The Network Watches You While Teaching You How to Look

AI, Rock–Paper–Scissors, and the social semiosis we are building together

Before talking about artificial intelligence, come back with me a few years.

Two children put their hands behind their backs.

Rock. Paper. Scissors. Shoot.

Paper beats Rock. Rock beats Scissors. Scissors beat Paper.

It looks like nothing more than a game. But notice what a child has to understand: no choice is absolutely superior. The response that wins now may lose in the next relationship.

In a study by Jie Gao, Yanjie Su, Masaki Tomonaga, and Tetsuro Matsuzawa, 38 children between 35 and 71 months of age learned precisely this circular structure. The experiment was not simply testing whether they knew how to play. It investigated whether they could solve a problem involving circular relations by combining paper > rock, rock > scissors, and scissors > paper. Performance in the condition in which the three relations were mixed improved with age and rose above chance from approximately 50 months, suggesting that this ability becomes more consistent around four years of age. (pubmed.ncbi.nlm.nih.gov)

Perhaps you can already see with us why this children's game matters to BrainLatam.

Not because a child is literally “activating three connectomes.”

But because, very early in life, the child encounters a small ecology of strategies in which:

No response wins in every encounter.

This may be one of the best entry points into Rock, Scissors, and Paper.

Rock is not a mistake. Rock can also be competence

BrainLatam uses Rock, Scissors, and Paper as operational metaphors for different functional modes of the Body-Territory, not as three independent anatomical structures in the brain. Rock approaches habit, rapid execution, and already embodied repertoires; Scissors, analysis, comparison, and deliberate planning; Paper, flow, broadened attention, and metacognition. (brainlatam.com)

Stop for a moment.

Did you have to consciously think about each finger movement required to scroll this page?

Probably not.

Fortunately.

If we had to deliberately reconstruct every movement we had already learned, living would be almost impossible.

In BrainLatam's extended reading of Yãy hã mĩy, observing, imitating, repeating, and embodying allow something initially effortful to become an available repertoire. In childhood, this process appears with particular clarity: before explaining, the child observes; before mastering, the child repeats; later, what was repeated may become gesture, language, skill, and belonging. (brainlatam.com)

Rock, therefore, can be high performance.

The musician does not calculate every finger movement.

The athlete does not reconstruct every action.

A child who has finally learned a rule does not need to rediscover it in every round.

The problem begins when the territory changes and Rock continues responding as though nothing had changed.

The automatism that was competence yesterday may become a local optimum tomorrow.

That is when confidence can harden into repetition, and repetition may approach what BrainLatam metaphorically calls blind faith. (brainlatam.com)

Now your phone vibrates

Do not look yet.

Or look.

That choice is already part of our experiment.

A study published in Biological Psychology used EEG to investigate what happens when smartphone notifications precede a cognitive-control task. Participants were assigned either to a brief mindfulness induction or to a neutral narration. In trials preceded by notifications, researchers found lower theta power and higher alpha and beta power, along with lower frontocentral activity associated with attentional shifting. In the mindfulness condition, some theta-related indices and the theta/beta ratio were compatible with greater cognitive-control engagement, particularly after notifications. The authors present these results as preliminary evidence that a brief intervention may buffer some of the effects of notifications on cognitive-control processes. (sciencedirect.com)

The study did not measure Rock, Scissors, or Paper.

Mindfulness is not synonymous with Paper.

But it allows us to ask a question that matters to our framework:

Does the same notification encounter the same Body-Territory when we are executing a habit, deliberating about it, or perceiving our own automatism?

In Rock, there may already be a familiar path:

vibration → picked it up → opened it.

In Scissors, something may interrupt the sequence:

“I am doing something else. Do I actually need to open this now?”

And in Paper:

“Interesting. My body was already preparing to pick up the phone before I decided.”

Here an important distinction appears.

The notification can acquire Qualia — it can pulse, disturb, awaken curiosity — without necessarily taking over the attentional channel.

What shines does not have to command the next action.

Perhaps agency is precisely the ability to preserve the possibility of changing moves.

What if the network learns which move you are in?

Now make a small shift in perspective with us.

For years, we have asked whether platforms can discover what we like.

But imagine that they can learn something even more useful:

when we respond quickly;

when we hesitate;

how long we remain;

when we return;

which type of stimulus interrupts another task;

which type of content receives an almost automatic click.

We do not need to imagine an AI reading thoughts.

Our actions only need to leave patterns.

We can then build what BrainLatam calls here the Algorithmic Semiotic Loop:

perception → Qualia/attention → action → data → algorithmic inference → new cut → new perception.

You encounter the feed.

But your encounter also modifies the next feed.

The network does not merely learn what you look at. It participates in selecting what you may be able to look at next.

But the algorithm does not have to reinforce Rock

Here there is an important opening for agency.

Khambatta and colleagues developed recommendation systems that personalized content according to users' current or ideal preferences. In a preregistered experiment with 6,488 participants, recommendations based on current preferences produced more clicks; recommendations aligned with the preferences people themselves wished to have generated fewer clicks, but a greater sense that their time had been well spent and that the service had benefited them. (nature.com)

Notice the difference.

The algorithm can ask:

“What makes this person click?”

Or:

“What direction does this person say they want to build for themselves?”

That changes our criticism profoundly.

The question is not simply:

algorithm versus autonomy.

It is:

Which Body-Territory are we teaching the algorithm to reinforce?

If the system optimizes only the most immediately available response, Rock can acquire a structural advantage.

But that is not technological destiny.

It is a design decision.

And now AI begins to write between us

Something new is happening on social networks.

Previously, algorithms mainly selected signs produced by other people.

Now AI also helps produce them.

In a 2026 experiment with 680 participants organized into small groups on an experimental social platform, different AI tools provided conversational assistance, response suggestions, discussion prompts, or feedback on comments. Some conditions increased participation and content production, but could also reduce perceived quality and authenticity and generate effects on people who were not even directly using the tool. (nature.com)

So answer with us:

If an AI suggested the sentence and I chose to send it, who spoke?

Me?

The AI?

Both of us?

And if you respond to that sentence, who are you responding to?

At this point, AI is no longer merely a filter of the environment.

It enters the circulation of the social sign itself.

Persuasion may also be a relationship between bodies

A dual-fNIRS study makes this question even more interesting.

Li and colleagues simultaneously measured persuaders and receivers during a naturalistic persuasion task. Arguments that successfully persuaded were associated with greater neural coupling in certain combinations of regions than unsuccessful arguments. Coupling measures also added predictive information beyond self-report. (academic.oup.com)

This does not mean:

synchrony = persuasion.

And it certainly does not mean that fNIRS measures meaning or Qualia.

But it allows a new question:

If human persuasion happens within a relationship between two Body-Territories, what changes when the message is written, personalized, or suggested by AI?

Perhaps Jiwasa is precisely here

Look at where we have arrived.

We began with two children playing.

No piece always wins.

Then we realized that our own functional modes should not possess permanent sovereignty either.

Rock can execute.

Scissors can interrupt and compare.

Paper can perceive the very mode in which perception is happening.

Now you and I have built a question that was not fully formed at the beginning of this text.

This is the direction in which BrainLatam uses Jiwasa: not as two consciousnesses becoming one, but as the possibility that something may emerge between us that neither participant possessed alone.

And now we place AI inside the circuit.

The question is no longer merely:

“Can AI persuade?”

It becomes:

When AI enters between us, does it merely offer signs — or does it begin to participate in the conditions through which we will produce the next signs together?

NeuroDesafio LATAM research question

We can test this.

The same message could be presented in three conditions: attributed to another human, attributed to AI, or explicitly personalized by AI.

The text could be identical.

What would change would be the existential declaration of the sign.

EEG could track the temporal dynamics of attention and updating. Dual-fNIRS could compare human-human interaction with AI-mediated conditions. ECG, breathing, and electrodermal activity could add other windows onto 5D Materiality. First-person reports could ask what acquired Qualia, where the attentional channel remained, and whether the participant perceived their own automatic response.

And we could add a fourth question:

At what moment did you realize that you could still choose another move?

Perhaps this is the question a child already begins to learn when discovering that Rock does not beat everything.

The network may know our habits.

AI may learn our clicks.

It may even predict which sign has the greatest chance of finding an already available path.

But as long as we can perceive that path, compare it, and create another one together with someone else, the story is not finished.

Rock learns a path. Scissors asks whether it still works. Paper notices that other paths exist.

And Jiwasa adds:

perhaps some paths only appear when we begin looking for them together.

Commented References

Gao, J.; Su, Y.; Tomonaga, M.; Matsuzawa, T. (2018). “Learning the rules of the rock–paper–scissors game: chimpanzees versus children”. Primates, 59(1), 7–17. DOI: 10.1007/s10329-017-0620-0.
Children aged 35 to 71 months were tested on learning the circular Rock–Paper–Scissors relation. Performance in mixed pairs rose above chance from around 50 months. Its relevance to BrainLatam is not to demonstrate “Rock–Paper–Scissors connectomes,” but to show that a relational structure with no absolute winner can already be learned in early childhood. (pubmed.ncbi.nlm.nih.gov)

Upshaw, J. D. et al. (2024). “Electrophysiological effects of smartphone notifications on cognitive control following a brief mindfulness induction”. Biological Psychology, 185, 108725.
Shows that preceding notifications altered electrophysiological measures related to cognitive control and that a brief mindfulness induction modulated some of those effects. It offers an experimental bridge for asking whether the effect of a sign also depends on the functional state in which it encounters the Body-Territory. (sciencedirect.com)

Khambatta, P.; Mariadassou, S.; Morris, J.; Wheeler, S. C. (2023). “Tailoring recommendation algorithms to ideal preferences makes users better off”. Scientific Reports, 13, 9325.
Across 6,488 participants, recommendations based on ideal preferences produced fewer clicks than recommendations based on current preferences, but greater perceived benefit and better-spent time. It shows that maximum engagement is not the only possible value function for algorithmic personalization. (nature.com)

Li, Y.; Luo, X.; Wang, K.; et al. (2023). “Persuader-receiver neural coupling underlies persuasive messaging and predicts persuasion outcome”. Cerebral Cortex, 33(11), 6818–6833.
Used dual-fNIRS during persuasive interaction and found associations between interpersonal neural coupling and persuasive success. It provides the basis for asking whether human messages, AI-attributed messages, or AI-personalized messages produce different neurophysiological relations. (academic.oup.com)

Møller, A. G.; Romero, D. M.; Jurgens, D.; et al. (2026). “The impact of generative AI on social media: an experimental study”. Scientific Reports, 16, 9376.
An experiment involving 680 people showing that generative tools can alter participation, production, quality, and perceived authenticity in social conversations. It reinforces that AI is no longer acting only by selecting signs: it is beginning to participate in their collective production. (nature.com)

BrainLatam — Rock, Scissors, Paper; Extended Yãy hã mĩy; 5D Materiality; and Jiwasa.
In the BrainLatam formulation, Rock, Scissors, and Paper are operational metaphors, not established anatomical modules: Rock emphasizes rapid and embodied repertoire; Scissors, deliberate analysis; Paper, flow and metacognition. Extended Yãy hã mĩy describes learning through observation, repetition, embodiment, and the later possibility of transformation; Jiwasa asks what may emerge when a relationship creates possibilities that no Body-Territory possessed alone. (brainlatam.com)