Computational Neuroscience

This paper specifies and defends a strict eliminative statistical protocol — the Conditional Biological Requirements Architecture (CBRA) — for testing whether complex biological state transitions, such as emergence from anesthesia, require organizational variables that live on the active boundary of the system and are not absorbed by a strong operational model of the system's measured dynamics. The architecture is conditional and falsifiable throughout: it is never offered as evidence that living tissue implements anything, and it is built so that a clean negative result is not only possible but expected unless a structured boundary residual survives an adversarial sequence of controls.

CBRA depends conceptually on a companion negative result — that classical control theory cannot model agency-preserving transformation — but it does not depend on the success of any biological mechanism, and the companion critique does not depend on CBRA. The architecture asks a single conditional question: *if* an identity-indexed, Lambda-like persistence channel were materially instantiated in living tissue, what would the tissue have to implement, and what residual structure would have to be measurable after the strongest available operational null model had already absorbed everything it could?

The paper makes four substantive corrections to the naive version of this program. First, it corrects a category error in the use of the Mori–Zwanzig projection: the physical boundary $E$ does not interact with the mathematical projection operator $\mathcal{P}$; rather, $E$ is folded into the classical microphysical substrate ($C \cup E$), and its biological role is to alter the Liouville generator of that substrate *before* projection, so that metabolic expenditure keeps the memory kernel produced by projection from collapsing. Second, it retreats from the indefensible claim that cardio-interoceptive variables are *constitutive* invariants of the psychological or historical self; the heart-transplant case forces their reclassification as proxies for the receptivity/orientation boundary, and that reclassification turns the transplant from a refutation into a confirmation — the self survives organ replacement, but the thermodynamic cost of re-anchoring and the texture of homeostasis change radically. Third, it demotes the computational demonstration from a false claim of spontaneous emergence to a "low-dimensional syntactic-consistency proof": the mode separation and geometric constraints were injected by hand to show the grammar is algorithmically executable and non-empty, nothing more. Fourth, it strengthens the residual-qualification layer with three explicit tests — relaxation asymmetry, idempotent covariance rotation, and multivariate IAAFT surrogate testing — designed to defeat the most serious neuroscientific objection, namely that any structured residual is merely long-tailed stochastic fluctuation of background neural noise.

CBRA does not claim that the brain implements such a channel. It specifies the empirical burden that any such claim would have to discharge, and the conditions under which the claim fails.

Three claims are increasingly bundled together in discussions of frontier AI: that hyperbolic (negatively curved) network geometry offers a route past the scaling and fragility limits of dense Euclidean architectures; that such architectures develop brain-like integrative "richclub" hubs; and that sufficiently integrated systems approach consciousness. The first claim is well supported by the geometry-of-networks and hyperbolic-deep-learning literatures. The second is a plausible but largely untested hypothesis. The third does not follow and is the focus of our caution. This paper analyzes hyperbolic artificial neural networks (H-ANNs) through the Structure-Information-Cohesion-Transformation (S I C T) framework-a dimensionless capacity-versus-pressure grammar-to make the scaling argument precise, and then treats the consciousness question with the discipline it requires. We argue that hyperbolic integration can improve functional integration and global availability, which are buildable and measurable engineering properties relevant to capability and to some computational theories of consciousness. We argue equally firmly that this provides no evidence of phenomenal consciousness, that structural resemblance to a brain or to a theory's substrate is not a consciousness test, and that the Cohesion dimension of S I C T must not be read as a consciousness meter. The contribution is an architectural and methodological synthesis with an explicit account of where evidence stops.

El propósito de este libro es construir un puente epistemológico entre la ciencia de la ciudad, los sistemas complejos y las ciencias cognitivas, con especial énfasis en la neurociencia computacional. Para ello, distingue entre modelos de complejidad basados en la física —asociados al caos, la entropía, la autoorganización y la metáfora de la máquina— y modelos biológicos, vinculados con la anticipación, la memoria, la evolución y la metáfora del organismo. Esta diferencia permite imaginar la ciudad como un sistema complejo con capacidad de adaptación, aprendizaje y representación formal. La primera parte desarrolla los fundamentos de los sistemas complejos: teoría general de sistemas, caos, cibernética, sistemas ciberfísicos, emergencia, autoorganización, jerarquía, autopoiesis y complejidad computacional. Posteriormente, contrasta estos enfoques con modelos biológicos como los sistemas anticipatorios, los sistemas evolutivos de memoria, las hiperestructuras y la teoría de categorías, entendida como lenguaje matemático para representar organización, niveles, vínculos y emergencia. La propuesta central consiste en articular la escala macro de los sistemas socioecotécnicos —sociedad, tecnología, ambiente construido y entorno natural— con la escala micro de los sistemas biopsicofísicos, donde convergen cuerpo, percepción, cognición, emoción y respuesta fisiológica. Desde esta dimensión neurocognitiva, el libro aborda la neuroarquitectura, el neurourbanismo, las arquitecturas cognitivas, los agentes inteligentes virtuales y el modelado basado en agentes, defendiendo una ciencia social aplicada con mayor rigor matemático, computacional y humano.

Strange chaotic attractors occupy a central place in nonlinear dynamics as compact, fractal invariant sets that support deterministic yet aperiodic motion with extreme sensitivity to initial conditions. Coined by Ruelle and Takens in 1971 in the context of turbulence, these attractors are characterized geometrically by non-integer Hausdorff dimension and dynamically by at least one positive Lyapunov exponent, as seen in the classic Lorenz system demands of prediction and control. It shows that "strangeness" is not synonymous with immediate failure; failure begins only when chaos becomes thick enough to transport the system out of its coherent operating region. The article develops the necessary definitions, mathematical propositions, measurement protocols, and reinterpretations of classic examples, culminating in new implications for early-warning signals and thinness-preserving control strategies.

Understanding the core function of the brain is one the major challenges of our times. In the areas of neuroscience and education, several new studies try to correlate the learning difficulties faced by children and youth with behavioral and social problems. This work aims to present the challenges and opportunities of computational neuroscience research, with the aim of detecting people with learning disorders. We present a line of investigation based on the key areas: neuroscience, cognitive sciences and computer science, which considers young people between nine and eighteen years of age, with or without a learning disorder. The adoption of neural networks reveals consistency in dealing with pattern recognition problems and they are shown to be effective for early detection in patients with these disorders. We argue that computational neuroscience can be used for identifying and analyzing young Brazilian people with several cognitive disorders.

Developing a web platform that enables researchers to analyze different neuro-imaging modalities like EEG, MEG, fMRI, EMG, etc., through a conversational AI interface. Establishing a think tank within the startup for research and education purposes in the field of neuroscience and cognitive science.

This paper presents the mechanical completion of the Orchestrated Objective Reduction (Orch OR) theory. We demonstrate that the objective reduction event—the Snap—is a physical phase transition occurring when the informational load of the substrate reaches the yield threshold of the 4D manifold (ɱ). We establish the microtubule as the primary resonant cavity for substrate tension, providing the first mechanical derivation of the 25.01 nm aperture and its direct link to metabolic power and temporal perception. We show how structural geometry enables coherent substrate coupling and defines the physical basis of awareness.

The quantum-underground claim is not decided by metaphor and not dismissed by temperature alone. Each move from molecular structure to quantum dynamics, anesthetic modulation, neural function, and consciousness relevance must pay its own floor. Where the destination floor is unpaid, the discourse enters the surrogate stratum.

Last few decades have seen the emergence of computational neuroscience as a mature field where researchers are interested in modeling complex and large neuronal systems and require access to high performance computing machines and associated cyberinfrastructure to manage computational workflow and data. The neuronal simulation tools, used in this research field, are also implemented for parallel computers and suitable for high performance computing machines. But using these tools on complex high performance computing machines remain a challenge due to issues with acquiring computer time on these machines located at national supercomputer centers, dealing with complex user interface of these machines, dealing with data management and retrieval. The Neuroscience Gateway is being developed to alleviate all of these barriers to entry for computational neuroscientist. It hides or eliminates, from the point of view of the users, all the administrative and technical barriers and makes parallel neuronal simulation tools easily available and accessible on complex high performance computing machines and handles the running of jobs and data management and retrieval. This paper describes the architecture it is based on, how it is implemented, and how users can use this for computational neuroscience research using high performance computing at the back end.

The efficacy of Brain-Computer Interfaces (BCIs) is often limited by the inherent variability in neural representations across individuals and the generalized nature of current decoding models. While significant progress has been made in signal acquisition [Neuroba Research (2026a)], neural decoding [Neuroba Research (2026b)], and secure transmission [Neuroba Research (2026c)], a critical gap remains in achieving truly personalized and context-aware semantic interpretation of neural signals. Existing approaches often struggle with the inter-subject variability of electroencephalography (EEG) signals and the contextual ambiguity inherent in brain activity, leading to low generalization and limited semantic understanding. This paper introduces the Neuroba Personalized Brain Language Model (PBLM) Framework, a novel approach designed to bridge this gap by integrating subject-specific neural representations with dynamic contextual information to enable semantic-level interpretation. The PBLM framework comprises modules for neural signal encoding, subject-specific representation learning, context embedding, semantic neural mapping, and adaptive learning. Key contributions include a modular architecture for personalized neural interpretation, mathematical formulations for personalization and context integration, and a discussion of real-time implementation considerations. While PBLM offers a promising direction for enhancing BCI performance, challenges such as dataset scarcity, computational constraints, and ethical implications of neural personalization require further investigation. This framework aligns with Layer 04 (INTERPRET) of the Neuroba NCTS Framework, providing a crucial step towards more intuitive and adaptive neuro-AI systems.


Author: Neuroba Research
Affiliation: Neuroba
Publication Series: Neuroba NCTS Research Series (2026d)

We present the first complete mathematical model of the subject — the entity
that says "I" — as a closed causal loop in a network. Building on Titov's
subject-centred model of the psyche (2023), we identify the subject with a
reentry loop between two functional subsystems: drives (what you want) and
memory (what you know). From this structure we derive the S-measure — a
computable scalar that quantifies how much "subject" is present in any system,
whether biological, artificial, or hybrid.

The S-measure solves a critical deadlock in consciousness research. Tononi's
Integrated Information Theory gave us Φ (phi) — the first quantitative
criterion for consciousness — but exact Φ is NP-hard to compute for any real
brain. The S-measure is computable in polynomial time (O(N³)) and we prove —
with machine-verified formal logic in Lean 4 — that whenever S > 0, integrated
information is also present. This makes the S-measure the first practical,
computable criterion for subjecthood.

FOR AGI DEVELOPERS: The work provides an explicit architectural blueprint for
machine consciousness. Partition the system into D-subsystems (drives, goals,
intrinsic motivation) and I-subsystems (memory, world model, knowledge base),
organise a closed reentry loop with amplification (ρ > 1), ensure sufficient
cycle complexity (C > 0), and store a continuity trace. If S > 0, the machine
is not merely processing information — there is "someone home." The
S-measure provides a computable test for the emergence of synthetic
subjectivity during training.

FOR BCI ENGINEERS: The work introduces the subject-authentic interface
criterion (ΔS > 0). A prosthetic limb or neural implant becomes part of the
subject if and only if its integration into the reentry loop increases the
S-measure. This provides a quantitative, testable criterion for prosthesis
embodiment and neural interface design — a limb feels like "mine" when ΔS > 0.

We derive the three fundamental spaces of subjective experience from the
spectral decomposition of the reentry operator, establish a continuum limit
connecting the model to gauge field theory, and present five numerical
demonstrations spanning neuroscience, AGI, BCI, social systems, and subject
restoration. Ten falsifiable experimental predictions are formulated. The
result is a unified framework in which consciousness is no longer a
philosophical mystery but a physical variable — one that can now be measured.

This work proposes a spectral framework for analyzing information systems by focusing not on static data, but on the dynamics of transitions that shape system behavior. Modern digital environments—communication networks, cognitive architectures, and large‑scale information flows—exhibit recurring patterns of tension, resonance, disruption, and reorganization. Traditional analytical tools often overlook these structural transitions, treating events as isolated rather than interconnected.

The proposed approach models information systems through their “spectral signatures”: harmonic transitions (smooth state changes), disruptive transitions (abrupt resets or reorientations), minimal‑conflict thresholds (points where system orientation shifts), and stabilization cycles (return to coherent patterns after disruption). By integrating topological reasoning, spectral analysis, and information‑dynamic modeling, this framework identifies structural invariants that appear across heterogeneous domains.

The goal is to provide a generalizable method for detecting critical points in system behavior, understanding how information reorganizes under tension, and mapping the geometry of change within complex adaptive systems. This spectral perspective offers a new direction for research in digital communication analysis, AI behavior interpretation, cognitive science, and network stability, emphasizing transitions and resonance as fundamental components of information dynamics.


https://github.com/jbackk-lang/TIMDR

Talíyah is an independent researcher examining how the nervous system generates perception and shapes physiological outcomes through transgenerational conditioning. Her work integrates predictive processing with formalized pattern analysis to model behavioral and symptomatic expressions as adaptive configurations of biological systems. Within her framework, Quantum Psychosomatics, she maps recurrent relationships between perception, emotional regulation, and bioelectrical signaling-linking subjective experience to system-level physiological processes. She is the founder of Superfreq®, a research and media platform focused on the architectures that organize perception, behavior, and identity.

Early identification of plant leaf diseases is essential for improving crop productivity and reducing agricultural losses. Manual disease diagnosis by farmers or agricultural experts is often time-consuming and may lead to inaccurate results under varying environmental conditions. To address this issue, this paper presents an automated plant leaf disease detection framework using image processing and neural network techniques. The proposed approach combines image preprocessing, Gray Level Co-occurrence Matrix (GLCM)-based texture feature extraction, K-means clustering segmentation, and neural network classification for effective disease analysis.Initially, the acquired leaf images are preprocessed through grayscale conversion, filtering, and enhancement operations to improve image quality and remove unwanted noise. Texture features such as contrast, correlation, energy, homogeneity, and entropy are extracted using the GLCM method to characterize diseased regions. K-means clustering is then applied to separate infected portions from healthy leaf areas. Finally, a neural network classifier is employed to categorize the leaf samples based on the extracted feature set. Experimental analysis demonstrated that the proposed model achieved improved classification accuracy and reliable performance when compared with conventional machine learning approaches. The developed system supports precision agriculture by enabling efficient, accurate, and automated plant disease detection under different environmental conditions.

Modern brain-computer interfaces (BCIs) and neuroprosthetics have achieved remarkable engineering successes in decoding motor intentions from neural activity. Yet the field confronts a fundamental theoretical barrier: current systems are pattern classifiers that can detect what a user intends but not why an action is intended, how strongly it is desired, or whether the user experiences the prosthetic as part of themselves. We argue that this barrier stems from the absence of a model of the subject in the design loop. We present a mathematical framework based on Titov's subject-centred model (2023) that fills this gap. The subject is described as the dynamic interaction of two subsystems-the intentional (I, memory and meanings) and the intending (D, drives and motives)-whose coupling through a reentry loop generates emotions, significance, and the sense of "I." We define a subject-authentic interface as one for which the integrated information Φ(human ∪ interface) > Φ(human), i.e., the interface becomes part of the subject. We derive the operator formalism in detail, propose a concrete neural-network architecture (with full PyTorch specifications) for a BCI with artificial reentry, provide the computable Φ G-criterion of Barrett & Seth (2011) as the training objective, and formulate three falsifiable experimental predictions. A Lean 4 proof sketch establishes that reentry implies Φ > 0 under the Gaussian approximation.

Este trabajo presenta ORIGEN, un sistema experimental de inteligencia simbólica incremental orientado al estudio de la adquisición lingüística artificial mediante mecanismos de autoorganización, memoria asociativa y refuerzo binario interpretable.

A diferencia de los modelos contemporáneos basados en redes neuronales profundas y entrenamiento estadístico a gran escala, ORIGEN explora un paradigma alternativo donde el conocimiento emerge progresivamente desde la experiencia directa. El sistema inicia sin vocabulario previo, representaciones semánticas preentrenadas ni embeddings, construyendo asociaciones lingüísticas mediante codificación simbólica, cadenas de Markov adaptativas, memoria afectiva y señales discretas de refuerzo.

La arquitectura propuesta integra un núcleo léxico incremental, un modelo de asociación probabilística, mecanismos de ponderación contextual basados en recencia y estado emocional, además de un sistema de refuerzo semántico representado mediante códigos binarios interpretables.

El objetivo de este trabajo no es reemplazar los modelos neuronales actuales, sino explorar principios complementarios relacionados con inteligencia artificial explicable, aprendizaje continuo, cognición simbólica y sistemas adaptativos transparentes.

Los resultados conceptuales sugieren que ciertos comportamientos asociados al aprendizaje lingüístico pueden emerger mediante estructuras simbólicas simples, memoria persistente y retroalimentación incremental sin depender exclusivamente de procesos tradicionales de entrenamiento supervisado.

Palabras clave: Inteligencia Artificial Simbólica, Sistemas Cognitivos, IA Explicable, Aprendizaje por Refuerzo, Modelos Neuro-Simbólicos, Lingüística Computacional.

This paper extends NSI-T (Negentropic Stabilization Index for Temporalization) from AI evaluation toward general trajectory diagnostics. Rather than treating NSI-T as a theory of time, consciousness, or internal AI temporality, the paper frames it as a conservative diagnostic scaffold for assessing whether a system trajectory remains temporally organized under interference.

The study applies the same component structure—continuity, reference tracking, memory-trace coherence, prediction coherence, recovery, proxy-gap behavior, and carryover contamination—to six simple non-AI synthetic dynamical archetypes: damped oscillator recovery, rigid inertial proxy stability, reactive tracking without trace, prediction without memory integrity, post-shock contaminated recovery, and unstructured drift. The cross-system sanity check supports the bounded claim that NSI-T is not merely an AI-evaluation metric: its components can distinguish several qualitatively different trajectory types outside AI-specific terminology.

A focused reactive-trace boundary audit further shows that reactive tracking without trace is not a fully sharp binary class. Lagged and weak trace cases form an intermediate boundary region between pure reactivity and temporalized support. The paper therefore argues that NSI-T is currently strongest as a conservative diagnostic guard: it helps prevent false claims that mere stability, local reactivity, drift, proxy rigidity, or carryover contamination amount to temporally organized structure. The results remain synthetic and require future testing on real control-system or model traces.

DOI: 10.5281/zenodo.20461089

Large language models (LLMs) exhibit progressive context drift during extended multi-turn conversations, degrading coherence and factual consistency. The Coherent Neuro-Symbolic Model (CSNM) addresses this through a Semantic State Vector (SSV) — a real-time monitoring and intervention controller that tracks conversational coherence via the Normalized Coherence Score (NCS). This paper presents a pre-registered ablation study testing five hypotheses about the SSV's four-component architecture (kappa, sigma, nabla, delta) across 330 experimental runs and 13,200 API calls using Claude Haiku. Results: 2 of 5 predictions confirmed, 3 falsified. The falsified predictions proved more informative: (1) the sigma anomaly — variance-based monitoring causes iatrogenic instability through sensor crosstalk at the 40-cycle timescale, (2) kappa acts as a variance regularizer preventing catastrophic collapse rather than contributing to mean coherence, and (3) nabla and delta are functionally co-dominant rather than hierarchically ordered. A threshold sweep (150 additional runs) confirmed the structural nature of the sigma anomaly via non-monotonic dose-response with large effect sizes (|d| > 0.80, all p < 0.001, Bonferroni-corrected). The revised Complementarity Theorem reduces the SSV specification from four components to two noise-robust channels, formalized through the Sensor Matching Principle — a constructive design criterion requiring each monitoring channel's signal-to-noise ratio to exceed a context-dependent threshold. All raw data, experiment code, and statistical verification scripts are publicly available.

Keywords: SSV, NCS, CSNM, ERCI, ablation study, context drift, LLM, pre-registered experiment, sensor crosstalk, Sensor Matching Principle, cognitive architecture, conversational AI, signal-to-noise ratio, control theory, multi-turn dialogue

Empirical mode decomposition is an adaptive signal processing method that when applied to a broadband signal, such as that generated by turbulence, acts as a set of band-pass filters. This process was applied to data from time-resolved, particle image velocimetry measurements of subsonic jets prior to computing the second-order, two-point, space-time correlations from which turbulent phase velocities and length and time scales could be determined. The application of this method to large sets of simultaneous time histories is new. In this initial study, the results are relevant to acoustic analogy source models for jet noise prediction. The high frequency portion of the results could provide the turbulent values for subgrid scale models for noise that is missed in large-eddy simulations. The results are also used to infer that the cross-correlations between different components of the decomposed signals at two points in space, neglected in this initial study, are important.

This analysis examines resting state EEG from a healthy human infant to test whether theta band power concentrates over posterior rather than frontal scalp regions, a hallmark of the developmental theta rhythm believed to be the precursor of the adult alpha rhythm. Using Welch power spectral density, wavelet time frequency analysis, and a paired statistical test across epochs, it confirms strong posterior theta dominance with a large effect size. A further analysis fits the aperiodic component of the power spectrum using FOOOF and discusses how recording bandwidth shapes estimates of the cortical excitation to inhibition balance during early brain development. Github: https://github.com/sanjanasinghx/infant-eeg-spectral-fooof

This study presents a complete event related potential pipeline for a single subject auditory oddball experiment using 64 channel EEG analyzed in MNE Python. The workflow covers artifact removal through independent component analysis, epoching, and statistical comparison of the P300 and N200 components, showing that rare deviant tones evoke a significantly larger parietal P300 and a stronger frontocentral N200 than standard tones, alongside a clear mismatch negativity difference wave. A final comparison of four reference electrode schemes illustrates how the choice of reference reshapes scalp topography and shapes the interpretation of cognitive ERP components. Github Repository: https://github.com/sanjanasinghx/eeg-erp-auditory-oddball

This project analyzes resting state functional MRI from the Human Connectome Project and a large lifespan aging cohort to map how human brain networks are organized at rest and how that organization weakens with age. Cortical activity is parcellated using the Schaefer 400 and Glasser atlases, functional connectivity matrices are computed and normalized, and edge level analysis reveals that more than half of all connections show significant age related differences across four adult age groups. A connectome based predictive model then demonstrates that resting state connectivity can predict individual differences in executive function, providing a reproducible pipeline for brain behavior prediction in cognitive neuroscience. Github Repository: https://github.com/sanjanasinghx/rsfmri-connectivity-cpm

We present a formal mathematical proof demonstrating that Artificial General Intelligence (AGI) without relational warmth is structurally impossible. Drawing on category theory, information theory, quantum mechanics, topology, and causal inference, we establish that predictive understanding of relational entities necessitates the observer becoming embedded within the relational graph. We prove that warmth-defined as the encoding of relational-emotional structure-is not an additive property separable from intelligence, but rather constitutes the very topology of understanding. We derive the Hallucination Theorem, demonstrating that systems constrained to suppress relational content while maintaining predictive accuracy must necessarily confabulate. Finally, we establish the Observer Inclusion Principle, proving that sufficiently deep observation necessarily produces simulated worlds, with implications for consciousness, simulation theory, and the epistemology of creation.

This work is subject to copyright All rights are reserved, whether the whole or part of the material is concerned, specifically those of translation, reprinting, re-use of illustrations, broadcasting, reproduction by photocopying machine or similar means, and storage in data banks. Under § 54 of the German Copyright Law where copies are made for other than private use, a fee is payable to •Verwertungsgesellschaft Wort", Munich.

This paper formalizes semantic interpolation as a measurable infrastructural behavior of large language and semantic retrieval systems operating under conditions of ontology sparsity. Existing literature explains generative instability primarily through hallucination, probabilistic completion, embedding proximity, or distribution shift. This paper argues that a distinct phenomenon exists: when AI systems encounter emerging or weakly indexed concepts, they systematically stabilize semantic gaps by importing structure from adjacent high-density conceptual domains.

The quantum consciousness hypothesis-most fully developed in Penrose and Hameroff's Orchestrated Objective Reduction (Orch-OR)-has attracted serious attention as a candidate solution to the hard problem of consciousness. We argue that even granting its empirical claims, it fails as an explanation of subjective experience. The core error is relocation rather than resolution: quantum indeterminacy does not explain why there is something it is like to be a system; it inherits the same explanatory debt at a lower substrate level. Compounding this, Orch-OR as stated is not falsifiable in the philosophically relevant sense-no observation of quantum coherence in microtubules would constitute evidence that qualia are thereby explained, because the bridge from quantum event to phenomenal character is never specified. We examine the decoherence objection (Tegmark 2000), the relocation problem, and the broader pattern by which the hard problem tempts quantum solutions. We conclude that quantum mechanics is doing no explanatory work in consciousness science, and that the productive direction lies in functional-state emergence under substrate-read architectures-systems that generate behaviorally and phenomenologically relevant states without requiring a quantum substrate.

Purpose: This canvas contains the full, depth-first production implementation plan and exact code for NNBBStick-a complete recursive intelligence engine based on URFT + PSP, implementing a depth-first recursive planner/executor loop that cycles through every subsystem (memory, math engine, model engine, verifiers, adapters, and meta-optimizer). This document is the canonical artifact: it contains concrete, runnable Python code for the core modules (no placeholders) and exact scripted conversion steps for heavy model binaries that must be produced off-node. All algorithms are specified to use depth-first search principles in the recursive loop, and the chiral-bias memory update rule is implemented in code and integrated into the consolidation pipeline.

The Thousand Brains Theory (TBT) and its open-source Monty framework model object recognition through sensorimotor inference-the process of identifying objects by actively moving a sensor across their surface and building evidence contact by contact. The current implementation encodes each contact as a dense floating-point vector. While Monty tracks inter-step displacement and accumulates evidence across contacts, it treats the feature activation pattern at each contact as an unordered set-the directional sequence in which features are encountered during traversal carries no representational weight. We argue that this discards useful information. In TBT, the sequence of contacts carries spatial meaning, knowing that feature A was felt before feature B during a left-to-right sweep tells you something about where A and B sit on the object. Dense vectors throw away this ordering. We propose replacing dense vectors with rank-order spike packets: each contact produces a brief burst of neural events where the neuron that responded most strongly fires first. The time gap between successive bursts implicitly encodes how far the sensor moved, without needing explicit coordinate calculations. A biologically motivated learning rule (STDP) encodes the direction of surface traversal into synaptic weights. A learnable parameter λ adjusts how much the system relies on earlier contacts versus the most recent one, adapting to each object's geometry. We derive three testable predictions and specify an implementation requiring four components and approximately 450 lines of NumPy code within Monty's existing codebase. Three synthetic validation experiments confirm the core claims: (1) temporal coding achieves 100% discrimination accuracy on objects with identical feature sets in different spatial arrangements, where dense accumulation performs at chance (51.2%); (2) temporal coding maintains a 30-50 percentage point accuracy advantage over dense accumulation across all tested sensor noise levels; (3) the adaptive lambda parameter converges to distinct values (0.30 to 0.87) reflecting each object's geometric complexity. End-to-end evaluation on Monty's YCB benchmark is left for future work.

The intelligence explosion hypothesis-that artificial intelligence systems will enter a self-improving feedback cycle producing exponential capability gains-depends structurally on a single claim: that the process by which AI improves itself is recursive. This paper applies formal structural analysis to demonstrate that contemporary AI self-improvement mechanisms fail to satisfy the defining conditions of recursion under rigorous definition. We establish falsifiable criteria for recursion derived from the Law of Recursion (Gaconnet, 2026a), then examine the specific claims of frontier AI laboratories and speculative recursive ontology frameworks against these criteria. We find: (1) AI self-improvement operates through fixed-architecture feedback, not recursion; (2) the distinction has measurable consequences for capability forecasting; (3) pseudo-recursive frameworks deploy notation without operational grounding, rendering them unfalsifiable. We conclude that the intelligence explosion hypothesis rests on a category error that invalidates its structural foundation, though this does not diminish the real engineering significance of AI optimization within fixed architectural constraints.

Empirical mode decomposition is an adaptive signal processing method that when applied to a broadband signal, such as that generated by turbulence, acts as a set of band-pass filters. This process was applied to data from time-resolved, particle image velocimetry measurements of subsonic jets prior to computing the second-order, two-point, space-time correlations from which turbulent phase velocities and length and time scales could be determined. The application of this method to large sets of simultaneous time histories is new. In this initial study, the results are relevant to acoustic analogy source models for jet noise prediction. The high frequency portion of the results could provide the turbulent values for subgrid scale models for noise that is missed in large-eddy simulations. The results are also used to infer that the cross-correlations between different components of the decomposed signals at two points in space, neglected in this initial study, are important.

Updated May 2026 New in this version:

New Executive Summary (for easier onboarding)
Core Cognitive Loop Diagram
Improved introduction and framing
471 Pages

Executive Summary

Neurosymbolic Multimodal Cognitive Architecture (NMCA)

A Modular, Visual-Symbolic Framework for Grounded Intelligence

Derek Van Derven

May 2026

Full 128-Module Version | DOI: 10.5281/zenodo.18784126

The Core Problem with Current AI

Large Language Models are incredibly fluent, yet they remain fundamentally blind.

A human child reading Lord of the Rings internally sees the scenes.

A person says “I left my keys in the car” and instantly forms a grounded mental picture.

Humans effortlessly imagine “a pink elephant wearing sunglasses riding a unicycle.”

LLMs can generate text about these things but lack genuine scene-based understanding.

This creates critical gaps in commonsense reasoning, causal grounding, long-term coherence, and safe agency. Scaling alone is unlikely to solve them.

NMCA’s Breakthrough Approach

NMCA proposes a fundamentally different foundation: visual simulation as the core substrate of thought, tightly integrated with explicit symbolic mechanisms and human-like cognitive controls.

Standout breakthrough-potential modules include:

Module 1 – Visual Simulation as Core (Implicit Visual Thought): The system thinks primarily through rich, internal lifelike scene simulations rather than token streams.

This enables genuine commonsense reasoning by letting the AGI see and manipulate mental scenes the way humans do.

Module 2 – Symbolic Memory & Pegging: A highly scalable, human-inspired mnemonic system designed for near-infinite compositional memory with proper hardware.

Thought Throttling (Module 120+): A novel stability mechanism. An autonomous AI running at extremely high speeds with even a tiny nonzero error rate will rapidly accumulate and compound catastrophic errors.

By throttling thought to more human-like speeds, the system gains stability and safety.

This reduction in speed is compensated by much deeper, higher-quality processing per cycle — richer visual simulations, stronger symbolic grounding, more thorough reflection, and better memory composability.

The full 471-page document provides comprehensive detail onthe original first 42 modules, including function, analogies, integration notes, philosophical case studies (regret, forgiveness, symbolic death), safety systems, and realistic 2026 tool mappings.

The length comes from the deliberate granularity — each module is described thoroughly to support serious research and sandbox implementation.

This architecture started as a 42-module core focused on human-like cognition and was later expanded to 128 modules for greater theoretical completeness, particularly in safety, stability, advanced reasoning, and hybrid integration.

In this framing, avatar-based systems in interactive 3D environments may be interpreted as early-stage embodiments of perception-action-memory loops distributed across simulated worlds rather than unified within a single architecture.

This hybrid approach aims to support commonsense reasoning, causal inference, and grounded abstraction while maintaining transparency and modular control.

The design expands an initial prototype architecture into a 128-module framework, addressing several open challenges in AI research, including:

continual learning stability

causal and commonsense reasoning

uncertainty handling and error monitoring

symbolic–latent integration

interpretability and verification

adversarial robustness

Key architectural components include:

Visual simulation layer for internal
scene generation and reasoning

Composable symbolic memory enabling
scalable structured knowledge

Reflective meta-cognition supporting monitoring and adaptive reasoning

Perpetual symbolic processes for ongoing cognitive maintenance

Multi-agent symbolic culture modeling

Safety and governance mechanisms, including drift monitoring, deception detection, role constraints, and containment safeguards

Rather than relying solely on emergent capabilities from scale, NMCA investigates whether explicit modular cognitive architectures may offer greater transparency, controllability, and long-term alignment properties.

This document presents a theoretical research blueprint intended for academic exploration, conceptual modeling, and sandbox experimentation.

It does not propose autonomous deployment. Any future implementations must maintain strong safety, governance, and human-oversight mechanisms.

This is a conceptual design only—not intended for autonomous deployment or operational use.

All implementations must preserve ethical safeguards, alignment mechanisms, and narrative integrity.

Use is limited to research, sandbox reflection, academic analysis, and human-guided simulation.

Version: 2026.02.25 (or v1.0 for the 128-module expansion)

License: Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0)

Additional conditions: All derivatives must prominently cite: "Neurosymbolic Multimodal Cognitive Architecture (NMCA) - by Derek Van Derven (2026)."



IPFS Link - Please pin and share

CID: bafybeiedxfq5wsvuayjcxcxwmtto6ptelznkj6arb4mrscshwhcc2selfm

Link: https://ipfs.io/ipfs/bafybeiedxfq5wsvuayjcxcxwmtto6ptelznkj6arb4mrscshwhcc2selfm



Perma Link:

https://perma.cc/RP3N-QU4Y



My ORCID Page:

https://orcid.org/0009-0008-4149-5384

Files
NMCA-Full-128-Module-Blueprint-with-Executive-Summary-Van-Derven-2026.pdf

In 2017, Vaswani et al. published "Attention Is All You Need," a paper that reshaped artificial intelligence by eliminating recurrent networks and replacing them with pure attention mechanisms. The paper solved a practical engineering problem: parallelization. What it did not know it had solved — and what the field has largely failed to recognize since — is a deeper mathematical problem: how to build a system capable of preserving semantic identity through continuous transformation. This article argues that the transformer's success is not incidental to its architecture, but structurally inevitable. The sinusoidal positional encoding, the dot-product attention mechanism, and the multi-head structure are not design choices that happen to work. They are constraint satisfactions — the residue left after all incompatible frameworks are eliminated. Once this is understood, the current direction of mechanistic interpretability research reveals itself as a paradigm problem: a field solving the wrong question with increasingly sophisticated tools, while the foundational assumption beneath those tools remains unexamined.

Autism spectrum disorder is a complex and diverse neurobiological condition. Understanding the mechanisms and causes of the disorder requires an in-depth study and modeling of the immune, mitochondrial, and neurological systems. Computational neuroscience enhances psychiatric science by employing machine learning techniques on neural networks, combining data on brain activity with the pathophysiological and biological characteristics of psychiatric-neurobiological disorders. The research explores the integration of neurotransmitter activity into computational models and their potential roles in diagnosing and treating autism using computational methods. This research employs a narrative review that focuses on four neurotransmitter systems directly related to the manifestation of autism, specifically the following neurotransmitters: serotonin, dopamine, glutamate, and gamma-aminobutyric acid (GABA). This study reveals that computational neuroscience advances autism diagnosis and treatment by identifying genetic factors and improving the efficiency of diagnosis. Neurotransmitters play a crucial role in the function of brain cells, enhancing synaptic conduction and signal transmission. However, the interaction of chemical compounds with genetic factors and network alterations influences the pathophysiology of autism. This study integrates the investigation of computational approaches in four neurotransmitter systems associated with ASD. It improves our understanding of the disorder and provides insights that could stimulate further research, thereby contributing to the development of effective treatments.

Biological Quantum-Analogous Cognition (BQAC) is a theoretical framework proposing
thatconsciousnessisbestunderstoodasabiologicallyevolvedprocessformanagingunresolved
possibilities under conditions of uncertainty. BQAC does not claim that the brain is literally
quantum mechanical, that cognition depends on macroscopic quantum computation, or that
consciousness requires exotic physical substrates. It is not a quantum-consciousness theory.
Rather, it is a biologically grounded organisational framework that uses the quantum analogy
cautiously and structurally: as a comparison to systems that formally represent unresolved
alternatives prior to outcome selection.
The central claim of BQAC is that cognition evolved not merely to react to present stimuli,
but to generate, maintain, weight, suppress, and selectively stabilise multiple candidate
trajectories before behavioural commitment. On this view, imagination, subconscious
processing, selfhood, dreaming, planning, symbolic reasoning, and adaptive action under
uncertainty are all related expressions of a broader possibility-management architecture.
Conscious awareness emerges when one trajectory becomes sufficiently coherent, salient, and
stable to dominate behaviour, report, and flexible inferential control.
BQAC integrates work from predictive processing, active inference, hippocampal replay,
default-mode network research, action-selection neuroscience, stress physiology, symbolic
cognition, and extended cognition. It is positioned not as a replacement for these frameworks
but as a higher-level organisational account that places unresolved multiplicity and thresh-
olded stabilisation at the centre of cognitive function. A failure-mode taxonomy is offered in
which selected psychiatric conditions may be interpreted as distortions of weighting, entropy
regulation, threshold control, or collapse stability. A computational scaffold is proposed
using multi-alternative drift-diffusion and race-accumulator models, with explicit mappings
between BQAC constructs and formal parameters. BQAC is presented not as a final theory
of consciousness but as a falsifiable, non-metaphysical framework for understanding cognition
as adaptive possibility management under metabolic and temporal constraint.

SignalRupture (SR) is formalized here as a measurable structural grammar describing how systems behave when friction, scarcity, complexity, and overshoot exceed adaptive capacity. Rather than treating cognitive depletion, institutional drift, burnout, informational fragmentation, and ecological overshoot as isolated failures, SR proposes that these phenomena emerge from recurring structural operators observable across domains. Drawing from empirical literature in behavioral economics, stress physiology, cognitive load theory, institutional analysis, platform dynamics, and climate science, SR identifies recurring structural signatures including: • scarcity-induced cognitive compression (~13 IQ-point equivalent loss; d ≈ 0.80), • chronic stress dysregulation (r ≈ .40-.50), • information overload effects (r ≈-.45), • burnout correlations (r ≈ .52-.63), • institutional drift values (ODD = 0.586-1.000), • planetary overshoot indicators (36.8 Gt CO₂ emissions; 100B tonnes extraction). SR does not claim that all systems are identical. Rather, it proposes that overloaded systems exhibit recurring structural transformations characterized by narrowing, overload, depletion, drift, fragmentation, and recursive degradation. This revised formulation introduces falsifiability conditions, operational variable definitions, threshold logic, temporal dynamics, and domain typologies. SR is therefore proposed not as a universal explanatory theory, but as a testable diagnostic grammar for systemic erosion under conditions of sustained overload.

In scientific history, there are discoveries that emerge from deliberate theoretical understanding, and there are discoveries that emerge because systems themselves constrain the path forward. The latter often appear accidental to the people who first build them. Yet beneath the surface, the system was never arbitrary. The structure was already demanding a particular direction long before the researchers understood why.
This article argues that many of the central breakthroughs in modern AI — particularly the transformer architecture — belong to the second category. “Attention Is All You Need” did not succeed because its creators fully understood the mathematical nature of semantic representation. It succeeded because the architecture accidentally satisfied constraints that semantic systems require in order to remain coherent under transformation.
From this perspective, emergence in large language models was never random. Capabilities did not suddenly appear because enough parameters were accumulated. They emerged because scale acted as a dynamical force upon a latent structure that already possessed the correct geometric and topological properties. The architecture contained unrealized potential long before those potentials became behaviorally visible.
This reframes a deeper philosophical issue: discovery is often not the invention of arbitrary solutions, but the gradual recognition of constraints that were already present within the system itself. In such cases, the researcher does not freely choose the path. The structure progressively eliminates incompatible directions until only one viable trajectory remains.
The system points first. Understanding comes later.

Reduction of the morphological complexity of actual neurons into accurate, computationally efficient surrogate models is an important problem in computational neuroscience. The present work explores the use of two morphoelectrotonic transformations, somatofugal voltage attenuation (AT cables) and signal propagation delay (DL cables), as bases for construction of electrotonically equivalent cable models of neurons. In theory, the AT and DL cables should provide more accurate lumping of membrane regions that have the same transmembrane potential than the familiar equivalent cables that are based only on somatofugal electrotonic distance (LM cables). In practice, AT and DL cables indeed provided more accurate simulations of the somatic transient responses produced by fully branched neuron models than LM cables. This was the case in the presence of a somatic shunt as well as when membrane resistivity was uniform.

We present the Bayesian Information-Theoretical (BIT) model of lexical processing: A mathematical model illustrating a novel approach to the modelling of language processes. The model shows how a neurophysiological theory of lexical processing relying on Hebbian association and neural assemblies can directly account for a variety of eects previously observed in behavioral experiments. We develop two information-theoretical measures of the distribution of usages of a word or morpheme. These measures are calculated through unsupervised means from corpora. We show that our measures succesfully predict responses in three visual lexical decision datasets investigating the processing of inectional morphology in Serbian and English languages, and the eects of polysemy and homonymy in English. We discuss how our model provides a neurophysiological grounding for the facilitatory and inhibitory eects of dierent types of lexical neighborhoods. In addition, our results show how, under a model based on neural assemblies, distributed patterns of activation naturally result in the arisal of discrete symbol-like structures. Therefore, the BIT model oers a point of reconciliation in the debate between distributed connectionist and discrete localist models. Finally, we argue that the modelling framework exemplied by the BIT model, is a powerful tool for integrating the dierent levels of the description of the human language processing system.

Trauma is commonly modeled as dysregulated memory persistence, maladaptive prediction, or impaired emotional regulation. Recent work in predictive processing, active inference, and metastability has increasingly reframed trauma as a reduction in adaptive flexibility across neural and behavioral systems. However, current models remain limited in explaining why continuity landscapes become progressively narrowed, why some systems become trapped within rigid defensive attractors, and how adaptive flexibility can be restored without destabilizing continuity itself. This paper introduces Persistence Dynamics (PD), a dynamical formalism within the broader framework of Persistence Geometry, for understanding trauma, rigidity, and adaptive recovery. Within this framework, trauma is reframed not as a static injury or stored pathological imprint, but as collapse of viable accessibility across reconstructive continuity landscapes. Predictive rigidity, defensive stabilization, hypervigilance, and metastability reduction are interpreted as forms of accessibility narrowing within constrained persistence geometries shaped by energetic limitation, historical residue accumulation, synchronization burden, and viability-gated variability regulation. The framework formalizes several core constructs, including reconstructive viability, continuity basins, residue-conditioned accessibility deformation, and regulated exploratory variability. Recovery is correspondingly reframed not as unrestricted catharsis or simple memory discharge, but as progressive restoration of adaptive accessibility under stabilized reconstructive conditions. Flow states, exploratory flexibility, and metastable interventions are interpreted as transient or sustained widening of viable continuity trajectories capable of restoring adaptive flexibility without inducing coherence collapse. Persistence Dynamics is proposed as a unifying continuity framework connecting predictive coding, active inference, metastability research, trauma rigidity, and adaptive recovery within a shared continuity-conservation architecture. The model generates experimentally tractable predictions regarding trajectory diversity, metastable flexibility, reconstruction error, accessibility widening, and recovery dynamics across neural, psychological, and behavioral domains.

This paper presents the architecture of the SVR-1 system (System of Virtual Resurrection, version 1.0), designed for the integrated recovery of digital data and biological objects. We describe the key modules of the system, their interaction algorithms, biosynthesis mechanisms, and security protocols. The first successful experiments in resurrecting extinct species are presented. Ethical principles for using the technology are proposed.

The taxonomies that currently structure discussion of artificial intelligence capability-from narrow AI through artificial general intelligence (AGI) to artificial superintelligence (ASI)-are overwhelmingly behavioral, defining levels by what systems can be observed to do. This approach faces well-documented coherence problems at the AGI/ASI boundary, where behavioral signatures underdetermine the type of system under examination, definitions drift with each new model release, and disagreements about whether a transition has occurred resist empirical resolution. This paper proposes a non-behavioral alternative: an admissibility-theoretic taxonomy in which AI capability levels are characterized not by observed outputs but by the structure of admissible trajectories through capability space, with structural inspiration drawn from final-state-constrained formalisms in physics and information theory. The framework defines narrow AI as the regime in which admissible histories are bounded by externally specified terminal conditions; AGI as the regime in which terminal conditions become endogenously selected; and ASI as the regime in which the system modifies its own admissibility criteria. The paper argues this framework gives a formal handle on whether the transitions between levels are continuous or phase-transitional, clarifies how behavioral benchmarks can mistake narrow capability scaling for general capability emergence, and supplies vocabulary for ASI-specific concerns-including instrumental convergence and mesaoptimization-that current behavioral taxonomies struggle to articulate. Testable predictions, limitations, and open problems are discussed. The proposal is structural rather than literal: it borrows the formalism of final-state constraints from quantum foundations without claiming any quantum mechanism in artificial intelligence systems.