Reproducibility in fNIRS

When can I trust the hemodynamic curve I see?

(First-Person Consciousness • Decolonial Neuroscience • Brain Bee • The Feeling-and-Knowing Taá)

The Feeling-and-Knowing Taá — opening a decolonial crack

I sit in front of a screen full of fNIRS traces.
Tiny red and blue waves sobe e descem along the prefrontal cortex. On paper they look precise, “scientific”, quase perfeitas.

But my corpo feels something else:
on some days the curves look clean and stable,
on others they look nervous, noisy, almost febrile.

Inside, a question appears as Taá — sentir antes de saber:

Can I really trust this curve?
Or am I just believing in a pretty drawing?

When I allow this question to expand, another layer emerges:

I notice how even my language about data was colonized.
How I learned to treat the body as “noise”, variability as “error”, and only what fits the model as “truth”.
In this mood, it’s easy to forget that every hemodynamic curve is a living river of blood, oxygen, CO₂, posture, emotion, história de vida.

And yet, most of the time, scientific training whispers:

“Ignore the rest. Fit the model. Trust the p-value.”

It is here que a fresta de descolonização se abre:
when I feel my own desconforto, I see that there is no real separation between Neuroscience, Politics and Spirituality — where spirituality is Utupe, Xapiri, living memory, not dogma.
The way I choose to call “noise” or “signal” is already a political act about which corpos count as real.

Every study on reproducibility in fNIRS becomes então uma oportunidade:
or I use it to reforçar a lente colonial (variability = defect),
or I use it to perguntar de novo:

What does this variability tell me about the living body and about the worlds we are building in Latin America?

The study in focus

This blog comments on a 2025 NeuroImage paper that investigates:

“Variability and reproducibility of fNIRS measurements across sessions and participants.”
(keywords for search: NeuroImage 2025 fNIRS reproducibility variability across-session cross-participant short-channels GLM hemodynamic response)

The authors ask, de forma direta:

If I measure the same person several times, and different people in the same task, how stable are the fNIRS signals really?

This is not a “side question”.
It is a foundational one: if the method is not reproducible, qualquer interpretação sobre atenção, cognição, emoção ou Zona 2 cai junto.

The core scientific question

In first-person terms, the question é assim:

“If I come back to the lab tomorrow, will my brain show the same pattern for the same task?
And if another person does the task, será que a curva se parece com a minha ou não?”

The study desmonta essa questão em três níveis:

1. Within-subject reproducibility
   
   

• Same person, different sessions.
  
  

2. Between-subject variability
   
   

• Different people, mesma tarefa.
  
  

3. Impact of preprocessing choices
   
   

• How much do GLM, short-channels, HRF modeling, ICA/PCA and filtering change the final picture?
  
  

Methods and analysis (Brain Bee–friendly)

The paper uses a robust fNIRS pipeline that hoje é quase obrigatório se quisermos falar em ciência séria:

• GLM (General Linear Model)
  The authors model the hemodynamic response using a GLM:
  
  

• Regressors for each condition of the task;
  
  

• Nuisance regressors for physiological noise and drifts;
  
  

• Betas that quantify how much each channel “responds” to the task.
  
  

• HRF (Hemodynamic Response Function)
  Em vez de assumir uma HRF única e fixa para todos, eles exploram como a shape (formato) da HRF muda:
  
  

• amplitude,
  
  

• time-to-peak,
  
  

• width and undershoot.
  
  

• Short-channels
  Curto caminho da luz, medindo principalmente sinais superficiais (pele, vasos extracorticais).
  Esses canais são usados como regressores para remover:
  
  

• blood pressure oscillations,
  
  

• systemic noise,
  
  

• scalp contamination.
  
  

• ICA and PCA
  
  

• ICA (Independent Component Analysis) separa componentes independentes do sinal:
  
  

• some clearly related to motion artifacts, breathing, heart beat;
  
  

• others candidate to “true” cortical activation.
  
  

• PCA (Principal Component Analysis) ajuda a reduzir dimensionalidade, identificar padrões globais e estimar variabilidade dominante across sessions.
  
  

• Reliability metrics
  
  

• test–retest correlations,
  
  

• intraclass correlation coefficients (ICC),
  
  

• spatial overlap of activation maps between sessions and between participants.
  
  

Tudo isso é apresentado com uma pergunta simples por trás:

How much of what we call “activation” is stable across time and people?

Main findings: reproducible, but not mechanical

Os resultados não são nem apocalípticos nem triunfalistas. Eles mostram que:

• Amplitude absolute of signals can vary bastante between sessions, mesmo na mesma pessoa.
  
  

• Relative patterns (which regions respond more than others, shape of the HRF) tend to be more stable, especially when:
  
  

• short-channels are properly used,
  
  

• GLM is bem especificado,
  
  

• ICA/PCA são aplicados com critério.
  
  

Em outras palavras:

fNIRS is not a chaotic, useless method.
But it is also not a rigid machine where every curve should look igual.

Reproducibility appears as coherent variation, not exact repetition.

Reading the results with our concepts

1. Damasian Mind and living variability

From a Damasian Mind perspective, cada sessão mede um corpo num dia diferente:

• different interoception,
  
  

• different proprioception,
  
  

• different emotional background,
  
  

• different metabolic state.
  
  

So, variability across sessions não é defeito; é o próprio metabolismo existencial se reorganizando.

2. Quorum Sensing Humano (QSH)

If humans are QSH systems — regulating-se mutuamente — then even small social changes:

• who is in the room,
  
  

• tone of the experimenter,
  
  

• institutional setting,
  
  

podem alterar a hemodinâmica pré-frontal.
The study mostra, sem usar esse nome, que QSH está sempre “vazando” nos dados.

3. Eus Tensionais and task demands

Cada tarefa convoca Eus Tensionais diferentes:

• a “task-self” that tries to perform;
  
  

• um “self” cansado;
  
  

• um self ansioso;
  
  

• um self curioso.
  
  

A reproducibility limited partly reflects which Eu Tensonal is being activated in each session. A mesma tarefa cognitiva não convoca o mesmo Eu se o contexto mudou.

4. Zones 1, 2, and 3

• Zone 1: automation — pode gerar respostas mais rápidas e stereotyped, mas nem sempre mais saudáveis.
  
  

• Zone 2: fruição, presença, criatividade — talvez HRFs mais suaves, estáveis, coerentes.
  
  

• Zone 3: ideologia, medo, rigidez — curves tensas, reativas, talvez mais ruidosas.
  
  

The paper não fala em zonas, mas os padrões que chama de “variabilidade” podem ser lidos assim: como mudanças de zona existencial.

5. DANA – DNA as living intelligence

Different HRF shapes across individuals lembram que:

DANA (DNA intelligence) não é blueprint rígido,
é um sistema fluido que produz corpos diversos, com ritmos hemodinâmicos singulares.

A ciência colonial tende a chamar isso de inter-subject noise.
Nós preferimos chamar de diversidade viva.

6. Yãy hã mĩy (Maxakali) — imitar-se para compreender

Origem Maxakali: imitar o animal que se quer caçar.
Aqui, “imitar-se” significa:

• ajustar nossos modelos aos dados reais,
  
  

• em vez de forçar o corpo a caber na nossa fórmula.
  
  

Modelos mais flexíveis de HRF e pipelines conscientes de variabilidade são um Yãy hã mĩy computacional: a ciência tentando imitar melhor o corpo.

How this science adjusts our views

Antes, poderíamos pensar:

• If the method was really good, every session would give the same curve.
  
  

Depois desse tipo de trabalho, fica mais honesto dizer:

• The method can be robust,
  
  

• but human physiology is plastic,
  
  

• and reproducibility precisa ser definida como coerência dentro da complexidade, não como repetição cega.
  
  

Isso desmonta a fantasia colonial de corpo-máquina perfeita
e nos obriga a ver o organismo como sistema complexo, sensível, historicamente situado.

Normative implications for education, health, city and politics in LATAM

1. Education with fNIRS and physiology
   
   

• School programs that usam fNIRS para avaliar atenção precisam de múltiplas sessões e protocolos rigorosos.
  
  

• Uma única medição não define “déficit” de ninguém.
  
  

2. Clinical use in our public health systems
   
   

• In SUS-like contexts, protocolos devem considerar variabilidade intra-sujeito antes de rotular alguém como “impairment”.
  
  

3. Neuro-rights and data justice
   
   

• Variability cannot be usado contra pessoas como “falta de disciplina” ou “falta de capacidade”.
  
  

• Leis precisam proteger corpos que fogem do padrão como legítimos, não como erro.
  
  

4. Urban and labor policies
   
   

• Cities and workplaces that reduzem estresse fisiológico (ruído, calor, precariedade) aumentam reproducibility não apenas dos dados, mas da própria capacidade cognitiva cotidiana das pessoas.
  
  

Keywords for scientific search

“NeuroImage 2025 fNIRS reproducibility variability across-sessions cross-participants GLM short-channels HRF ICA PCA test–retest reliability”

When Two Brains Receive the Same World - Cooperation, synchrony, and the shared rhythm of attention

Embodied Singing -Voice, interoception, and Body-Territory in vocal expertise

Pleasant Odors and the Breath that Organizes Us - How smell organizes brain–body coupling

Architecture That Thinks With Me - Turning corners and the attentional cost of built environments

Auditory Approach Bias From Birth - How newborns and adults code the desire to listen

Beta Waves and the Moment I Truly Decide - The prefrontal cortex as the space where "feeling" becomes "choosing"

How My Brain Encodes Voice in Midlife - F0, listening effort, and the vitality of human hearing

Learning Beside Another Brain - Hyperscanning and the pedagogy of co-presence

Reproducibility in fNIRS - When can I trust the hemodynamic curve I see?

HRfunc and the True Shape of the Hemodynamic Response - Why every brain breathes light in its own way

Mixed Reality and Decision-Making - How the brain evaluates prototypes and hybrid worlds

Intense Exercise and the Awakening of Zone 2 - The hemodynamics of effort and the body that generates intelligence

Buttoning a Shirt - Everyday actions as windows into attention, gesture, and consciousness

Depression, tDCS, and the Prefrontal Cortex - Reigniting silent circuits

Designing fNIRS Studies in Real-World Environments - Why science must step outside the laboratory to exist

Transformers and Virtual Short-Channels - AI cleaning brain signals and retelling hemodynamics

Mental Fatigue and Performance - When the head gives up before the body

Cold Water and the Brain - Oxygenation, cold, and the consciousness of the limit

Walking After Stroke - Cognitive–motor interference in everyday life

Balance and the Cerebellum in Parkinson’s Disease - Movement, tensions, and reorganization of the Body-Territory

Freezing of Gait and the Loss of the Body’s Own Quorum - When the body stops trusting the next step

Children With Cochlear Implants - Learning to hear through the brain, not just the device

Emotional Processing in Children With Oppositional Behavior - Regulation, conflict, and the birth of Tensional Selves

Mild Cognitive Impairment - Early hemodynamic signs and presence in the world

Pain, Apathy, and Depression in Dementia - When feeling and thinking stop walking together

Cognitive Load - How much does fNIRS really feel my mental effort?

The Brain in Daily Life -Assisted horsemanship, sport, and embodied enjoyment

Linguistic Jiwasa - When language thinks the world

Dialogical Multiplication and Indigenous Psychology - How to let psychology listen without erasing the Other

The Feeling and Knowing Taá of Christmas 

Republican Capitalism of Spirits without Bodies

NIRS fNIRS EEG ERP Multimodal NIRS-EEG

#Decolonial
#Neuroscience
#NIRSfNIRS
#Multimodal
#NIRSEEG
#Jiwasa
#Taa
#CBDCdeVarejo
#DREX
#DREXcidadão