About NCI
A proposed framework for how the brain decides what reaches awareness, what that selection costs, and why it differs between people.
Two open preprints · Hypotheses stated so they can fail · Not a diagnostic system
The Name
Neurobiological Cognitive Integration names a proposed connection between biological constraints and lived experience:
Neurobiological
The circuits that regulate admission and action, and the cellular, vascular, and endocrine processes that support them.
Cognitive
The construction of percepts and alternative models from admitted signals. Information not selected for conscious use may remain available elsewhere; how it can be recovered is part of the model, not an assumed absence from the brain.
Integration
The relationship between a relatively stable processing profile (P), changing metabolic headroom (Range), and the behavior their interaction permits.
From Part 2, §1.3. On this site, NCI always means Neurobiological Cognitive Integration. The NCI Foundation is not affiliated with other organizations that use the abbreviation.
The Question
Sensory systems gather information on the order of a billion bits per second. Human behavior runs at about ten (Zheng & Meister, 2024).
The selection between the two is made below awareness. The thalamic reticular nucleus gates what is admitted and basal-ganglia loops arbitrate which action proceeds, within tens to hundreds of milliseconds, too fast for deliberation to take part (Halassa & Acsády, 2016; Frank, 2005). What reaches awareness is a compressed summary, which Part 2 calls the autonomic brief. It arrives with little record of what was left out, or why.
Why do two people given the same information come away with different summaries, each sure that theirs is simply the truth? Why can the same person, rested and exhausted, seem like two different people? How much of what we read as character, or as disorder, is really the setting of that choice and the energy available to run it? And if the setting is fixed as the brain matures, is it ever reset? Part 2, §1.1
The findings behind these questions are already published, by other laboratories, in eight fields that rarely cite one another: sensory, systems, cellular and developmental neuroscience, clinical neuropsychology, metabolic physiology, neuroimmunology, and infectious disease. The Science page sets them out with their sources. Read together, they make a picture but not yet a model: no parameters, no predictions, nothing that could be wrong. Supplying those three is what NCI does.
The Proposal
Position: the stable setting
The setting of the filter: what is usually called temperament. It is built from two cortical networks, which turn processing inward toward internal modeling or outward toward external action, and three admission tags, which flag novelty, threat, and resonance with other people. Their combinations are proposed to settle into five recurring profiles, the same five Part 1 identified from behavior. Every person is a blend.
Range: the budget that runs it
The energy left to run the setting once the brain’s fixed costs are paid. It rises and falls with sleep, stress, and inflammation, and it is why one person, rested and exhausted, can seem like two.
Four consequences follow. Every strength carries a matching blind spot. A filtered view feels like plain reality. Exhaustion should cost flexibility first and the filtering itself last. And questionnaires record how people describe the brief, not the setting underneath it.
Part 2’s most consequential proposal concerns change. Position is largely settled by the mid-twenties and held in place by physical brakes, among them perineuronal nets, which inflammation dissolves in animal studies. NCI proposes that while the brakes are off, a setting can drift toward its extremes and be held there once they re-form: a change in weighting, not learning.
Part 2 marks what it does not know in the same place it makes each claim. Four of its boundaries:
| Established by others | What NCI proposes | Not yet shown |
|---|---|---|
| The thalamic reticular nucleus gates what is admitted; basal-ganglia loops select which action proceeds (Halassa & Acsády, 2016; Frank, 2005). | Two parameters. A stable setting (Position) and a variable budget (Range). | That the two can be measured separately is a requirement, not a demonstrated fact. |
| Total brain energy use barely changes between rest and effort, and the discretionary share differs between people (Raichle, 2015; Duffy et al., 2026). | Unfunded function. Dysfunction examined as a metabolic shortfall: what is left unpaid once the fixed costs of filtering are met. | The measurements are at rest and between people. They do not establish a depletion cascade within one person. |
| Fast-spiking inhibitory interneurons are among the most energy-demanding cells in cortex; when an energy deficit is engineered into them alone, sensory gating fails (Kann, 2016; Inan et al., 2016). | An order of failure. Under load, flexibility goes first, sustained abstraction next, filtering last. | The cell-level evidence establishes vulnerability, not the order. |
| Perineuronal nets mark the close of critical periods, and removing them restores plasticity; inflammation dismantles them in animals (Carceller et al., 2023; Crapser et al., 2020; Zhang et al., 2022). | The Three Locks. Revising a settled default needs budget, structural release and attribution together. | Their joint necessity in humans is unestablished; the structural evidence comes from rodents. |
Adapted from Part 2, §1.5, which draws eleven.
In Practice
Each of these is a problem researchers already study, with the test that could count against NCI’s reading of it.
Neuroimmunology · Critical care · Neurology
“I haven’t been the same since”
Some people describe lasting change after a serious illness. In animal studies, inflammation, even when it starts outside the brain, dissolves the brakes that hold adult circuits in place. NCI proposes that a setting released this way can drift toward its extreme and be held there. The evidence so far is from rodents.
Test: Position measured before and after a documented inflammatory episode, against matched intervals without one (Part 2, §9.6).
Sleep · Fatigue · Cognitive neuroscience
Rested and exhausted, two different people
As the budget tightens, NCI predicts that switching between tasks fails first, sustained abstract reasoning next, and sensory filtering last, as disinhibition rather than slowing.
Test: Prediction 1: progressive load or sleep restriction, with high-density EEG and standard gating, switching and reasoning tasks. The paradigms exist, so the order can be tested now.
Occupational health · Nursing research
Burnout is not one thing
In high-acuity bedside care, one patient is at once the object of care and the source of acute threat. NCI predicts that this wears people down differently from work that alternates between new settings and new people.
Test: Prediction 8: validated burnout scales should show different factor structures across role-stratified cohorts, with roles defined by measured task properties, not job titles.
Personality science · Psychometrics
What a questionnaire can see
A trait questionnaire asks people to summarize the brief. Its domain scores add together parameters NCI treats as different in kind, and one sitting mixes trait with state. Part 2 specifies what an instrument that separates them would have to do; none yet exists.
Test: Prediction 7: working-memory training should not shift Position, and autonomic training should not raise the working-memory ceiling.
Cerebral energetics · Metabolic physiology
A rate, not a reserve
Total brain energy use barely changes between rest and effort; allocation does (Raichle, 2015). NCI treats the flexible share as a rate, and proposes that long-term load damages the machinery that produces it. That may be why some people recover fully from severe depletion while others level off below where they were.
Test: Prediction 10: phosphorus magnetization-transfer spectroscopy at rest and under load, then after sleep deprivation with and without creatine.
Aging · Longitudinal epidemiology
Decline that follows transitions, not birthdays
Tissue change predicts late-life decline that tracks age. Part 2 adds a second account: losing external supports predicts step-changes timed to retirement, bereavement, relocation or sensory loss, partly recoverable when what was lost is supplied another way.
Test: Within-person comparisons around a dated transition, with age-matched controls (Part 2, §9.5). Existing cohorts can decide it without new instruments.
Clinical trials · Neuromodulation
A null measured at the wrong stage
In rodents, 40 Hz light flicker restores plasticity in visual cortex; human studies have read the same kind of stimulation through recall tests, several processing stages downstream. NCI calls this the Stage Error, and names what should move instead: input-stage rates such as perceptual-learning slopes.
Test: If those rates do not change under the same stimulation, the mechanism does not transfer and the null was accurate (Part 2, §4.6).
Psychiatry · Clinical classification
Dimensions, and what generates them
NCI shares the dimensional, mechanism-first commitment of the Research Domain Criteria and proposes a candidate generator for the dimensions. It uses clinical presentations only to describe what depletion looks like downstream; it does not define or replace diagnoses.
Test: Prediction 5, in ethically approved studies: lasting change to a settled default should need budget, structural release and attribution together.
Scope
Not a diagnostic system. Part 2 states the limit in its own words:
NCI models underlying computational, metabolic, and neurobiological parameters. It does not define medical disease entities, does not substitute for clinical psychiatric or neurological assessment, and is not a clinical diagnostic instrument. Part 2, §1.2
Not advice for individuals. The monographs are written for researchers, and nobody should alter pharmacological or therapeutic treatment on the basis of a theoretical framework. Establishing biological validity, clinical usefulness, and any diagnostic or interventional application remains work ahead.
Not finished. Both parts are preprints and have not been peer reviewed. They report no new data. No validated instrument exists; Part 2 specifies one rather than reporting one.
The Monographs
NCI is set out in two monographs, both open preprints on Zenodo. They are companions: Part 1 sets out the big picture, Part 2 its mechanism. Authors and affiliations are listed in each.
Part 1: the big picture
Neurobiological Cognitive Integration (NCI): A Precision-Weighted Predictive-Processing Framework for Continuous Temperament Dimensions, Cognitive Complexity, and the Phenotype–Pathology Continuum
Five continuous dimensions, with the five profiles proposed as density peaks in that space; the Two-Axis model, separating temperament from cognitive complexity; a predicted alignment with five Big Five aspects; and the phenotype–pathology continuum.
NCI-MONO-2026-V1.1 · 66 pages · doi:10.5281/zenodo.22002985
Part 2: the mechanism
Neurobiological Cognitive Integration (NCI): The Autonomic Brief, the Budget That Constrains It, and the Life It Renders
The autonomic brief; Position and Range; the grid from which the five profiles are derived; the Three Locks on revision; ten falsifiable predictions; and a specification for a measurement instrument.
NCI-MONO-2026-02 · 196 pages · 245 references · doi:10.5281/zenodo.22258778 · Abstract and contents
The two are separately citable. Cite them by section number (§).
The Foundation
If the account holds, much of what we attribute to character and vulnerability is the setting and the energy to run it, not volition alone. People are built to differ, no one is built to be complete, and difficulty is more often a shortfall to be funded than a failure of will. Part 2, §1.1
Finding out will take methods no single group holds: electrophysiology and spectroscopy, animal models, longitudinal cohorts, psychometrics, and clinical research. The NCI Foundation exists to keep the framework open, keep each claim tied to its evidence, and bring together the researchers who can test it.
Open. The monographs are free to read on Zenodo, with permanent DOIs.
Marked. Hypotheses are labeled as hypotheses, kept apart from the findings they build on, and stated so that a specific result would count against them.
Corrected. When a claim is shown to be wrong, it changes. Corrections are recorded and carried into the next revision.
Collaborate
The findings NCI rests on are not in doubt. The architecture drawn from them is a first sketch, and better instruments will show which lines hold, which move, and which were never there. If you can test a prediction, show where an argument fails, or hold a finding that belongs in the structure, we would like to hear from you.