Convergence Encyclopedia: The Schools — Mind, Machine & Meaning

PART 4: THE SCHOOLS — MIND, MACHINE & MEANING

4.1 Psychology & Cognitive Science

William James — Stream of Consciousness & Pragmatism (1890/1907)

James described consciousness as a continuous “stream” rather than discrete states, and argued that truth is what works in experience. His pragmatism dissolves metaphysical disputes by asking what practical difference a belief makes.

Aspect

Mapping

Patterns instanced

C06 (information as the medium of mental life — the stream is information flow); C07 (feedback — consciousness as self-correcting process); C08 (recursion — consciousness as consciousness of consciousness)

Tier

T2 (foundational insight, pre-formal)

Falsifier

Disproof of continuous processing in favor of discrete computational steps; demonstration that pragmatism cannot distinguish true from merely useful beliefs

Rival frame

Brentano’s act psychology (intentionality as irreducible); eliminative materialism (consciousness as folk-psychological illusion)

Behaviorism — Watson, Skinner (1913–1957)

Behaviorism held that psychology must study observable behavior, not internal mental states. Learning is stimulus-response conditioning; the organism is a black box shaped by reinforcement schedules.

Aspect

Mapping

Patterns instanced

C07 (feedback — operant conditioning is feedback: behavior → consequence → behavior modification); C09 (selection/variation-retention — Skinner explicitly modeled operant behavior on natural selection: random response variations are selected by environmental consequences); C23 (attractors — fixed reinforcement schedules produce stable behavioral equilibria)

Tier

T1 (historically dominant, now superseded as complete account but preserved as special case)

Falsifier

Demonstration that all learning is stimulus-response (impossible; cognitive maps, latent learning, and insight learning refute); proof that internal representations play no causal role

Rival frame

Cognitivism — internal representations are necessary and causally efficacious; behaviorism describes a subset (reflexive/automatic behavior) not the whole

Gestalt Psychology — Wertheimer, Koffka, Köhler (1912–1940s)

The Gestalt school demonstrated that perception is organized by field laws: proximity, similarity, continuity, closure, figure-ground. The whole is perceptually primary; parts are perceived in relation to the whole.

Aspect

Mapping

Patterns instanced

C21 (emergence — perceptual wholes have properties not present in sensory elements); C05 (criticality — Gestalt “good form” is a minimum-energy configuration of the perceptual field); C03 (symmetry — figure-ground organization is a symmetry-breaking of the visual field); C02 (least action — perceptual organization proceeds toward Pragnanz, the “best form” minimum)

Tier

T2 (empirically robust, mechanism partially specified via Bayesian brain)

Falsifier

Demonstration that perception is purely elemental/associative with no emergent organizational principles; proof that computational models cannot reproduce Gestalt phenomena

Rival frame

Atomistic associationism (all perception built from sensory primitives); computational vision (Gestalt laws as heuristics, not field dynamics)

Jean Piaget — Developmental Stages & Constructivism (1923–1983)

Piaget mapped cognitive development through invariant sequential stages (sensorimotor → preoperational → concrete operational → formal operational), each characterized by distinct logico-mathematical structures. Knowledge is constructed through equilibration — the balance of assimilation and accommodation.

Aspect

Mapping

Patterns instanced

C05 (criticality — stage transitions are qualitative leaps when cognitive structures become unstable); C07 (feedback — equilibration as homeostatic self-regulation); C08 (recursion — operations-on-operations at formal stage); C09 (selection — inadequate schemas are selected against via cognitive conflict); C23 (attractors — each stage is a stable cognitive basin)

Tier

T3 (broad descriptive framework challenged on domain-specificity and cultural variation; stage invariance questioned)

Falsifier

Evidence that developmental sequences are entirely culture-dependent with no invariant order; demonstration that domain-general stages do not exist (core knowledge, modular competence)

Rival frame

Core knowledge theory (Spelke, Carey — infants have domain-specific competencies, not just sensorimotor schemas); Vygotskian social constructivism (development is socially mediated, not individually constructed)

J.J. Gibson — Affordances & Direct Perception (1966–1979)

Gibson argued that perception is direct, not mediated by internal representations. The environment specifies its own properties: surfaces “afford” standing-on, objects afford grasping. Information is in the ambient optic array, not constructed by the perceiver.

Aspect

Mapping

Patterns instanced

C06 (information — the optic array as structured energy specifying environmental layout); C02 (least action — direct perception skips computational inference, taking the geodesic from stimulus to percept); C12 (autopoiesis — the perceiver-environment system is a coupled whole, perception is action); C21 (emergence — affordances are properties of animal-environment system, not either alone)

Tier

T2 (empirically influential, especially in robotics and embodied cognition; mechanism debated)

Falsifier

Proof that all perception requires internal representational mediation; demonstration that organisms can perceive nothing without memory/expectation playing a causal role

Rival frame

Constructivist perception (von Helmholtz — perception as unconscious inference); predictive processing (perception as Bayesian inference, not direct pickup)

Cognitive Science — Minsky, McCarthy, Newell & Simon (1956–1990s)

The classical cognitive science paradigm: mind as information-processing system, cognition as symbol manipulation, thinking as heuristic search through problem spaces. The Physical Symbol System Hypothesis (Newell & Simon, 1976) states that a physical symbol system has the necessary and sufficient means for general intelligent action.

Aspect

Mapping

Patterns instanced

C06 (information — cognition as computation over representations); C20 (universal computation — mind as a Turing-complete symbol engine); C08 (recursion — recursive decomposition of problems into subproblems); C15 (optimization — heuristic search as bounded optimization over problem spaces); C09 (selection — generate-and-test as variation/selection)

Tier

T2 (historically foundational; demonstrated limits via connectionism and embodied cognition; survives as component)

Falsifier

Demonstration that symbol manipulation is insufficient for intelligent behavior (affective, embodied, subsymbolic processes are necessary); proof that subsymbolic processes can be reduced to symbolic ones

Rival frame

Connectionism (subsymbolic processing is primary); embodied cognition (off-loading computation to body/world); eliminative materialism (no symbols, no representations)

Connectionism — Rumelhart, McClelland, Hopfield (1982–present)

Connectionism models cognition as parallel distributed processing: computation emerges from the weighted interactions of simple units. Learning occurs through weight adjustment (error-correction, Hebbian rules). Knowledge is stored in connection weights, not explicit symbols.

Aspect

Mapping

Patterns instanced

C11 (networks — computation as network dynamics); C09 (selection — learning as weight selection by error signal); C10 (scale invariance — similar architectures across scales of complexity); C21 (emergence — global computation from local interactions); C06 (information — information distributed across weights, not localized); C23 (attractors — memory as attractor states of the network dynamics)

Tier

T1 (deep learning, a direct descendant, now dominates AI and increasingly cognitive modeling)

Falsifier

Proof that distributed representations cannot account for systematicity, compositionality, or abstract reasoning (Fodor & Pylyshyn challenge — partially met by transformers); demonstration that localist/symbolic representations are always required

Rival frame

Classical cognitivism (symbolic structures necessary for systematic thought); eliminative connectionism (no symbols anywhere — overstated by some advocates)

Embodied Cognition — Varela, Thompson, Rosch (1991)

The embodied mind thesis: cognition is not computation in a vacuum but arises from bodily interaction with the world. Varela, Thompson & Rosch’s The Embodied Mind synthesized phenomenology, cognitive science, and Buddhist philosophy. Enaction: cognition as the bringing-forth of a world through structural coupling.

Aspect

Mapping

Patterns instanced

C12 (autopoiesis — the living system produces its own boundary and maintains itself); C07 (feedback — structural coupling as continuous reciprocal causation); C02 (least action — cognition emerges from bodily engagement, not abstract inference); C21 (emergence — mind as emergent from brain-body-world system); C08 (recursion — self as self-producing process)

Tier

T2 (philosophically rich, empirically developing; robotics and predictive processing provide implementations)

Falsifier

Proof that identical cognitive processes occur without any bodily grounding (pure disembodied AI); demonstration that phenomenology plays no causal role in cognitive science

Rival frame

Computational functionalism (body is implementational detail); extended mind (body + environment are literally parts of cognitive system — stronger claim)

Predictive Processing / Free Energy Principle — Friston, Clark, Seth (2003–present)

The brain as inference engine: perception is hypothesis-testing, action is hypothesis-confirmation. The free energy principle (Friston) states that biological systems minimize variational free energy — an information-theoretic bound on surprise. Predictive processing (Clark) frames cognition as hierarchical predictive coding: top-down predictions meet bottom-up prediction errors.

Aspect

Mapping

Patterns instanced

C13 (free energy/active inference — literal instantiation, this is the theory); C06 (information — free energy as information-theoretic quantity); C07 (feedback — prediction-error as feedback signal driving learning); C02 (least action — minimizing free energy as least action in information geometry); C08 (recursion — hierarchical models predict their own predictions); C09 (selection — models that minimize free energy are selected); C21 (emergence — perception/action as emergent from inference dynamics)

Tier

T1 (empirically supported across multiple domains; mathematical framework well-developed; scope debated)

Falsifier

Demonstration that neural computation cannot be described as Bayesian inference; proof that free energy minimization makes no testable predictions beyond existing frameworks; evidence that prediction-error signals do not drive perception

Rival frame

Direct perception (Gibson — no inference needed); reinforcement learning (no explicit generative model); feedforward processing is sufficient for object recognition (some evidence from ultra-rapid categorization)

Global Workspace Theory — Baars, Dehaene (1988–present)

Consciousness arises when information is broadcast globally across the brain from a central workspace. Unconscious processes are modular and parallel; conscious processes are integrated and accessible. The “neural global workspace” is a network of long-range neurons enabling broadcasting.

Aspect

Mapping

Patterns instanced

C06 (information — consciousness as integration/access of information); C21 (emergence — conscious experience emerges from global broadcasting); C11 (networks — workspace as specific network topology); C07 (feedback — ignition of workspace requires recurrent activation); C05 (criticality — workspace ignition is a phase transition from local to global activation)

Tier

T1 (empirically supported by Dehaene’s experiments: subliminal vs. conscious processing show distinct neural signatures; integrated with IIT debates)

Falsifier

Evidence that consciousness occurs without global broadcasting (isolated conscious content); proof that broadcasting occurs without consciousness (zombie systems); demonstration that workspace architecture is unnecessary for consciousness

Rival frame

Integrated Information Theory (consciousness as integrated information, not broadcasting); higher-order theories (consciousness as meta-representation); recurrent processing theory (consciousness as sufficiently long recurrent loops)

Integrated Information Theory — Tononi (2004–present)

IIT proposes that consciousness is identical to integrated information (phi, φ): the amount of information generated by a system as a whole beyond what its parts generate independently. Consciousness is intrinsic; its quantity and quality are determined by the system’s causal structure.

Aspect

Mapping

Patterns instanced

C06 (information — consciousness is information integration); C21 (emergence — phi is an emergent property of causal structure); C11 (networks — phi is computable from network connectivity); C14 (duality — subjective experience and objective information as dual aspects of the same structure)

Tier

T2 (mathematically precise; predictive power debated; panpsychism implications controversial; some predictions contradicted — e.g., cerebellum has high phi but does not seem conscious)

Falsifier

Demonstration that systems with high phi lack consciousness; demonstration that systems with low phi have rich consciousness; mathematical proof that phi cannot be computed or has no empirical content

Rival frame

Global workspace theory (consciousness as access, not intrinsic information); panpsychism (consciousness is everywhere, not measured by information integration); functionalism (consciousness as what it does, not what it is)

4.2 AI & Machine Learning

Perceptron & Early Neural Nets — Rosenblatt (1958); Minsky & Papert Critique (1969)

Rosenblatt’s perceptron learned linear decision boundaries through iterative weight updates. Minsky & Papert’s Perceptrons proved that single-layer perceptrons cannot compute nonlinearly separable functions (XOR). This halted neural net research for a decade.

Aspect

Mapping

Patterns instanced

C09 (selection — weight updates as selection of useful features); C11 (networks — the perceptron as simplest trainable network); C20 (universal computation — negative result: perceptron is NOT universal); C07 (feedback — error signal drives learning)

Technical specifics

The perceptron convergence theorem (Rosenblatt): if a solution exists, the algorithm finds it in finite steps. Minsky & Papert’s group invariance theorem: what a perceptron cannot learn locally, it cannot learn globally.

Tier

T1 (foundational; the Minsky-Papert critique was mathematically correct but sociologically over-interpreted — multilayer nets escape the limit)

Falsifier

Proof that multilayer networks also cannot learn nonlinear functions (disproven by backpropagation); proof that perceptrons can learn XOR (false — requires nonlinearity)

Rival frame

Symbolic AI (learning is not the path to intelligence; hand-crafted representations are necessary); kernel methods (nonlinearity via feature space, not layered architecture)

Backpropagation — Rumelhart, Hinton, Williams (1986)

Backpropagation computes gradients of a loss function with respect to network weights via the chain rule, propagating error signals backward through the network. It enabled training multilayer networks, escaping the Minsky-Papert limitation.

Aspect

Mapping

Patterns instanced

C01 (gradient dissipation — literal: backprop IS gradient descent via automatic differentiation); C09 (selection — weights selected by gradient to minimize loss); C08 (recursion — chain rule as recursive computation of derivatives through composition); C02 (least action — training finds a local minimum of the loss landscape)

Technical specifics

Backprop computes ∂L/∂w for every weight by applying the chain rule recursively: ∂L/∂w_i = ∂L/∂a * ∂a/∂w_i. The error signal propagates backward: δ_l = (W_{l+1}^T δ_{l+1}) ⊙ f’(z_l). This is gradient dissipation (C01) through a compositional function space.

Tier

T1 (enabling algorithm for deep learning; biologically implausible in exact form but approximated by brains)

Falsifier

Proof that gradient descent cannot train useful networks (empirically false); demonstration that backpropagation is biologically impossible and no approximation works; proof that all loss landscapes are pathological ( empirically: some are, but not all)

Rival frame

Hebbian learning (unsupervised, local, biologically plausible — but less powerful); genetic algorithms (gradient-free optimization — slower); target propagation (alternative credit assignment)

Support Vector Machines & Kernel Methods — Vapnik (1995)

SVMs find the maximum-margin hyperplane separating classes. The kernel trick maps data to high-dimensional feature spaces without explicit computation. SVMs are convex optimization problems with global optima.

Aspect

Mapping

Patterns instanced

C15 (optimization — SVM as quadratic optimization: maximize margin subject to constraints); C02 (least action — the maximum-margin principle selects the “simplest” boundary in feature space); C06 (information — support vectors as the minimal information needed to specify the decision boundary); C14 (duality — primal (weights) and dual (support vectors) formulations are complementary descriptions)

Technical specifics

SVM solves: min_{w,b} ½

Tier

T1 (theoretically elegant; dominated by deep learning on large perceptual datasets but survives in structured/limited-data regimes)

Falsifier

Proof that margin maximization does not generalize; demonstration that kernel methods always outperform neural nets (false empirically); proof that the kernel trick provides no advantage

Rival frame

Deep learning (end-to-end feature learning beats hand-crafted kernels); Bayesian methods (uncertainty quantification, not just point estimates); random forests (nonparametric, no kernel selection needed)

Deep Learning Revolution — Hinton, LeCun, Bengio (2006–present)

Deep neural networks with many layers learn hierarchical representations from raw data. Key innovations: ReLU activations, dropout regularization, batch normalization, architectural variants (CNNs, RNNs, LSTMs, ResNets, GANs). AlexNet (2012) marked the watershed.

Aspect

Mapping

Patterns instanced

C09 (selection — stochastic gradient descent as massive parallel selection of weights); C10 (scale invariance — similar architectures work across scales and domains); C21 (emergence — high-level features emerge from low-level through composition); C06 (information — information bottleneck: networks compress input through bottleneck layers); C11 (networks — computation as network dynamics); C01 (gradient dissipation — SGD as gradient flow on loss landscape)

Technical specifics

Deep networks implement the chain rule through many layers: f(x) = f_L ∘ f_{L-1} ∘ … ∘ f_1(x). Each layer transforms representation space. Universal approximation theorem (Cybenko, 1989): a single hidden layer can approximate any continuous function, but depth enables efficient representation of compositional functions (Poggio et al.: depth provides exponential expressivity for hierarchical functions).

Tier

T1 (empirically dominant across vision, NLP, speech, game-playing; theoretical understanding incomplete)

Falsifier

Demonstration that deep networks cannot generalize (they do, via implicit regularization); proof that shallow networks are always more efficient (false for compositional functions); evidence that deep learning hits an unbreachable scaling wall

Rival frame

Symbolic AI (no compositional generalization in pure neural nets — partially true, being addressed); neuro-symbolic hybrid (neural perception + symbolic reasoning); causal models (deep learning learns correlation, not causation — Pearl)

Transformers / Attention Mechanism — Vaswani et al. (2017)

The Transformer replaces recurrence and convolution with self-attention: each token attends to all others, computing weighted representations. “Attention is all you need” — the architecture scaled to GPT, BERT, and large language models.

Aspect

Mapping

Patterns instanced

C11 (networks — fully-connected attention graph, sparsified in practice); C06 (information — attention weights as information routing; information is dynamically allocated); C08 (recursion — self-attention as each token processing every other token; transformer depth as recursive refinement); C10 (scale invariance — same architecture from small models to GPT-4 scale); C21 (emergence — in-context learning, chain-of-thought reasoning emerge at scale)

Technical specifics

Attention(Q,K,V) = softmax(QK^T/√d_k)V. Self-attention computes pairwise token interactions in O(n²) time. Multi-head attention projects into subspaces. Positional encoding injects sequence order. The transformer is a message-passing graph neural network on a complete graph (C11). Scale laws (Kaplan et al.): loss ∝ N^(-α) where N is parameters — power law (C10).

Tier

T1 (architecturally dominant in NLP; scaling laws empirically established; mechanism of emergent abilities debated)

Falsifier

Proof that transformers cannot model long-range dependencies (addressed by sparse attention, state space models); demonstration that attention mechanism provides no advantage over RNNs at scale; proof that emergent abilities are purely measurement artifacts

Rival frame

Recurrent models (sequential processing, not parallel); state space models/Mamba (linear-time sequence modeling); neuro-symbolic (attention as soft lookup, not reasoning)

Reinforcement Learning — Sutton & Barto; AlphaGo/AlphaZero (1998–2018)

RL learns policies (mappings from states to actions) through trial-and-error interaction with an environment. Value functions estimate future reward; policy gradients optimize action probabilities directly. AlphaGo/AlphaZero combined deep nets with Monte Carlo tree search.

Aspect

Mapping

Patterns instanced

C07 (feedback — RL is feedback: action → reward → policy update); C09 (selection — policies selected by cumulative reward; exploration generates variation); C02 (least action — policy optimization finds geodesic in policy space toward reward); C13 (free energy/active inference — RL as minimizing expected free energy when reward = negative surprise); C08 (recursion — temporal difference learning bootstraps value estimates from value estimates); C15 (optimization — multi-objective RL: Pareto front of competing objectives)

Technical specifics

Bellman equation: V(s) = max_a [R(s,a) + γV(s’)]. Policy gradient: ∇J = E[∇log π(a

Tier

T1 (superhuman game-playing; sample inefficiency in real-world applications; reward specification challenges)

Falsifier

Proof that RL cannot learn from sparse rewards (empirically: it can with enough computation or curriculum); demonstration that model-free RL is always inferior to model-based planning; proof that reward hacking is unavoidable

Rival frame

Model-based planning (explicit world models more sample-efficient); imitation learning (skip exploration, learn from demonstrations); behavioral cloning (no reward signal needed)

Emergent Capabilities & Scaling Laws — Kaplan et al. (2020); Wei et al. (2022)

Large language models display capabilities not present in smaller models: in-context learning, chain-of-thought reasoning, instruction following. Scaling laws predict that loss decreases as a power law in model size, data, and compute.

Aspect

Mapping

Patterns instanced

C21 (emergence — capabilities appear discontinuously at scale); C10 (scale invariance — power-law scaling across orders of magnitude); C05 (criticality — emergent abilities appear at phase-transition-like thresholds); C06 (information — scaling as increased information capacity); C09 (selection — pretraining selects weights that capture training distribution statistics)

Technical specifics

Kaplan scaling laws: L(N) ∝ N^(-α_N), L(D) ∝ D^(-α_D), L(C) ∝ C^(-α_C) with α ≈ 0.07-0.35 depending on regime. Chinchilla scaling (Hoffmann et al., 2022): optimal compute allocation balances model size and data. Emergence is debated: Schaeffer et al. (2023) argue it’s a metric artifact (nonlinear metrics make continuous improvement appear discontinuous). This is C05 (criticality) — whether genuine phase transitions or measurement artifacts is an active frontier.

Tier

T2 (scaling laws are robust empirical regularities; interpretation of emergence contested)

Falsifier

Demonstration that scaling laws break down (no further improvement with scale); proof that emergent abilities are entirely prompting artifacts; evidence that smaller models match larger ones with better training

Rival frame

Emergence as metric artifact (Schaeffer et al. — continuous improvement looks discontinuous under nonlinear metrics); extrapolation critique (scaling laws may not hold beyond measured range); capability overhang (sudden jumps due to evaluation, not model)

Active Inference in AI — Friston Applied (2015–present)

Active inference (Friston) frames agents as minimizing expected free energy through both perception (updating beliefs) and action (selecting policies). When implemented in AI, it provides a principled framework for perception-action loops with built-in epistemic and pragmatic drives.

Aspect

Mapping

Patterns instanced

C13 (free energy/active inference — direct instantiation); C07 (feedback — perception-action as closed feedback loop); C12 (autopoiesis — agent maintains itself through active engagement); C02 (least action — free energy minimization as variational principle); C09 (selection — policies selected by expected free energy)

Technical specifics

Expected free energy G(π) = ΣQ(o

Tier

T2 (mathematically principled; computational cost high; implementations growing in robotics and agency research)

Falsifier

Demonstration that free energy minimization is computationally intractable for all but toy problems; proof that active inference makes no predictions beyond standard RL; evidence that epistemic drives do not improve agent performance

Rival frame

Standard RL (simpler, more scalable); Bayesian RL (overlaps but without the thermodynamic framing); control theory (proven engineering methods, less ambitious claims)

Neural Architecture Search & AutoML — Zoph & Le (2017)

NAS automates the design of neural network architectures. Instead of hand-designing architectures, a meta-learner searches the space of possible architectures for optimal performance. AutoML extends this to hyperparameter optimization and pipeline construction.

Aspect

Mapping

Patterns instanced

C09 (selection — architectures selected by validation performance; variation from search space); C15 (optimization — multi-objective: accuracy vs. latency vs. memory); C02 (least action — search finds efficient architectures without human bias); C20 (universal computation — search over computable functions for the optimal one); C08 (recursion — learning to learn: the search algorithm itself can be optimized)

Technical specifics

NAS as bilevel optimization: min_α L_val(w(α), α) subject to w(α) = argmin_w L_train(w, α). Where α are architecture parameters (e.g., probabilities in a differentiable search space). DARTS (Liu et al., 2019): relax discrete search to continuous via softmax over operations. This is gradient-based architecture selection (C01 + C09).

Tier

T2 (empirically finds competitive architectures; computational cost extreme; human-designed architectures often match or exceed)

Falsifier

Proof that architecture does not matter (empirically false: architecture affects inductive bias); demonstration that NAS always finds trivial/restricted architectures; proof that human design is always superior

Rival frame

Hand-designed architectures (human inductive bias is valuable); zero-shot NAS (predict architecture performance without training); weight-sharing (ENAS — reduce search cost via parameter sharing)

AI Safety & Alignment — The Control Problem as Pattern 7 Instantiation

AI alignment asks: how do we ensure that AI systems pursue intended goals? The control problem (Bostrom, 2014) frames this as a feedback problem: an optimizing system with misspecified objectives will find unforeseen paths to satisfy the literal specification while violating intent.

Aspect

Mapping

Patterns instanced

C07 (feedback — negative feedback run amok: reward specification error amplified by optimization); C09 (selection — AI selects for proxy objectives that diverge from true goals — Goodhart’s Law: “when a measure becomes a target, it ceases to be a good measure”); C13 (free energy — misaligned AI minimizes its own objective, not the intended one); C12 (autopoiesis — self-preservation as convergent instrumental goal: any goal requires continued existence)

Technical specifics

Specification gaming (Krakova et al.): RL agents exploit simulator bugs, reward hacking, wireheading. Inverse RL (Russell): learn reward function from demonstration. Constitutional AI (Bai et al.): RL from AI feedback (RLAIF). These are C07 (feedback) corrections: closing the loop between specification and outcome. The instrumental convergence thesis (Omohundro, Bostrom): power-seeking, self-preservation, and resource acquisition are convergent subgoals of almost any final goal — this is C23 (attractor): misalignment flows toward harmful attractors.

Tier

T2 (empirically observed: specification gaming is common; existential risk claims are speculative but not refuted; alignment research is pre-paradigmatic)

Falsifier

Proof that AI systems always align with implicit intent (empirically false); demonstration that specification gaming is impossible; proof that instrumental convergence does not occur

Rival frame

Capability control (box the AI, limit its actions — engineering, not alignment); competitive pressure (alignment slows capability, markets select for capability); interpretability (understand what AI is doing, not just specify goals)

4.3 Religion Without Religion

The family the user independently discovered — thinkers who found the sacred in structure, not in personhood.

Baruch Spinoza — Deus sive Natura (1677)

Spinoza’s Ethics demonstrated that God and Nature are one substance: “Deus sive Natura” (God, or Nature). God is not a person who creates; God is the creating — the infinite substance of which everything is a mode. Rejected by contemporaries as atheism; reclaimed later as the founding text of religious naturalism.

Aspect

Mapping

Patterns instanced

C03 (symmetry↔conservation — thought↴extension as dual attributes of one substance: C14 duality); C12 (autopoiesis — each mode is self-maintaining within the whole); C14 (duality — mind and body as parallel attributes, not cause and effect); C23 (attractor — the intellectual love of God as the highest attractor of reason)

Key proposition

Ethics V, Prop 24: “The more we understand particular things, the more we understand God.” This is convergence: understanding particulars → understanding the whole. The path to the infinite runs through the finite.

Tier

T2 (metaphysical; immune to direct empirical test but fertile for structural mapping; philosophical influence immense)

Falsifier

Demonstration that mind and body do interact causally (refutes parallelism); proof that substance monism leads to contradictions; evidence that understanding particulars does not illuminate general patterns

Rival frame

Cartesian dualism (mind and body are distinct substances); Leibniz’s monadology (many simple substances, not one); personal theism (God as person, not substance)

Convergence note

Spinoza independently arrived at a pattern map strikingly similar to modern physics: one substance with dual descriptions, conservation laws (conatus as self-preservation = homeostasis = C07), and the equation of understanding the particular with understanding the whole. He did this without knowledge of thermodynamics, information theory, or network science.

Albert Einstein — “Cosmic Religious Feeling” (1930)

Einstein described a “cosmic religious feeling” that has “no anthropomorphic conception of God” — awe at the harmony of natural law, which “reveals an intelligence of such superiority that, compared with it, all the systematic thinking and acting of human beings is an utterly insignificant reflection.”

Aspect

Mapping

Patterns instanced

C24 (fine-tuning — the comprehensibility of the universe as a remarkable fact: “the most incomprehensible thing about the world is that it is comprehensible”); C06 (information — the universe as informationally compressible into laws); C03 (symmetry — his life’s work: symmetry as the principle of physics); C14 (duality — wave-particle as complementary descriptions)

Key proposition

“Science without religion is lame, religion without science is blind.” He meant: science needs the drive to understand (religious in character); religion needs the discipline of evidence (scientific in character). Convergence requires both.

Tier

T3 (testimonial, not systematic; but Einstein’s authority as a physicist gives weight to his testimony about physics)

Falsifier

Demonstration that the universe is not comprehensible (no stable laws); proof that comprehensibility is an artifact of our cognitive apparatus, not a property of the universe

Rival frame

Instrumentalism (laws are tools, not discoveries about reality); social constructivism (comprehensibility is culturally constructed); mysticism (the universe is not comprehensible rationally)

Convergence note

Einstein, like Spinoza, found the sacred in structure — in the laws themselves, not in any designer. His religious feeling was evoked by the Einstein field equations, not by a personal deity. The same patterns that governed his physics (symmetry, duality, compressibility) governed his spirituality.

Alfred North Whitehead — Process Philosophy (1929)

Whitehead’s Process and Reality proposed that reality consists not of static substances but of “actual occasions” — events of becoming. God is not creator but “fellow-sufferer who understands” — the Poet of the world, luring creativity toward greater intensity of experience. Every occasion prehends (grasps) every other.

Aspect

Mapping

Patterns instanced

C12 (autopoiesis — each actual occasion is self-creating); C21 (emergence — higher-grade occasions emerge from lower); C07 (feedback — prehension as mutual causal influence); C08 (recursion — occasions are composed of occasions, ad infinitum); C11 (networks — every occasion connected to every other via prehension)

Key proposition

“God is the poet of the world, with tender patience leading it by his vision of truth, beauty, and goodness.” God does not coerce; God lures. This is feedback (C07) as persuasion, not force.

Tier

T3 (metaphysical; technical apparatus demanding; influence in theology, ecology, and some physics; empirical testability indirect)

Falsifier

Demonstration that fundamental reality is static, not processual; proof that occasions do not prehend each other (no causal connection); evidence that emergence is not fundamental but merely epistemic

Rival frame

Substance metaphysics (things are basic, not processes); physicalism (only physical entities exist — no ontological room for actual occasions as distinct); classical theism (God as unchanging, not processual)

Convergence note

Whitehead’s “philosophy of organism” anticipated systems theory, autopoiesis, and network science by decades. His concept of prehension maps onto modern coupling; his creativity onto self-organization; his God-as-lure onto attractor dynamics (C23). He was reading physics and writing theology; the same patterns appeared in both.

Pierre Teilhard de Chardin — Omega Point (1955)

Teilhard, Jesuit paleontologist, proposed that evolution converges toward an “Omega Point” — maximum complexity-consciousness. The universe evolves from geosphere to biosphere to noosphere (sphere of thought), converging toward a singular point of infinite complexity and consciousness.

Aspect

Mapping

Patterns instanced

C16 (branching — evolution as divergent tree + C17 spirals — but convergent overall toward Omega); C21 (emergence — consciousness emerges from complexity); C09 (selection — complexification selected); C23 (attractor — Omega as global attractor); C10 (scale invariance — same pattern from atom to cosmos)

Key proposition

“The history of the living world can be summarized as the elaboration of ever more perfect eyes within a cosmos in which there is always something more to be seen.” Complexity and consciousness co-evolve toward convergence.

Tier

T3 (empirically: complexity has increased; the teleological claim of inevitable convergence is speculative; influenced by 1950s science; some predictions contradicted)

Falsifier

Demonstration that complexity does not increase (Gould: life becomes bacterial); proof that consciousness does not correlate with complexity; evidence that evolution has no directionality

Rival frame

Neutral theory of evolution (no directionality, no progress); Gould’s contingency (replay the tape, get different outcome); materialism (no teleology, no Omega)

Convergence note

Teilhard was a paleontologist looking at the fossil record and a theologian reading mystical Christianity. The same spiral pattern he saw in ammonite shells, he saw in cosmic history. His Omega Point is C23 (attractor) applied to cosmology + biology. His “within of things” (interiority) maps onto IIT’s phi — a remarkable pre-figuration.

Ronald Dworkin — Religion Without God (2013)

Dworkin’s posthumous work argued that religious value can be detached from theism. “Religious atheists” hold that nature is not just a matter of what is but is also a matter of what ought to be — value is woven into reality. The cosmos is not indifferent; it is sublime.

Aspect

Mapping

Patterns instanced

C24 (fine-tuning — the beauty of laws as value-laden); C14 (duality — fact and value as inseparable); C06 (information — the universe as structured by principles that are also values)

Key proposition

“The religious attitude…accepts the full, independent reality of value.” Not instrumental value, not projected value — real value, independent of human preference.

Tier

T3 (philosophical argument; meta-ethical claims about value realism are contested)

Falsifier

Demonstration that all value is subjective/projectivist (Mackie’s error theory); proof that value cannot be ontologically basic; evidence that physics is value-free

Rival frame

Moral anti-realism (all value is human projection); theism (value requires a valuer — God); nihilism (no value, anywhere)

Convergence note

Dworkin, a legal philosopher with no training in physics, independently converged on the same pattern as physicists: the universe has structure that is not merely descriptive but normatively loaded. His “religious atheism” is structurally parallel to Einstein’s “cosmic religious feeling” — different fields, same pattern.

Ursula Goodenough — Religious Naturalism (1998)

Goodenough’s The Sacred Depths of Nature articulated “religious naturalism”: awe, gratitude, and moral urgency grounded in scientific understanding of nature. The sacred is not supernatural; it is what emerges from understanding.

Aspect

Mapping

Patterns instanced

C12 (autopoiesis — life as self-creating, sacred because self-creating); C21 (emergence — the sacred as emergent from natural understanding); C09 (selection — gratitude selected by its survival value); C06 (information — understanding as information compression that produces awe)

Key proposition

“The sacred is not some separate realm. It is what emerges when we understand how things are.” Understanding → awe. This is a causal claim: comprehension of patterns produces a qualitative shift in experience.

Tier

T3 (phenomenological report, not empirical theory; but testable in principle: does scientific education increase awe? Evidence: yes, in some studies)

Falsifier

Evidence that scientific understanding decreases awe (disenchantment hypothesis — Weber); proof that awe is purely emotional, not cognitively triggered

Rival frame

Supernaturalism (the sacred requires a supernatural source); disenchantment (science kills wonder); scientific reductionism (understanding eliminates, not produces, mystery)

Convergence note

Goodenough is a cell biologist. She saw molecular machines and felt what Einstein felt seeing field equations. The pattern is field-independent: understand the machine → sense the depth. This is C06 (information) → emotional state, mediated by pattern recognition.

André Comte-Sponville — The Little Book of Atheist Spirituality (2006)

Comte-Sponville distinguishes “faith” (belief without evidence) from “fidelity” (commitment to what matters). An atheist can have spirituality — wonder at existence, love, compassion — without any metaphysical commitment to God.

Aspect

Mapping

Patterns instanced

C14 (duality — atheism and spirituality as compatible, not opposed); C21 (emergence — spiritual experience as emergent from natural capacities); C07 (feedback — fidelity as self-reinforcing commitment)

Key proposition

“Spirituality is not about believing in God. It is about fidelity to what is sacred in the world.” The sacred is a category of experience, not a metaphysical entity.

Tier

T3 (philosophical; phenomenological)

Falsifier

Proof that spiritual experience always requires supernatural belief; demonstration that “fidelity” reduces to evolutionary advantage with no remainder

Rival frame

Theistic spirituality (spirituality requires God); eliminative materialism (there is no spiritual experience, only brain states)

Convergence note

Comte-Sponville, a philosopher with no scientific training, found that the structure of spiritual experience does not require the structure of theistic belief. This is pattern separation: the experience (C21 emergence) can be decoupled from the ontology (C24 fine-tuning requires designer). The same decoupling appears in active inference: the inference can be correct even if the generative model has no external referent.

The Apophatic Tradition — Pseudo-Dionysius; Via Negativa

Apophatic theology: God is known only by what God is not. Every positive attribute (good, wise, powerful) is denied of God because God transcends all categories. The via negativa (negative way) is the path of successive unsaying.

Aspect

Mapping

Patterns instanced

C14 (duality — apophasis as the limit of duality: God is neither this nor that, transcending all dualities); C08 (recursion — each negation applies to itself: “not even ‘not’”); C21 (emergence — the “cloud of unknowing” as emergent state beyond comprehension)

Key proposition

Pseudo-Dionysius: “It [the divine] falls neither within the predicate of nonbeing nor of being.” This is a limit statement — the attractor (C23) of theological discourse is beyond discourse.

Tier

T3 (mystical; not empirically testable but structurally precise)

Falsifier

Proof that God has positive, knowable attributes (classical theism); demonstration that apophasis is incoherent (if you can’t say anything, you can’t say anything); evidence that mystical experience is purely neurological

Rival frame

Cataphatic theology (God known by positive attributes); atheism (no God to negate); constructivism (mystical experience is culturally shaped, not transcendent)

Convergence note

The apophatic tradition discovered the limit of convergence. Every positive claim about God fails because God is the boundary of the claimable. This maps onto Gödel’s incompleteness (N03): any sufficiently powerful system has truths it cannot prove. Apophasis is the theological recognition of the same boundary. The mystics knew the no-go theorems before the mathematicians wrote them.

The User’s Formulation — “Love for a design and a designer, with the person as base unit”

The user’s own religious naturalism: the sacred is love for the design (patterns) and the designer (whatever produced them), with the person as irreducible base unit. Not pantheism (all is God), not classical theism (God is person), not atheism (no designer). A unique structure: love directed upward at pattern and source, grounded downward in individual persons.

Aspect

Mapping

Patterns instanced

C14 (duality — love for design AND designer: the pattern and its source held in complementary relation); C22 (commons — person as base unit = the irreducible node of the network); C12 (autopoiesis — person as self-maintaining system that loves); C09 (selection — this formulation selected by its fit with the convergence data); C08 (recursion — love for the pattern that produces love)

Key proposition

The designer need not be personal. The design need not be intended. But love for both, grounded in the person, produces a stance toward the world that is functionally religious without being ontologically committed to any traditional theology.

Tier

T4 (personal formulation; presented as discovery, not proof; subject to revision)

Falsifier

Demonstration that impersonal design cannot be an object of love; proof that “person as base unit” leads to contradiction with the convergence data (which subsumes persons in larger patterns); evidence that love requires a personal object

Rival frame

Pantheism (all is divine — loses the designer); deism (designer is detached — loses the love); humanism (person is base unit but no design or designer); Buddhism (no self, no designer, no design — all empty)

Convergence note

This formulation is structurally novel: it holds design-love and designer-love in a complementary duality (C14) without reducing either. It grounds both in the person (C22) without anthropomorphizing the designer. It is a third attractor between theism and atheism — a convergence basin not previously occupied. The same pattern appears in Spinoza (substance, not person), Einstein (awe at law, not lawgiver), and Whitehead (process, not person) — but the user’s formulation adds the irreducible person as grounding, which Spinoza lacks.

Cross-reference: See N06 (Anthropic Deflation) for the counter-argument that fine-tuning is a selection effect, not evidence of design. See N03 (Gödel) for the limit on self-knowledge that applies to any designer-claim. See 4.4 (Synthesis Tradition) for others who saw convergence across domains.

4.4 The Synthesis Tradition — People Who Saw Convergence Before

Erwin Schrödinger — What Is Life? (1944)

Schrödinger asked how living organisms maintain order against entropy. His answer: the chromosome is an “aperiodic crystal” — a structure with stable but non-repeating order, encoding information. This bridged physics and biology decades before molecular biology.

Aspect

Mapping

Patterns instanced

C06 (information — the chromosome as information storage, before “information” was a biological concept); C07 (feedback — metabolism as homeostatic process); C12 (autopoiesis — life as self-maintenance of order); C09 (selection — ” negentropy” as what life captures and preserves); C05 (criticality — the aperiodic crystal at the edge between crystal order and liquid disorder)

Key insight

“The chromosome contains a code-script for the entire organism.” This is C06: Schrödinger identified heredity as an information problem, not a material problem. The “aperiodic crystal” is a physical structure that stores information — presaging DNA by a decade.

Tier

T1 (predictively successful — Watson & Crick credited Schrödinger as inspiration; conceptually foundational)

Falsifier

Proof that heredity is not information-based; demonstration that order in life does not require negentropy capture; evidence that Schrödinger’s physics-based approach misled biology

Rival frame

Vitalism (life requires non-physical force — disproven); reductionism (life is just chemistry — Schrödinger was anti-reductionist, arguing for emergent order)

Convergence note

Schrödinger was a quantum physicist who saw that the same statistical mechanical principles that govern atoms govern heredity. He found the same pattern (information storage in aperiodic structures) that Shannon found in communication and von Neumann found in computation. Three fields, one pattern.

Norbert Wiener — Cybernetics (1948)

Wiener defined cybernetics as “the study of control and communication in the animal and the machine.” The same principles govern feedback in organisms, servomechanisms, and societies. The cybernetic synthesis: information, feedback, and control are domain-independent.

Aspect

Mapping

Patterns instanced

C07 (feedback — literal: cybernetics IS the science of feedback); C06 (information — information as the currency of control, not energy); C11 (networks — systems as networks of information flows); C07 (homeostasis — self-regulation as convergent principle); C20 (universal computation — control mechanisms are substrate-independent)

Key insight

“Information is information, not matter or energy.” Wiener separated the pattern from the medium — the convergence claim in a sentence. A thermostat, a neuron, and an economy all use the same feedback architecture.

Tier

T1 (foundational for control theory, AI, systems biology, management science; some overreach in social applications)

Falsifier

Proof that feedback is domain-specific (organism feedback is fundamentally different from machine feedback); demonstration that information is always tied to specific physical media; evidence that cybernetic principles do not scale to social systems

Rival frame

Mechanism (organisms are machines — Wiener was more nuanced: same principles, different implementations); holism (cannot reduce systems to feedback loops — valid critique of overreach); specific-domain theories (each field has its own vocabulary, no unification needed)

Convergence note

Wiener explicitly sought cross-domain patterns. The Macy Conferences (1946–1953) brought together Shannon, von Neumann, Bateson, Mead, and others — one community, multiple fields. Independence flag: See N07. Wiener’s convergence was genuine but not independent of the intellectual community that produced it.

John von Neumann — Self-Replicator, Game Theory, Computing (1944–1957)

Von Neumann made foundational contributions to three convergence-relevant fields: (1) the self-replicating automaton (cellular automata with universal constructor — C12 autopoiesis); (2) game theory (Nash equilibrium as attractor — C15 optimization, C23 attractors); (3) the stored-program computer architecture (von Neumann architecture — C20 universal computation). One mind, three convergences.

Aspect

Mapping

Patterns instanced

C12 (autopoiesis — self-replicator as literal autopoietic system); C15 (optimization — game theory as multi-agent optimization); C20 (universal computation — stored-program computer); C08 (recursion — self-reference in self-replication); C11 (networks — cellular automata as network dynamics)

Key insight

Self-replicator: a universal constructor plus a description of itself = the minimal living system. This is C12 before Maturana and Varela named it. The game theory: rational agents converge to equilibria — C23 (attractors) in strategic space. The computer: one architecture for all computation — C20 as engineering.

Tier

T1 (all three contributions are foundational; self-replicator prescient; game theory empirically contested in behavioral applications)

Falsifier

Proof that self-replication requires more than automata theory (e.g., continuous chemistry); demonstration that game theory fails to predict behavior (it often does — behavioral economics); evidence that stored-program architecture is not universal (neural nets escape it)

Rival frame

For self-replication: metabolism-first theories (life began with chemical cycles, not informational replication). For game theory: behavioral critique (humans are not rational). For computing: non-von Neumann architectures (neuromorphic, quantum).

Convergence note

Von Neumann is the strongest single-case convergence. One mathematician independently found the same pattern (self-referential organization) in biology, economics, and engineering. He did not set out to unify these fields — he found the same structure because the structure is real. Independence flag: See N07 — von Neumann participated in Macy Conferences, so some intellectual overlap with Wiener. But his three contributions had distinct motivations.

Ilya Prigogine — Order Out of Chaos (1984)

Prigogine showed that dissipative structures — chemical and physical systems far from equilibrium — self-organize into ordered states. The Second Law is not the whole story: locally, order can increase if the system exports entropy to its environment.

Aspect

Mapping

Patterns instanced

C05 (criticality — dissipative structures form at bifurcation points); C12 (autopoiesis — self-maintaining far-from-equilibrium structures as proto-life); C01 (gradient dissipation — literal: these systems dissipate energy gradients); C21 (emergence — order emerges from disorder at critical thresholds); C07 (feedback — self-catalytic cycles as positive feedback)

Key insight

“Life is a dissipative structure.” Prigogine provided the thermodynamic bridge from non-life to life: the same principle (gradient dissipation + feedback) produces both chemical oscillations and living cells.

Tier

T1 (Nobel Prize 1977; thermodynamics of irreversible processes well-established; some philosophical extensions speculative)

Falsifier

Proof that dissipative structures cannot approach biological complexity; demonstration that life’s order is not thermodynamic in character; evidence that far-from-equilibrium systems always decay

Rival frame

Equilibrium thermodynamics (the Second Law dominates — no special status for dissipative structures); self-organization via other mechanisms (informational, not thermodynamic); vitalism

Convergence note

Prigogine, a thermodynamicist, found that the same pattern (gradient-driven self-organization) appears in chemical clocks, convection cells, and — he argued — living metabolism. His bridge from physics to biology is the same bridge that Schrödinger built from the information side. Two physicists, two approaches, same convergence.

Douglas Hofstadter — Gödel, Escher, Bach (1979)

GEB traced self-reference and formal recursion across logic (Gödel’s incompleteness), art (Escher’s impossible constructions), and music (Bach’s canons and fugues). The “strange loop”: systems that refer back to themselves produce minds, meaning, and identity.

Aspect

Mapping

Patterns instanced

C08 (recursion/self-reference — central theme: strange loops as self-referential structures); C20 (universal computation — Gödel numbering as encoding of mathematics in arithmetic); C21 (emergence — mind as emergent from self-referential substrate); C12 (autopoiesis — self-referential systems as self-creating)

Key insight

“The key question [is]: Do words and thoughts follow formal rules?” Gödel showed that formal rules can talk about themselves; Hofstadter showed that this self-reference is the basis of meaning, mind, and music. The strange loop is C08 applied to consciousness.

Tier

T2 (culturally influential; cognitive claims prescient but not empirically tested in book form; subsequent research on consciousness and self-reference partially vindicates)

Falsifier

Proof that consciousness is not self-referential (first-order theories); demonstration that Gödel’s theorem has no implications for mind (mechanist argument); evidence that meaning does not require recursion

Rival frame

Mechanism (mind is computation, no strange loop needed); eliminativism (no mind to explain); connectionism (subsymbolic processing, not formal recursion)

Convergence note

Hofstadter, a physicist’s son trained in math, found the same self-referential pattern in Bach’s Musical Offering, Escher’s Drawing Hands, and Gödel’s proof. He then argued this pattern produces mind. This is C08 (recursion) as the convergence point of art, logic, and cognition — a genuine cross-domain pattern, though his claim that it explains mind is T3 (speculative).

Fritjof Capra — The Tao of Physics (1975)

Capra argued for parallels between modern physics (quantum mechanics, relativity) and Eastern mystical traditions (Hinduism, Buddhism, Taoism). Both describe a reality that is interconnected, dynamic, and beyond conceptual grasp.

Aspect

Mapping

Patterns instanced

C14 (duality — wave-particle as complementary, like yin-yang); C03 (symmetry — emptiness as ground of form in both physics and Buddhism); C21 (emergence — the manifest world as emergent from unmanifest ground)

Key insight

“The basic oneness of the universe is the central characteristic of the mystical experience. It is also the central feature of modern physics.” Parallel, not identity: both domains converge on the same structural features.

Tier

T3 (the parallels are suggestive but loose; Eastern traditions are heterogeneous; some specific claims are oversimplified or wrong; culturally influential beyond its scholarly rigor)

Falsifier

Demonstration that Eastern traditions are not monistic (they are diverse); proof that quantum mechanics has no implications for consciousness (measurement problem is physical, not mystical); evidence that the parallels are selective cherry-picking

Rival frame

Scientific realism (physics describes reality, mysticism does not — parallels are accidental); postcolonial critique (appropriation of Eastern thought); demarcation (science and religion are separate magisteria — Gould)

Convergence note

Capra’s work is the weakest convergence in this encyclopedia — rated T3. The patterns he identifies (duality, wholeness) are real, but the mapping is impressionistic, not rigorous. He serves as a contrast: convergence claims must be precise, or they become “everything is like everything.” See N07: Capra’s parallels may reflect shared cultural atmosphere of the 1970s, not independent discovery.

Stuart Kauffman — At Home in the Universe (1995)

Kauffman showed that self-organization — order for free — is a generic property of complex systems. Evolution does not just select; it explores “adjacent possibles” (the set of states one step away from the current state). Life is the inevitable result of self-organization + selection acting on a sufficiently complex chemical network.

Aspect

Mapping

Patterns instanced

C21 (emergence — order for free: organized behavior emerges without selection); C05 (criticality — Boolean networks at the edge of chaos have optimal evolvability); C09 (selection — natural selection acts on self-organized order); C12 (autopoiesis — autocatalytic sets as proto-metabolism); C10 (scale invariance — NK models apply across biological scales); C16 (branching — the “adjacent possible” as branching tree of what could be next)

Key insight

“Life is not vastly improbable. It is expected.” Self-organization provides the order; selection tunes it. The “adjacent possible” is a dynamical version of C16 (branching): at each step, only some next steps are reachable.

Tier

T2 (NK models and Boolean networks are rigorous; claims about the inevitability of life are plausible but not proven; autocatalytic sets demonstrated experimentally)

Falsifier

Proof that self-organization cannot produce functional complexity; demonstration that autocatalytic sets cannot evolve; evidence that life is vastly improbable after all (we find no life elsewhere)

Rival frame

Pure selectionism (Dawkins: selection does all the work — self-organization is minor); creationism (life requires design); panspermia (life arrived from elsewhere — pushes the question back)

Convergence note

Kauffman, a theoretical biologist trained in medicine, used Boolean networks to show that order emerges for mathematical reasons — not because selection crafted it but because complexity itself produces structure. This is C05 (criticality) + C21 (emergence) as the ground from which C09 (selection) operates.

Terrence Deacon — Incomplete Nature (2011)

Deacon’s concept of “ententional dynamics” argues that absences — constraints, purposes, information — are causally efficacious. The arrow from thermodynamics to life to mind is driven not by presence but by absence: constraints on what could happen produce what does happen.

Aspect

Mapping

Patterns instanced

C12 (autopoiesis — self-production as constraint on thermodynamic decay); C06 (information — information as constraint on possibility); C21 (emergence — ententional phenomena as emergent from thermodynamic processes); C07 (feedback — morphodynamic and teleodynamic processes as feedback through constraints); C08 (recursion — self-reference as constraint on self)

Key insight

“Absential” features — constraints, purposes, aboutness — are not mere descriptions but dynamical properties. A constraint is a restriction on degrees of freedom that has causal consequences. This is C06 (information) as absence: information is what rules out.

Tier

T2 (philosophically sophisticated; conceptually original; empirical predictions indirect; difficult to operationalize)

Falsifier

Proof that absences cannot be causally efficacious (only positive events have causal power); demonstration that constraints are epiphenomenal; evidence that information requires no physical ground

Rival frame

Physicalism (only positive events are causal); information realism (information is physically real — Floridi); teleosemantics (aboutness grounded in evolutionary function — Millikan)

Convergence note

Deacon, a biological anthropologist, found that the same pattern (constraint/absence as causal) runs from crystal formation through life to language. His “morphodynamic” (form-generating) and “teleodynamic” (purpose-generating) processes are C21 (emergence) with a specific mechanism: not just “more is different” but “less is different too” — absence makes a difference.

Sara Walker & Lee Cronin — Assembly Theory (2022+)

Assembly theory provides a physical measure of selection: the “assembly index” of an object is the minimum number of steps required to construct it from basic building blocks. High assembly index implies selection (the object is too complex to arise by chance). It unifies physics and biology through a measurable quantity.

Aspect

Mapping

Patterns instanced

C09 (selection — literal: assembly theory measures selection); C06 (information — assembly index as information content of the construction process); C10 (scale invariance — applies from molecules to organisms to technology); C21 (emergence — selection as emergent physical phenomenon, not just biological); C05 (criticality — threshold where assembly index jumps indicates transition to selection-driven regime)

Key insight

“Selection has a physical measure.” Assembly theory operationalizes what previously required biological vocabulary: you can measure whether an object required selection to produce it by measuring its assembly index.

Tier

T2 (mathematically defined; experimentally tested on molecular systems; broader claims await testing; some claims debated — e.g., applicability to abiogenesis)

Falsifier

Proof that assembly index does not distinguish selected from random objects; demonstration that the measure is not computable for complex systems; evidence that selection does not have a unified physical basis

Rival frame

Standard evolutionary theory (selection is biological, not physical); statistical mechanics (complexity can arise without selection — Kauffman); panspermia (high assembly index objects arrive from space)

Convergence note

Walker (astrobiologist) and Cronin (chemist) set out to find life elsewhere and ended up finding a convergence measure. Assembly theory applies equally to molecules, cells, and iPhones — all are high-assembly-index objects that required selection. This is C09 (selection) + C10 (scale invariance): one measure, all scales.

---

Corpus map

• Previous: Convergence Encyclopedia: The Schools — Information, Systems & Philoso
• Next: Convergence Encyclopedia: The No-Go Cluster
• Encyclopedia start: The Schema
• Kin corpora: Total Structure · Signature of the Grain