The Right to Change the Data About Me

If the brain can change, why should the algorithm keep us in the past?

In Blog 3, we stated:

the data is not you.

But there is an even more important consequence:

the data about you also has a date.

Imagine a Personalissima AI accompanying someone for ten years.

It learns where that person goes, what they read, how they write, their schedules, relationships, choices, and thousands of other regularities.

After some time, the model becomes very good.

Then the person changes.

Perhaps they learned something.

Moved to another city.

Met someone.

Lost someone.

Started a profession.

Left another.

Encountered another territory.

Grew older.

Or simply did something different.

The machine has thousands of pieces of evidence about who that person was.

But how much new evidence does it need before acknowledging who that person is becoming?

This may be one of the fundamental questions for Personalissima AI:

if the Body-Territory can change, who gives the algorithm the right to keep it trapped in its own past?

Plasticity: change is not an exception

Neuroplasticity occupies a central place in Roberto Lent’s scientific trajectory.

In a recent synthesis produced within the IBRO/IBE-UNESCO framework, Lent defines neuroplasticity as the brain’s capacity to modify its structure and function through interaction with the environment, occurring at different levels, from molecular processes to neural circuits.

This prevents an overly simplistic interpretation.

Plasticity does not mean:

“we can become anything.”

Nor does it mean that every change is beneficial.

It means something fundamental to our discussion:

the nervous system has a history without being completely frozen by its history.

What happened participates in what exists now.

But what happens now also participates in what may exist later.

For an AI that learns people, this difference is enormous.

When memory becomes a statistical prison

AI systems learn regularities because regularities are useful.

If I chose coffee for one hundred mornings, predicting coffee the next morning seems reasonable.

But tomorrow I might try tea.

The system may consider it noise.

Five days may still look like an exception.

What about twenty?

At some point, what seemed like a deviation may begin to represent a new regularity.

The technical question is:

how much weight should be given to the past, and how much to what is new?

When the model is about a person, however, the question also becomes ethical and epistemological.

Because the algorithm does not merely remember.

It can act upon what it remembers.

If it believes I like certain content, it offers me more of it.

By offering more, it changes my exposure.

My next behavior then occurs within an environment partially produced by the model of my previous behavior.

A loop emerges:

behavior → data → model → recommendation → modified environment → new behavior.

The system no longer merely predicts.

It begins to participate in the conditions of the next choice.

When the past keeps arriving in the present

Imagine someone who showed intense interest in a particular subject during a difficult period.

The algorithm learns.

Months later, that person has changed.

But the platform continues to offer the same universe of content.

The past keeps arriving in the present.

The person changed.

Their algorithmic environment may not have.

A historically accurate representation can become currently inadequate and still continue to influence the Body-Territory.

In Blog 3, we argued that the representation is not the Being.

Now we add:

an old representation should not acquire the power to make it harder for the Being to produce new differences.

Connectomic reserve: what exists without yet being manifested?

In 2026, Roberto Lent, Fernanda Tovar-Moll, and Diego Szczupak proposed a particularly provocative hypothesis: connectomic reserve.

The authors suggest that connections broadly distributed throughout the brain, some originating during development and later becoming less functionally evident, could remain available for plastic modulation or remodeling according to internal or external demands.

They explicitly state that this proposal still needs experimental testing.

Previous work by the group had already shown that our understanding of connectivity may be less simple than traditional models suggested. A 2023 study on callosal dysgenesis found significant reorganizations and considerable variability in heterotopic connections; another study showed that heterotopic connections represent a substantial proportion of callosal connectivity across different species.

We should not transform these neuroanatomical findings into proof about human identity or freedom.

But we can formulate a BrainLatam question:

how many possibilities of a Body-Territory do not appear in its data simply because they have not yet manifested?

An AI learns what has happened.

But what has happened does not necessarily contain everything that can still happen.

There is a difference between:

probability calculated from the past

and

possibility not yet manifested.

The exception may be the beginning

Imagine someone who has repeated a particular behavior a thousand times.

Then they do something different.

Statistically, it may simply be an outlier.

But that difference may also be the beginning of a transition.

The AI does not yet know.

The person may not know either.

A question therefore emerges:

should Personalissima AI quickly eliminate exceptions, or preserve them long enough to discover whether some of them represent the beginning of a new regularity?

Not every exception is transformation.

But every transformation must, at some point, produce some difference from what was happening before.

That difference needs room to exist.

Do not erase the past — temporalize it

Here we must avoid the opposite extreme.

The right to change the data about me does not mean:

“erase everything that no longer corresponds to the person I believe I am.”

The past happened.

Legitimate primary records have historical value.

Perhaps the solution lies less in erasing and more in temporalizing.

A Personalissima AI could distinguish:

what happened → when it happened → how often → whether it is still happening → which inference was produced → whether that inference is still supported by current data.

Instead of:

“You prefer X.”

we might have:

“Between 2023 and 2025, X appeared as a recurring preference. Over the last six months, this pattern has significantly decreased.”

The first statement brings behavior closer to identity.

The second describes a trajectory.

And the Body-Territory is also a trajectory.

Being able to say: “this changed”

We now return to the question that originated this series:

each person’s data needs to return to each person, in each person’s own language, so that each person can interact with and change the data mined from them.

“Changing the data” needs to be understood carefully.

It does not mean falsifying the primary record.

It means being able to say:

“This happened, but it no longer represents my current situation.”

“This inference has lost its context.”

“This behavior belonged to that period.”

“Consider this new information.”

“Show me why this classification still applies.”

Personalissima AI would need to preserve simultaneously:

enough memory not to falsify history

and

enough plasticity not to transform history into destiny.

Body-Territory: it is not only the brain that changes

There is another consequence.

Plasticity does not occur independently of experience and environment. Lent’s own definition emphasizes the brain’s interaction with the environment.

This is why we speak of Body-Territory.

The territory changes.

Relationships change.

Work changes.

Food changes.

The available words change.

Technologies change.

Institutions change.

And new differences become available for the body to perceive.

A Personalissima AI perhaps should not ask only:

“What changed in you?”

It may also need to ask:

“What changed around you?”

There is a fundamental difference here between building a digital profile and accompanying a Body-Territory.

A 2023 Brazilian study involving Lent offers a useful warning against easy reductions: when comparing men with different educational levels, researchers found no significant differences in the absolute numbers of cells, neurons, and non-neuronal cells in the medial temporal lobe attributable to schooling.

Relationships between experience, brain, and trajectory cannot be reduced to a single environmental variable.

BrainLatam Hypothesis: a computational right to plasticity

We therefore arrive at the fourth hypothesis of W41/2026:

Personalissima AI should preserve not only the memory of the Body-Territory, but also its permanent possibility of producing differences that contradict the model constructed from its past.

For now, we call this:

a computational right to plasticity.

It is not the right to falsify the past.

It is the right not to be statistically condemned to repeat it.

The AI may say:

“For a long time, you followed this path.”

But it must remain capable of noticing:

“Perhaps another path is beginning to emerge.”

And this begins to take us beyond the individual.

Because no one changes completely alone.

We change through relationships.

We learn from others.

We are affected by institutions, territories, languages, economies, and technologies.

If every Body-Territory has the right to produce new differences, how can we build something collective without transforming the regularity of the majority into an obligation for each individual?

How can we belong without having to reduce ourselves in order to fit?

This is where the word we have been approaching since the beginning of this series must move to the center:

Jiwasa — “We / Us”: to feel that we belong without having to reduce the Being.

We belong to and are part of the State and its institutions, yet each Body-Territory continues to contain new possibilities of being the State.

Perhaps a collective future does not depend on discovering how to make everyone follow the same path.

Perhaps it depends on building a State capable of perceiving when new paths begin to emerge.

Commented References — Latin America, post-2021

LENT, Roberto; TOVAR-MOLL, Fernanda; SZCZUPAK, Diego. The secret connections of the brain: a connectomic reserve for neuroplasticity? Brain, 2026.
The central reference for this essay. It proposes the connectomic reserve as a hypothesis related to neuroplasticity and cognitive reserve: distributed connections may remain available for remodeling or modulation in response to internal and external demands. The authors emphasize that the hypothesis still requires experimental validation.

LENT, Roberto. Neuroplasticity, for better and for worse. IBRO / IBE-UNESCO, 2025.
Presents neuroplasticity as the brain’s capacity to alter structure and function through interaction with the environment and across different levels of organization. It provides the basis for our distinction between having a history and being condemned to repeat it.

SZCZUPAK, Diego; LENT, Roberto; TOVAR-MOLL, Fernanda; SILVA, Afonso C. Heterotopic connectivity of callosal dysgenesis in mice and humans. Frontiers in Neuroscience, 2023.
Shows changes and considerable variability in heterotopic connections in brains with callosal dysgenesis. For this essay, it is relevant as evidence that different structural organizations and processes of reorganization challenge excessively rigid models of connectivity.

SZCZUPAK, Diego et al. The relevance of heterotopic callosal fibers to interhemispheric connectivity of the mammalian brain. Cerebral Cortex, 2022/2023.
The study found a predominance of heterotopic callosal connections in mice, marmosets, and humans, expanding a view traditionally centered on homotopic connections. It reminds us that even consolidated scientific maps remain open to revision when new measurements become available.

DE MORAIS, Viviane A. Carvalho et al. Resilience of Neural Cellularity to the Influence of Low Educational Level. Brain Sciences, 2023.
A Brazilian study involving Roberto Lent that investigated education and the cellular composition of the medial temporal lobe. It found no significant differences in absolute numbers of cells, neurons, and non-neuronal cells attributable to education in the groups studied. It is useful as a warning against simplistic causal relationships between a single environmental variable and a brain characteristic.

W41/2026 — CEPID: From Personalissima AI to the Collective Future

BrainLatam Hypothesis: Personalissima AI needs memory without transforming memory into a sentence. If the Body-Territory can change, its computational model must remain open not only to what it has been, but also to the differences that begin to announce what it may still become.