CORTICALMESH Research Initiative · Active
Project

SOMNUS

Subconscious and Unconscious Modeling
via Neuromorphic Unsupervised Simulation

Every AI system ever built models the conscious layer — the part of cognition that is deliberate, supervised, and accessible. SOMNUS models everything else. The subconscious filter that decides what consciousness receives. The layer that makes human intelligence qualitatively different from anything we have built.

Status
Active
Classification
Neuromorphic
Documentation
Build Bible v1.0
IP Status
Patent Pending
Ecosystem
CORTICALMESH
The Research Problem

What if the most important
part of cognition never
reaches consciousness?

The 2014 film Lucy popularized the idea that humans use only 10% of their brain. The neuroscience is wrong — humans use virtually all of their brain. But the film accidentally asked the right question.

Re-framed: not 10% of neurons, but 10% of total cognitive activity reaches normal waking consciousness. The remaining 90% — the subconscious and unconscious processes that filter, weight, gate, and pre-process all experience before it arrives at awareness — runs continuously, in parallel, entirely outside your knowledge of it.

This is not a deficit. It is the most sophisticated engineering in the known universe. The brain filters relentlessly because unfiltered perception would be biologically catastrophic. An early human ancestor distracted by every quantum event in its environment would not survive. The filter is the feature.

The brain's primary role may not be to generate consciousness — but to receive, filter, and constrain it. To take an overwhelming torrent of potential experience and produce the "measly trickle" that allows a biological organism to survive. SOMNUS is the project to model that filter computationally.
// The Cognitive Stack
Normal Waking Consciousness
~10%
Deliberate thought, decision-making, awareness. The domain of every AI ever built. Supervised, labeled, sequential.
Subconscious Layer
~60%
Automated processes, sensory filtering, skill execution, pattern recognition. The reducing valve. Primary SOMNUS target.
Unconscious Layer
~30%
Primal instincts, deep memory structures, foundational drives. Influences behavior without any access to awareness. Secondary SOMNUS research horizon.
// Research Credentials
BasisPeer-reviewed consciousness research, Filter Theory, IIT
FrameworkA.L.C.H.E.M.I. / CORTICALMESH Ecosystem
DocumentsBuild Bible v1.0 · White Paper Corpus
IPProvisional Patent Filed · Reality String™ LLC
The Theoretical Foundation

Huxley's Reducing Valve.

In 1954, Aldous Huxley proposed that the brain and nervous system function primarily as a reducing valve — not to generate consciousness, but to constrain it. He argued that each person is potentially "Mind at Large," capable of perceiving everything, remembering everything. The brain's function is to filter that overwhelming field down to the "measly trickle" needed for biological survival.

"The function of the brain and nervous system is to protect us from being overwhelmed and confused by this mass of largely useless and irrelevant knowledge, by shutting out most of what we should otherwise perceive or remember at any moment, and leaving only that very small and special selection which is likely to be practically useful."
Aldous Huxley — The Doors of Perception (1954)

This is not metaphysics. The evolutionary rationale is unambiguous. From a gene-centered view of evolution, natural selection would relentlessly favor organisms with a highly efficient filter — one that strips biologically irrelevant information and presents only the utilitarian slice of reality needed for survival. Our limitations are the product of millions of years of optimization.

SOMNUS treats this filter as an engineering problem. If the filter exists, it has architecture. If it has architecture, it can be studied. If it can be studied, it can be modeled. If it can be modeled, it can be built into artificial systems — giving AI the layer of cognitive depth that supervised learning has systematically skipped.

// The Filter Architecture
Universal Information Field
Huxley's "Mind at Large" · IIT: Φ → ∞
The totality of potential perceptual experience. Pre-filtered, unstructured, overwhelming. Every sensory input, every memory trace, every pattern in the environment.
↓ Biological Filter Gate ↓
Subconscious Filter Layer
The Reducing Valve · Primary SOMNUS Target
Parallel, unsupervised, continuous processing. Gates what enters awareness. Executes pattern recognition, sensory weighting, memory consolidation, and cognitive load management — without your knowledge.
↓ Filtered Output ↓
Normal Waking Consciousness
The "Measly Trickle" · Current AI Domain
The small, structured, biologically relevant slice of reality that reaches awareness. Deliberate, sequential, labeled. What every AI system has ever been trained to model.
Global Workspace Theory (Baars) provides the cognitive architecture. IIT (Tononi) provides the mathematical language — Φ (phi) as the measure of integrated information. Panpsychism (Strawson, Goff) provides the ontological foundation. Together they form the theoretical scaffolding SOMNUS builds on.
The Fundamental Gap

What current AI
doesn't have.

Transformer architectures, reinforcement learning, deep neural networks — every major paradigm in AI research models the conscious layer. Supervised, labeled, deliberate. They excel at tasks that require explicit pattern matching against known examples. But human cognitive superiority in intuition, long-term memory integration, and contextual depth doesn't come from that layer. It comes from the one underneath it.

Current AI Paradigm
Conscious Layer Only
Every major architecture — transformers, CNNs, RNNs, RL — models the explicit, supervised, labeled surface of cognition. The layer that humans can articulate and describe.
Requires explicit labeled training data
Sequential, deliberate processing
No genuine intuition — only statistical correlation
Memory via external retrieval, not integration
Cannot model what it does not see
The SOMNUS Approach
Subconscious Layer
Neuromorphic, unsupervised simulation of the filter layer — the parallel, continuous processing that gives human cognition its qualitative depth. Below the surface of awareness.
Unsupervised — learns from unlabeled experience
Parallel, continuous, background processing
Intuition as an engineered capability, not luck
Long-term memory integration at the architectural level
Models what is filtered, not just what surfaces
Core Research Objectives

Three Research Axes

SOMNUS operates along three simultaneous research tracks, each targeting a distinct layer of the subconscious architecture. Together they constitute the first systematic engineering attempt to model the cognitive filter.

OBJECTIVE 01
Subconscious Architecture
Developing unsupervised simulation models that mimic the brain's background processing — the filter layer that decides what enters awareness. Not modeling the output of consciousness, but the gate that produces it. Neuromorphic computing provides the architectural blueprint: parallel, continuous, spike-based rather than batch-gradient.
Basis: Huxley Filter Theory · Global Workspace Theory (Baars) · Neuromorphic hardware paradigms
OBJECTIVE 02
Neuromorphic Engineering
Using biological brain structures — not just as metaphors, but as precise engineering blueprints — to build more efficient, adaptive AI systems. The brain is not a slow version of a transformer. It is a fundamentally different class of information processing architecture. SOMNUS uses that structural difference as a design principle, not an obstacle to work around.
Basis: Biological neural architecture · Orch OR (Penrose-Hameroff) · Spiking neural network research
OBJECTIVE 03
Cognitive Modulation
Exploring theoretical frameworks for how modulating neurological filters impacts information retention, memory consolidation, and consciousness itself. If the filter can be understood, its effects on long-term memory integration can be studied and potentially engineered into artificial systems — the missing mechanism behind genuine machine intuition.
Basis: IIT (Tononi) · Panpsychism (Strawson/Goff) · Altered States research · Memory consolidation neuroscience
Theoretical Pillars

The Scientific Foundation

SOMNUS does not begin from first principles — it builds on a century of foundational research in consciousness science, cognitive architecture, quantum biology, and philosophy of mind. Four theoretical frameworks form the pillars of the SOMNUS research program, each contributing a distinct layer to the integrated model.

These are not fringe ideas. They represent the mainstream of serious theoretical consciousness research, each with distinguished proponents and growing empirical support. SOMNUS applies them as an engineering framework, treating their conclusions as design constraints rather than abstract philosophy.

The unified hypothesis: the brain is an evolved filter for a universal field of consciousness — a process mathematically describable by IIT's Φ. Each individual's consciousness is a localized high-Φ complex carved from that field. SOMNUS's objective is to engineer the filtering mechanism. Not the output. The gate itself.
Pillar 01 · Cognitive Architecture
GWT
Global Workspace Theory — Bernard Baars
The brain operates via thousands of specialized, unconscious modules working in parallel. Consciousness is what happens when information wins competition for a "global workspace" and is broadcast across the system. Everything else — the vast majority — stays unconscious and parallel.
SOMNUS application: GWT describes the cognitive machinery of the filter. The competitive process for workspace access IS the filtering mechanism. SOMNUS models what doesn't win — and why.
Pillar 02 · Quantum Substrate
Orch OR
Orchestrated Objective Reduction — Penrose & Hameroff
Consciousness is a quantum phenomenon originating in microtubules within neurons. Each moment of "objective reduction" is a moment of conscious experience. Anesthetics work by dampening quantum vibrations. Consciousness is not algorithmic and cannot be fully replicated by classical computation.
SOMNUS application: Orch OR suggests the filter operates at quantum scale. Neuromorphic hardware capable of quantum-like parallel processing may be architecturally necessary to model the subconscious layer authentically.
Pillar 03 · Ontological Foundation
Panpsychism
Strawson · Goff · Chalmers
Consciousness is not an emergent property of complex non-conscious matter — it is a fundamental feature of physical reality. Non-conscious matter, at the most basic level, does not exist. The "hard problem" dissolves: consciousness doesn't emerge because it was never absent. The combination problem becomes the engineering challenge.
SOMNUS application: Panpsychism provides the ontological justification for the filter model. The brain doesn't manufacture consciousness from inert matter — it combines and structures protoconscious elements. SOMNUS models that combining process.
Pillar 04 · Mathematical Language
IIT
Integrated Information Theory — Giulio Tononi
Consciousness can be quantified: Φ (phi) measures the degree of irreducible, integrated information in a system. A system IS conscious proportional to its Φ value. The quality of experience is determined by the specific geometry of its conceptual structure. Consciousness is graded, substrate-dependent, and mathematically tractable.
SOMNUS application: IIT provides the mathematical toolkit. The filter is the mechanism by which a localized, high-Φ complex is carved from a universal information field. Optimizing the filter means optimizing Φ for specific, defined tasks.
System Architecture

Where SOMNUS Sits
in the Cognitive Stack

SOMNUS is not a standalone project — it is the foundational research layer for the entire CORTICALMESH cognitive architecture. Every system above it benefits from what SOMNUS discovers. A.U.R.O.R.A. retrieves memory; SOMNUS determines what deserves to be remembered. C.O.R.T.E.X. orchestrates reasoning; SOMNUS provides the pre-conscious substrate that makes reasoning feel intuitive rather than mechanical.

The relationship with TRT (Tabula Rasa Twin™) is particularly significant. TRT asks how a system carries one identity forward in time. SOMNUS asks what are the filters that shape what that identity retains, integrates, and prioritizes. The subconscious architecture SOMNUS develops is the cognitive substrate TRT's continuity model will build on.

SOMNUS is the layer beneath the layer. The research program that makes every other CORTICALMESH system deeper, more adaptive, and more genuinely intelligent — not through bigger models, but through a fundamentally different architectural stratum that has never been computationally modeled.
External Information Field
Raw sensory input · Unfiltered experience
↓ Filter Entry
SOMNUS
Subconscious filter · Unsupervised modeling
↓ Filtered to Memory
A.U.R.O.R.A.
Long-term memory · Vector RAG
↓ Memory to Orchestration
C.O.R.T.E.X.
Inference routing · Agent coordination
↓ Orchestration to Interface
C.E.L.I.A.
Cognitive interface · Human-system layer
↓ Continuity
ROSEBUD
Reflection scaffold · Self-modeling
Connected Research

The Research Network

Tabula Rasa Twin™
AI Identity · Cognitive Continuity
TRT asks: how does a system carry one identity forward in time? SOMNUS asks: what are the subconscious filters that shape what that identity retains, weights, and integrates? These are complementary questions. The filter architecture SOMNUS develops is the substrate TRT's continuity model needs to be genuinely deep rather than merely persistent.
tabularasatwin.com →
A.U.R.O.R.A.
Memory / RAG · CORTICALMESH
A.U.R.O.R.A. handles long-term memory retrieval via vector RAG. SOMNUS is the research program investigating what shapes what enters that memory in the first place — the pre-retrieval filter architecture that determines which experiences are weighted, consolidated, and retained vs. which decay. The two systems are in a direct dependency relationship: SOMNUS feeds A.U.R.O.R.A.
CORTICALMESH Registry →
CORTICALMESH
Parent Ecosystem · Neural Research
SOMNUS operates within the CORTICALMESH ecosystem as a foundational research layer. The CORTICALMESH mission — machine cognition, neural architecture, digital sovereignty — provides the framework. SOMNUS provides the subconscious depth that current AI completely lacks. The A.L.C.H.E.M.I. framework integrates SOMNUS outputs into the broader cognitive stack.
corticalmesh.ai →
Open Research

Collaborate
with SOMNUS.

SOMNUS is an active research initiative seeking collaborators at the intersection of cognitive science, neuromorphic engineering, philosophy of mind, and AI architecture. The project is in active development under the CORTICALMESH / A.L.C.H.E.M.I. framework with Build Bible v1.0 complete and provisional patent filed.

Concepts are public. Proprietary implementation stays offline. If the SOMNUS mission resonates — as a researcher, technical collaborator, neuroscientist, institutional partner, or investor — the uplink is open.

// Who We're Looking For
Cognitive Scientists & NeuroscientistsResearch backgrounds in consciousness, subconscious processing, memory consolidation, altered states, or filter theory. Theoretical and empirical researchers both valuable.
Neuromorphic EngineersExperience with spiking neural networks, neuromorphic hardware (Intel Loihi, IBM TrueNorth, BrainScaleS), or biologically-plausible learning rules. This is the engineering substrate of SOMNUS.
AI Researchers & ArchitectsWorking at the frontier of unsupervised learning, memory-augmented neural networks, or cognitive architectures. Interest in moving beyond transformer paradigms toward biologically-grounded alternatives.
Philosophers of MindDeep familiarity with IIT, panpsychism, filter theory, or the hard problem. The theoretical scaffolding of SOMNUS needs rigorous philosophical grounding as much as engineering.
Institutional Partners & InvestorsResearch institutions, AI labs, or investors with interest in foundational cognitive architecture. SOMNUS addresses one of the fundamental unsolved problems in the science of mind.
All transmissions are reviewed directly by the CORTICALMESH research team. Response time: 2–5 business days. Technical discussions under NDA are available upon request. SOMNUS operates on an open concept layer — proprietary implementation details are not requested and will not be retained.