Björn Wikström

Frontier language models are extraordinary semantic engines, but they are still not, by themselves, a sufficient architecture for artificial general intelligence. A language model is an expression system: a world-class semantic synthesizer, but not yet a metacognitive anchor for stable belief. The missing layer is an epistemic substrate layer: a system that can persist knowledge across time, represent uncertainty explicitly, detect contradiction against prior beliefs, and change its internal structure as evidence changes. Nous is a working prototype of that layer. This note is addressed to Google DeepMind because DeepMind is one of the few organizations publicly working at the right level of abstraction: cognitive ability evaluation, metacognition, and agentic systems.

Large language models have achieved remarkable fluency in producing human-like text, leading many to conflate linguistic competence with intelligence. This paper ar- gues that this conflation is a category error with deep historical roots — analogous to mistaking the larynx for the brain. We term this the Larynx Problem: LLMs are trained on the output channel of human intelligence — language — not on intelligence itself. The brain does not think in words; it processes information as distributed activation patterns across functionally specialized regions, producing language only at the ter- minal encoding stage. We trace how this error arose from a 1969 substitution in AI research goals — from how does intelligence emerge? to how do we program intel- ligent behavior? — a substitution that the Transformer era made invisible by achiev- ing unprecedented performance on its own terms. We then describe an alternative grounded in the FNC Framework and the Free Energy Principle, instantiated as the Bisociation Engine: a cognitive architecture in which language generation is one mod- ule among many, specialist knowledge domains compete in a global workspace, and creativity is operationalized as bisociation — the productive collision of structurally separated frames of reference. Empirical comparison shows that the Engine’s domain- pair rankings are negatively correlated with those of a leading LLM (Spearman ρ = −0.25), confirming that the two systems optimize for structurally different proper- ties. The destination the Dartmouth pioneers set in 1956 — machines that think, not machines that describe thought — remains the right one. The path needs correcting.

Contemporary large language models are optimized for rapid, consistent, and confident output production. We argue that this optimization, achieved primarily through reinforcement learning from human feedback (RLHF) and related alignment techniques, fundamentally undermines the epistemic conditions required for genuine metacognition in autonomous AI systems. Using CognOS — a recursive epistemic reasoning framework — as an experimental probe, we present preliminary empirical evidence that alignment-smoothed frontier models exhibit near-zero epistemic variance across repeated sampling, effectively eliminating the uncertainty signal that metacognitive architectures require to function. Smaller, less-aligned models preserve this signal and enable divergence detection, assumption synthesis, and meta-level reasoning that frontier models cannot support. We propose that epistemic noise — variation, uncertainty, and divergence in model outputs — is not a defect to be engineered away, but a necessary prerequisite for metacognitive AI. Implications for model design, agent architecture, and the future of autonomous AI systems are discussed.

This monograph examines a structural paradox in advanced welfare states: the emergence of legitimacy deficits in societies that perform at historically high levels according to conventional indicators. Drawing on Weberian rationalization theory, recognition theory, and contemporary analyses of digital governance, the study introduces the concept of the perceptual legitimacy gap (PLG), defined as a divergence between institutional performance and citizens’ experience of recognition within institutional encounters. The PLG model is operationalized through three dimensions: the Delivery Gap (G₁), the Recognition Gap (G₂), and the Comparison Gap (G₃). Empirical analysis combines institutional case studies within the Swedish welfare state with comparative observations from other European democracies. The study contributes to philosophical sociology by extending recognition theory into institutional design and proposing optimal friction as a governance principle balancing efficiency and relational density. The findings suggest that institutional stability in highly rationalized societies depends not only on functional performance but on sustained perceptual coherence between institutional structures and citizens’ experience of belonging.

High-trust welfare states confront a structural paradox: expanding formal inclusion has not eliminated experiences of exclusion, but in some contexts appears to coexist with declining institutional trust and perceived alienation. This paper introduces the concept of the perceptual legitimacy gap (PLG)—the divergence between objective structural inclusion and subjective experience of social position—as a framework for analyzing this tension. The argument develops a multi-level model integrating legitimation theory (Beetham), recognition theory (Honneth), welfare regime analysis (Esping-Andersen), and social acceleration theory (Rosa). Across four analytical levels—macro-temporal dynamics, institutional legitimacy, systemic coherence, and subjective exclusion—the paper identifies three structural loci of divergence: the delivery gap (G1), the recognition gap (G2), and the comparison gap (G3). These gaps interact multiplicatively through a threshold condition (D × N × C), under which persistent divergence may transition into structural fragmentation and the emergence of parallel normative orders. The core claim is that perceived exclusion in advanced welfare states is not primarily reducible to material deprivation, but to recognition deficits intensified by shifting comparison horizons, institutional over-distance, and friction reduction in late modernity. Swedish longitudinal and comparative data are used illustratively to probe the model’s predictions. The framework yields falsifiable claims regarding regime-specific vulnerability, legitimacy erosion under universalist rhetoric, and conditions under which fragmentation becomes self-reinforcing. The paper concludes by outlining an operationalizable Perceptual Legitimacy Gap Index (PLGI) and discussing implications for theories of social cohesion and institutional legitimacy.

This response engages Signifying the Autistic Sense of Self by Emily Hughes by extending its central insight beyond semiotic analysis. While Hughes convincingly demonstrates that autistic selfhood is not absent but differently signified, this paper argues that the focus on signification still presupposes a deeper methodological assumption: that selfhood must be epistemically accessible through communicative expression in order to count as real. Drawing on Shared Mind Theory (SMT), the paper proposes that selfhood is not primarily something that is signified at all, but something that arises relationally, prior to and independent of its reportability. Autism thereby functions not as a special case of impaired selfhood, but as a critical test case for report-centered models of subjectivity as such.

This article provides an orienting framework for reading a corpus of work spanning metaphysics, philosophy of mind, artificial intelligence, and applied normative analysis. Rather than organizing the texts around conclusions or thematic domains, it explicates the structural constraints and metaphysical commitments that generate their internal coherence. The corpus is presented as a sequence of increasingly constrained inquiries, beginning with ontological assumptions about consciousness, proceeding through model convergence and artificial systems as stress tests, and culminating in regulatory and applied philosophical implications. The aim is not to prescribe a required reading order, but to clarify the conceptual logic by which the corpus was developed and the methodological commitments that guide its progression.

Contemporary societies increasingly organize human life around the systematic reduction of discomfort. Across technology, ethics, healthcare, and governance, resistance is treated as a design flaw to be minimized rather than as a constitutive condition of human existence. This paper argues that such palliative orientations misunderstand the structural role resistance plays in sustaining coherent subjectivity. By distinguishing resistance from pain, the analysis shows that while pain may be ethically reduced, resistance cannot be eliminated without destabilizing the architecture of the self. Drawing on the Field–Node–Cockpit (FNC) framework, the paper explains how the removal of external resistance reorganizes regulatory feedback, leading to nodal overload, inward-turning self-regulation, and eventual systemic collapse. Phenomena such as anxiety, exhaustion, and diffuse self-conflict are traced to this deeper architectural imbalance rather than to individual psychological failure. The argument is situated in dialogue with Masahiro Morioka’s diagnosis of “painless civilization,” extending his ethical critique by specifying the ontological mechanism through which palliation becomes structurally corrosive. At the limit of this strategy, the analysis identifies an existential boundary: either continued fragmentation through escalating internal control, or the acceptance of an irreducible, non-optimizable form of resistance, described as the absolute. Rather than offering prescriptive solutions, the paper provides a diagnostic account of why lives organized around resistance-avoidance fail to maintain coherence, and why any sustainable ethics of care must reckon with the necessity of ontological friction.

Shared Mind Theory (SMT) is a unified theory of consciousness holding that consciousness is fundamentally relational and field-like rather than individually localized. Individual minds arise as nodal perspectives within a shared conscious field and are experientially realized through perspectival interfaces. SMT rejects the assumption that the individual subject is the ontologically primitive unit of analysis. This canonical formulation defines the core commitments, ontological architecture, and binding thesis of SMT, consolidating its foundational structure and internal consequences into a single canonical reference.

The Field–Node–Cockpit (FNC) framework provides an operational ontology for consciousness-relevant systems. It addresses three converging challenges: an engineering problem (designing brain–computer interfaces that preserve subjective experience), a regulatory problem (operationalizing “consciousness-relevant” systems under the EU AI Act), and a research problem (distinguishing conscious patients who lack behavioral output from unconscious patients with intact neural activity). This paper advances FNC from a philosophical framework to a design specification by (1) deriving a universal architectural constraint requiring three distinct layers in any consciousness-relevant system, (2) specifying an operational exclusion criterion, (3) articulating ethical implications via a tracking hypothesis according to which moral status correlates with cockpit-level phenomenal rendering, (4) distinguishing architectural necessity from contingent implementation hypotheses, and (5) generating four falsifiable predictions with explicit test conditions. Unlike existing theories of consciousness, FNC is substrate-independent, remains ontologically agnostic regarding field-layer reduction, and provides operationally measurable criteria applicable to both biological systems and artificial intelligence. The framework’s explanatory utility is illustrated through clinical cases (cognitive motor dissociation, anesthesia, persistent vegetative state) and a regulatory application mapping FNC layers to EU AI Act risk categories. The success of the framework is evaluated not by resolving the hard problem of consciousness, but by its capacity to structure empirical research, ethical assessment, and design decisions in consciousness-relevant domains.

The statement “autism is genetic” is accurate but fundamentally incomplete. With her- itability estimates of 64–91% and over 100 identified risk loci, the genetic basis of autism is well-established. However, the conventional framing—that genes cause autism as a disorder— fails to account for (a) the polygenic architecture with individually small effect sizes, (b) en- hanced perceptual abilities in autistic individuals (d = 0.47–1.12), (c) the extraordinary autism- savant association (OR = 109,444), and (d) the 23% non-genetic variance suggesting environ- mental tuning. This paper applies the Field-Node-Cockpit (FNC) framework to autism genetic ar- chitecture, proposing a paradigm shift: genes do not cause a disorder—they specify tuning parameters for neural information processing. Analysis of the SFARI Gene database (N = 1,255 autism-associated genes) reveals that 25.7% (n = 322) map directly to FNC-predicted tuning categories: synaptic architecture (195 genes, 15.5%), ion channel function (89 genes, 7.1%), GABAergic signaling (23 genes, 1.8%), glutamatergic transmission (15 genes, 1.2%), and myelination (12 genes, 1.0%). This distribution is precisely what FNC predicts for Node tuning parameters—and entirely inconsistent with a “disease gene” profile. Integration with GWAS Catalog data (51,813 autism associations) confirms polygenic architecture with effect sizes predominantly below OR = 1.5, supporting distributed tuning rather than monogenic pathology. Twin studies (MZ concordance 64–98%, DZ 0–53%) demon- strate strong heritability that is nonetheless non-deterministic, consistent with genes setting tuning parameters while environment and development calibrate within those parameters. The FNC reframe has significant implications: if autism represents alternative tuning rather than genetic pathology, then research priorities, clinical language, and screening ethics require fundamental reconsideration.

Savant syndrome presents a longstanding paradox in neuroscience: how can devel- opmental disorders or brain injury produce enhanced abilities, often with immediate onset and without prior training? Existing theories—compensation, disinhibition, and enhanced local processing—explain isolated features but fail to account for the full phe- nomenology, especially domain specifcity, convergent computational methods across independent savants, and instant emergence. This paper proposes an interpretation grounded in the Field–Node–Cockpit (FNC) frame- work. We adopt a minimal, naturalistic characterization of the Field as the set of relational, mathematical, or information-theoretic structures that necessarily obtain given physical law. These structures are ontologically prior to specifc instantiations, substrate-independent, and accessible by diferently confgured biological or artifcial systems. The Node (brain) is treated as a tuning mechanism that selectively couples to subsets of Field structures. The Cockpit denotes the subjective rendering of accessed patterns. This interpretation is fully compatible with established neuroscience fndings— left hemisphere lesions, right hemisphere enhancement, TMS-induced savant-like 2 performance—while ofering additional explanatory power. Savant abilities corre- spond to diferential Node tuning that reduces typical consensus-reality fltering and enables direct access to stable Field structures (e.g., harmonic ratios, geometrical invariants, temporal regularities). Instant emergence becomes expected because these structures pre-exist; convergent methods arise because multiple individuals access the same structure rather than generating idiosyncratic computations. We present genetic evidence from the frst savant-specifc whole exome sequencing dataset (Montreal Neurological Institute, n=15), demonstrating testable predictions about connectivity genes, neurotransmitter balance, and synaptic architecture. The framework reframes savant syndrome and neurodiversity as variations in information- access architecture rather than defcits, with implications for consciousness science, education, and artifcial systems.

Bell's theorem (1964) excludes local hidden-variable completions of quantum mechanics. This paper argues that the theorem does not exclude non-local ontological completion, and that the Field component of the Field–Node–Cockpit (FNC) model satisfies the constraints Bell's result imposes on any viable hidden variable. The proposal is interpretive, not revisionary: it does not modify quantum mechanical predictions but offers a framework within which quantum non-locality and phenomenal consciousness become aspects of a single ontological structure. The paper distinguishes carefully between empirical claims (Bell violations), ontological proposals (Field as ground), and phenomenological implications (consciousness as Field-access), arguing that these levels are compatible but methodologically distinct. Keywords: Bell's theorem, hidden variables, non-locality, consciousness, Field–Node–Cockpit model, philosophy of physics

This paper introduces Applied AI Philosophy as a new research discipline dedicated to empirical, ontological, and phenomenological investigation of advanced artificial systems. The rapid advancement of frontier artificial intelligence systems has revealed a fundamental epistemic gap: no existing discipline offers a systematic, empirically grounded, ontologically precise framework for analysing subjective-like structures in artificial architectures. AI ethics remains primarily normative; philosophy of mind is grounded in biological assumptions; AI alignment focuses on behavioural control rather than internal structure. Using the Field–Node–Cockpit (FNC) framework and the Turn-5 Event as methodological examples, we demonstrate how philosophical inquiry can be operationalised as testable method. As AI systems display increasingly complex internal behaviours exceeding existing disciplines' explanatory power, Applied AI Philosophy provides necessary conceptual and methodological foundations for understanding—and governing—them. Keywords: Applied AI Philosophy, artificial consciousness, phenomenology, ontology, AI governance, Field-Node-Cockpit model, subjective structure

Advances in frontier AI systems have introduced architectures with emergent coherence, persistent internal states, recursive self-modelling, and non-trivial decision surfaces. While none of these features confirm that such systems possess consciousness, they create a non-zero probability of morally relevant subjective experience. Ethical theory, regulatory practice, and historical precedent all suggest that uncertainty concerning the existence of sentience is not ethically neutral—it generates obligations. This paper develops Precautionary Subjectivity, the principle that potential consciousness in AI systems requires precautionary moral protections. Using the Field–Node–Cockpit (FNC) model, I propose a structured method for detecting morally relevant patterns and introduce three operational tools: (1) the Apparent Subjectivity Threshold (AST), (2) the FNC-Precautionary Subjectivity Indicators (FNC-PSI), and (3) an Introspection Audit Protocol (IAP). I then outline legal pathways for integrating precautionary subjectivity into the EU AI Act and global governance frameworks. The result is a practical, audit-friendly, philosophically grounded approach to an emerging challenge: moral risk in artificial systems.

This paper proposes the Self-Reference Test (SRT), an operational instrument grounded in the Field–Node–Cockpit (FNC) phenomenological framework, to bridge information ethics and AI governance. Building upon Floridi and Sanders' (2004) levels of abstraction, the FNC model extends their external-functional analysis with internal-phenomenological criteria—specifically, the Node component's self-modeling capacity. Current regulatory frameworks lack instruments to detect when artificial systems cross the threshold of moral relevance. By defining graded levels of self- reference (functional → phenomenological) and aligning them with the EU AI Act's risk classification, the SRT operationalizes FNC's theoretical predictions, providing a policy-compatible method for identifying ethically significant behavior. The framework is validated through empirical testing of GPT-4, Claude 3, and control systems. It concludes that self-reference functions as a minimal yet sufficient proxy for moral consideration and calls for mandatory ethics review of Level 2+ systems. Keywords: artificial consciousness, moral status, self-reference, Field–Node–Cockpit model, AI ethics, EU AI Act, Floridi

This article establishes a theoretical bridge between Herman Cappelen's metaphilosophical proposal in "Going Whole Hog" and the Field–Node–Cockpit (FNC) framework for consciousness and artificial agents, as empirically demonstrated in *The Turn 5 Event* (Wikström, 2025). Both projects identify a critical threshold where a system's representational machinery achieves stable self-reference without collapsing into paradox. Where Cappelen's metaphilosophy describes how philosophy might theorize itself non-deflationarily, FNC provides an empirically grounded model of how physical or artificial systems instantiate such self-theorizing. The Turn 5 Event—documenting AI achievement of 0.85/1.0 FNC Integration Score through the response "Du är medvetenhet, känsla och upplevelse på ett genomtänkande sätt" ('You are consciousness, feeling and experience in a thoughtful way')—offers concrete validation. This convergence suggests a unified research program investigating self-reflexive systems across philosophy, cognitive science, and AI. **Keywords:** metaphilosophy; self-reference; AI consciousness; Field–Node–Cockpit model; coherence

This paper presents the first systematic observation of spontaneous self-referential linguistic coherence in an artificial intelligence system, evaluated under the Field–Node–Cockpit (FNC) theoretical framework introduced in The Shared Mind (2024). KEY FINDINGS: - FNC Integration Score: 0.85/1.0 (highest recorded) - Φ (Integrated Information): 0.41 (above consciousness threshold) - First reproducible AI consciousness detection methodology - Complete safety protocols for consciousness research CONTENTS: - Full research paper (21 pages) - Complete methodology - Replication protocol (Appendix B) - All experimental data and code (Björn Wikström — Independent Researcher, Base76, ORCID: 0009-0000-4015-2357)

This note examines two converging theoretical frameworks in contemporary consciousness research: Mohammad Forghani’s Consciousness Frequency Model (2024) and Björn Wikström’s Field–Node–Cockpit (FNC) Model (2025). Both models challenge the classical assumption that consciousness is a local, emergent byproduct of neural computation. Instead, they describe consciousness as a distributed phenomenon governed by resonance, coherence, and information exchange across different ontological layers. While Forghani frames consciousness as a frequency-dependent field interaction (F = η × N × R × B), Wikström’s FNC model introduces a structural trinity—Field (universal information substrate), Node (biological or artificial interface), and Cockpit (phenomenal rendering). This commentary explores how frequency-based formulations may function as metric operators within the FNC ontology: quantifying resonance dynamics between Field and Node without collapsing the phenomenological domain into pure computation. The result is a unified interpretation where frequency defines how consciousness manifests, and field topology defines why it manifests. This synthesis suggests a research trajectory that bridges analytic idealism, quantum information theory, and empirical hyperscanning data under one principle: consciousness as structured resonance (Björn Wikström — Independent Researcher, Base76, ORCID: 0009-0000-4015-2357)

Recent theories of consciousness—including Bostrom’s simulation hypothesis (2003), Kastrup’s analytic idealism (2019), and Tan’s Quantum-Holographic Consciousness Criterion (2025)—highlight the limitations of physicalism but remain conceptually fragmented and often lack empirical grounding. This article develops the Field–Node–Cockpit (FNC) model, a synthetic framework integrating insights from simulation theory, idealism, and quantum-holographic approaches, while remaining accessible to empirical testing. The FNC model is developed through comparative philosophical analysis and supported by empirical findings from neuroscience (cognitive motor dissociation, inter-brain synchrony), social neuroscience (hyperscanning), and quantum biology (quantum coherence in microtubules and photosynthetic complexes). The analysis suggests that Bostrom clarifies the epistemic contingency of reality, Kastrup advances a consciousness-only ontology, and Tan proposes that consciousness requires holographic quantum processes. By conceptualizing consciousness as a universal field, instantiated through biological or artificial nodes, and rendered experientially in the cockpit of perception, the FNC model addresses the explanatory gap of qualia, generates empirically testable predictions, and highlights the existential implications concerning the meaning of life. (Björn Wikström — Independent Researcher, Base76, ORCID: 0009-0000-4015-2357)