can produce true knowledge without becoming what it represents.

Archaeology can know a great deal about a past society.

But it cannot return to the experience of waking up in that village.

Neuroscience may need to preserve the same humility.

What if the map begins to modify the territory?

Here a second question appears.

Jean Baudrillard returned to Borges’s image to think about situations in which representations stop merely describing reality and begin to participate in what is experienced as real.

We do not need to accept his entire theory to recognize something simple.

A diagnosis is a representation.

But it can change how someone perceives themselves.

A financial score is a model.

But it can determine access to credit.

An algorithm constructs a probabilistic representation of a person.

Then it uses that representation to decide what the person will see.

So we get:

Body-Territory
→ data
→ model
→ selection
→ new experience
→ new data.

The map has returned to the territory.

Brazilian anthropologist Letícia Cesarino helps us understand these recursive relationships among people, algorithms, and digital environments.

This allows us to formulate an important distinction:

An AI does not need to feel what it represents for its representation to participate in what we will feel.

When words begin to trap us

Perhaps now we can return to the title of this series.

First there is an experience.

Then we give it a name.

Then we create a category.

Then a model.

And when the model works very well, we may begin to see the experience only through it.

Danilo Silva Guimarães helps us remember that categories can be useful without fully coinciding with the worlds of experience they attempt to represent.

Ailton Krenak also offers an important window.

Compare:

water resource

with:

a river that participates in the continuity of life.

They may refer to the same water.

But they produce different representations of our relationship with it.

No word contains the entire river.

Perhaps no word needs to.

Decolonial Neuroscience is not about abandoning models

We need EEG.

fNIRS.

MRI.

Equations.

AI.

Categories.

Models.

Decolonial Neuroscience does not propose destroying them.

It proposes that we continue asking:

What can this model see?

What did it need to leave out?

And what happens when we return this representation to the Body-Territory it claims to represent?

Because there is a fundamental difference between:

using a model to understand life

and

demanding that life fit inside the model.

BrainLatam Question

If we build a model capable of predicting everything we will say, everything we will do, and every brain pattern associated with what we feel, will we finally have reproduced the experience of being alive — or will we simply have reached Borges’s perfect map?

References

Borges, J. L. (1946). Del rigor en la ciencia / On Exactitude in Science.

Neves, E. G. (2022). Sob os tempos do equinócio: oito mil anos de história da Amazônia Central.

Krenak, A. (2022). Futuro ancestral.

Cesarino, L. (2022). O mundo do avesso: Verdade e política na era digital.

Guimarães, D. S. (2022). A Tarefa Histórica da Psicologia Indígena diante dos 60 anos da Regulamentação da Psicologia no Brasil.

Gilson, M.; Tagliazucchi, E.; Cofré, R. (2023). Entropy production of multivariate Ornstein-Uhlenbeck processes correlates with consciousness levels in the human brain. Physical Review E, 107, 024121.

Di Paolo, E. A.; Lawler, D.; Vaccari, A. P. (2023). Toward an Enactive Conception of Productive Practices: Beyond Material Agency. Philosophy & Technology, 36, 31.

Luppi, A. I.; Cofré, R.; Tagliazucchi, E.; Raimondo, F.; Orio, P.; Sanz Perl, Y.; Sitt, J. D. et al. (2023). Computational modelling in disorders of consciousness. NeuroImage, 275, 120162.

Aparício, M.; Moraes, C. de P.; Py-Daniel, A. R.; Neves, E. G. (2024). Amazônia em simbiose: marcas de humanidades que enfrentam o Antropoceno.

Complementary philosophical reference

Baudrillard, J. (1981). Simulacres et Simulation.