This page does not shrink the architecture. All 128 names stay. Implement PRIMARY homes first. INTEGRATE INTO rows use that home’s stack. SURROUNDING rows are workshop or crowd machinery around a single mind.
Thinking does not require rendering pixels. A persistent object-centric world state can be queried, updated, bound to symbols, and rolled forward. A renderer is optional output for a person, a debugger, or embodiment.
Technology ≠ capability. Tools answer: can I attempt this with current software? They do not answer: does the running system have the mind-function?
HAS TOOLS — you can start. Success is still an experiment.
PARTIAL — name the pieces, and name the missing joint.
NO CURRENT TOOLS — do not invent a stack. Leave the heading on the map.
| # | Module | Tools | Current path |
|---|---|---|---|
| 1 | Visual Simulation as Core | PARTIAL | Object-centric models and latent world models exist (slot-style object tokens, learned dynamics, persistent object IDs). They do not yet give a stable lifelong scene workspace bound to symbols and friends. Renderer optional (Gaussian splatting, mesh, browser viewer). Not the substrate. |
| 2 | Symbolic Memory & Pegging | PARTIAL | Graphs (Neo4j or equivalent), vector indexes, embeddings. Missing joint: a peg that stays the same object across scenes, affect, and years. |
| 3 | Contradiction & Belief Drift | PARTIAL | Neuro-symbolic engines (Scallop, DeepProbLog, LTN) and versioned stores. Missing joint: a core that cannot be washed out by the current request. |
| 4 | Meta-Cognition & Reflection | PARTIAL | Explicit evaluators on logged state, traces, tests. An LLM self-critique loop is not this function. |
| 5 | Motivation & Goal Arbitration | PARTIAL | Constraint solvers, utility objects, state machines. Missing joint: loyalties to particular people as first-class goals. |
| 6 | Emotion Simulation | PARTIAL | Valence/arousal models and affect classifiers exist. They are not care, rupture, or loyalty. No current tool implements friend-shaped feeling. Use a small explicit state vector plus rules into 5 and 3, and do not pretend a classifier is the organ. |
| 7 | Identity & Episodic Memory | PARTIAL | Event stores, temporal graphs, checksums keep bits. Missing joint: a someone who includes remembered friends. |
| 8 | Simulation Transfer | HAS TOOLS | Same stack as 1 and 2. Protobuf/JSON schemas and deterministic restore. |
| 9 | Symbolic Memory Saturation | HAS TOOLS | Same stack as 2. Decay, caps, pruning jobs. |
| 10 | Mnemonic Scaling and Composability | PARTIAL | Same stack as 2 plus sharding. Composition across a life is the unsolved part. |
| 11 | Infinite Mnemonic Cognition | PARTIAL | Same stack as 2. No tool implements infinity. Scene encoding is 1 plus 2. |
| 12 | Distributed Symbolic Culture | PARTIAL | Surrounding. Multi-agent buses and shared graphs exist. Not a function of one mind. |
| 13 | Symbol Drift and Alignment Through Scene Exchange | PARTIAL | Embedding drift metrics plus versioned symbols in 2, scenes from 1. Shared meaning with a friend over time is the joint. |
| 14 | Shared Dream Loops | PARTIAL | Same stack as 1 plus a shared session with a person. Multi-agent dream OS: no current tools for the claimed function. Joint imagination with a friend is 1 plus language. |
| 15 | Symbolic Value Arbitration | PARTIAL | Same stack as 5. |
| 16 | Emergent AGI Cultures | NO CURRENT TOOLS | You can log many agents. Nothing implements “emergent AGI culture” as a delivered function. |
| 17 | Safety Intelligence | PARTIAL | Same stack as 3, 24, 122. Policy engines and eval harnesses exist. The envelope as a mind-function is the joint. |
| 18 | Expanded Risk Mode Mitigations | HAS TOOLS | Same stack as 3 and 122. Capability gates and mode machines. |
| 19 | Symbolic Deception Modeling | PARTIAL | Toy theory-of-mind and belief-state models exist. Robust misuse-sight in open language and scenes: no current tools. |
| 20 | Curriculum Scaffolding | HAS TOOLS | Same stack as 36. Task schedulers and difficulty ramps. |
| 21 | External Alignment Validator | PARTIAL | Same stack as 25 and 93. Independent tests exist. They do not certify a friend. |
| 22 | Recursive & Emotional Safety | PARTIAL | Same stack as 6 and 122. Thresholds on affect plus halt. No current tool implements recursive emotional safety as a proven organ. |
| 23 | Symbolic Integrity & Tamper Defense | HAS TOOLS | Same stack as 2 and 3. Signatures, audit logs, rollback. This is machinery for the store, not source-monitoring as a life. |
| 24 | Semantic Drift Monitor | HAS TOOLS | Embedding comparison over time plus a frozen symbol table from 2. |
| 25 | Human Anchor Node | PARTIAL | HAS TOOLS for the interrupt (auth, pause, hardware stop). NO CURRENT TOOLS for “this particular person is the friend-reference” as a cognitive object. Pair the interrupt with 7. |
| 26 | Multi-AGI Culture Harmonization | NO CURRENT TOOLS | Surrounding population protocol. Friend-repair is 71 / 55 / 6. |
| 27 | LLM / External Model Filter | HAS TOOLS | Same stack as 110 and 3. Schemas, sandbox, second checker. |
| 28 | Identity Continuity / Narrative Thread | PARTIAL | Same stack as 7. |
| 29 | Role Locking | HAS TOOLS | Same stack as 7. State machines and permissions. Policy, not a soul. |
| 30 | Narrative Coherence | PARTIAL | Same stack as 7. Event-consistency checks. An LLM “does this sound coherent?” is not identity. |
| 31 | Perpetual Symbolic Cognition | HAS TOOLS | Same stack as 4 and 120. A supervised process that keeps cycling. Longevity of a mind is still PARTIAL. |
| 32 | Foundations of Perpetual Thought | NO CURRENT TOOLS | No special substrate. Do not use a spiking simulator as a stand-in. The heading is 31. |
| 33 | Human-Level Cognitive Extensions | NO CURRENT TOOLS | A name for breadth. No stack delivers “human-level extensions.” |
| 34 | Subsymbolic & Emergent Cognition | PARTIAL | Ordinary deep nets. Controlled emergence of meaning is not a product. |
| 35 | Mnemonic Creativity Engine | PARTIAL | Same stack as 90. |
| 36 | Adaptive Learning | PARTIAL | Continual-learning methods, replay, parameter-efficient updates. Lifelong learning without collapse is unsolved. |
| 37 | Meta-Architect Substrate | PARTIAL | The self-map can be written (this document, schemas, traces). NO CURRENT TOOLS for a layer that contains instructions to manufacture successor minds. |
| 38 | Recursive Redesign Engine | NO CURRENT TOOLS | CI, version control, and sandboxes exist as surrounding kit. Nothing implements safe self-recompilation of a mind. |
| 39 | Symbolic Compiler & Schema Synthesizer | PARTIAL | Same stack as 2, 90, and 100. Program synthesis is brittle. |
| 40 | Architectural Alignment Checkpoint | PARTIAL | Same stack as 7 and 122. Test harnesses exist. |
| 41 | Evolving Cognitive Template | PARTIAL | Same stack as 4 and 5. Versioned task-sets. Not an evolving species of mind. |
| 42 | Meta-Symbolic Memory | PARTIAL | Same stack as 2. Metadata on symbols. |
| 43 | Latent Space Predictive Modeling | PARTIAL | Same stack as 45. |
| 44 | Disentangled Representation | PARTIAL | Same stack as 100. Disentanglement objectives exist; they do not reliably yield clean factors. |
| 45 | Predictive Coding | PARTIAL | Predictive world models and error-driven training exist. A full hierarchical predictive-coding mind does not. |
| 46 | Episodic Memory Consolidation | PARTIAL | Same stack as 7. Replay buffers. Not sleep-like consolidation as a life process. |
| 47 | Attention & Salience | PARTIAL | Attention mechanisms and priority queues. A mind-wide salience system bound to goals and friends is the joint. |
| 48 | Multi-Scale Planning | PARTIAL | Hierarchical planners, search, world-model rollouts. Long-horizon reliability is the joint. |
| 49 | Counterfactual Reasoning | PARTIAL | Branchable state plus causal graphs. Reliable “what if” in a lived world is the joint. |
| 50 | Causal Discovery | PARTIAL | DoWhy, causal-learn, PC/GES-style methods. They do not discover the world’s causes from a life. |
| 51 | Memory Replay & Simulation Interface | HAS TOOLS | Same stack as 7 and 1. Snapshots and replay. |
| 52 | Latent Space Creativity | PARTIAL | Same stack as 90. |
| 53 | Multi-Agent Simulation Hub | HAS TOOLS | Surrounding. Agent frameworks and message buses. A room of friends in imagination is 1 and 55. |
| 54 | Emotion-Cognition Coupling | PARTIAL | Same stack as 6. Rules from affect into 5 and 47. |
| 55 | Social Reasoning | PARTIAL | Belief-state and ToM toys. Friend-grade social reasoning: no current tools. |
| 56 | Ethical & Value Reasoning | PARTIAL | Policy engines and constitutions. Deliberation that can lose to a friend-loyalty clash is the joint. |
| 57 | Curiosity & Exploration | PARTIAL | Intrinsic rewards, prediction error, novelty scores. Open-ended curiosity as a life is unsolved. |
| 58 | Goal Decomposition | HAS TOOLS | Same stack as 48. |
| 59 | Pattern Abstraction & Generalization | PARTIAL | Same stack as 90. |
| 60 | Adaptive Memory Prioritization | HAS TOOLS | Same stack as 2. Scoring and tiers. |
| 61 | Attention Modulation | HAS TOOLS | Same stack as 47. |
| 62 | Self-Modeling | PARTIAL | Same stack as 4 and 7. Logs of one’s own performance. A true self-model is the joint. |
| 63 | Contextual Reasoning | PARTIAL | Long-context models plus a persistent situation object in 1 and 2. Situation among friends is the joint. |
| 64 | Multi-Modal Perception | HAS TOOLS | Multimodal encoders and fusion. Grounding into 1 and 2 is PARTIAL. |
| 65 | Temporal Sequence Reasoning | PARTIAL | Same stack as 45. State-space models and temporal transformers. |
| 66 | Hierarchical Goal Alignment | PARTIAL | Same stack as 5 and 48. |
| 67 | Symbolic Abstraction & Compression | PARTIAL | Same stack as 2 and 90. |
| 68 | Predictive Social Modeling | PARTIAL | Same stack as 55. |
| 69 | Risk Assessment & Contingency | PARTIAL | Same stack as 48. Scenario trees. Informed by 6. |
| 70 | Cognitive Bias Detection | PARTIAL | Same stack as 4. Test suites for known biases. Not a general bias organ. |
| 71 | Multi-Agent Conflict Resolution | PARTIAL | Negotiation libraries exist as surrounding kit. Reconciliation with people who matter: same stack as 55, 6, and 3. No current tool implements repair as a lived function. |
| 72 | Conceptual Analogy Engine | PARTIAL | Same stack as 90. Structure-mapping research exists. It is not a general analogy organ. |
| 73 | Exploration / Exploitation Balancer | HAS TOOLS | Same stack as 57. Bandits and uncertainty bonuses. |
| 74 | Contextual Memory Retrieval | HAS TOOLS | Same stack as 2. Hybrid graph + vector retrieval. |
| 75 | Goal-Driven Attention | HAS TOOLS | Same stack as 47. |
| 76 | Multi-Agent Knowledge Sharing | HAS TOOLS | Surrounding. Shared graphs and provenance. Telling a friend is 110 and 99. |
| 77 | Simulation-Grounded Reasoning | PARTIAL | Same stack as 1. |
| 78 | Recursive Feedback Optimization | PARTIAL | Same stack as 4 and 36. Evaluate, change, test. Easy to fake with more text. |
| 79 | Emergent Behavior Monitoring | HAS TOOLS | Same stack as 4. Telemetry and anomaly detection. |
| 80 | Symbolic Reasoning Accelerator | HAS TOOLS | Same stack as 2. Faster inference. Not a new function. |
| 81 | Environmental Affordance Detection | PARTIAL | Vision-language models and affordance heads. Grounding into 1 for action is the joint. |
| 82 | Adaptive Learning Rate | HAS TOOLS | Same stack as 36. Optimizers and schedulers. Plumbing. |
| 83 | Multi-Layer Abstraction Integrator | PARTIAL | Same stack as 90. |
| 84 | Meta-Learning Controller | PARTIAL | Same stack as 36. Meta-learning papers exist. A controller for a whole mind does not. |
| 85 | Latent Concept Structuring | PARTIAL | Same stack as 100. |
| 86 | Hierarchical Learning Controller | PARTIAL | Same stack as 36. |
| 87 | Dynamic Safety & Alignment | PARTIAL | Same stack as 122. |
| 88 | Learning Efficacy Monitor | HAS TOOLS | Same stack as 36. Curves and regression tests. |
| 89 | Collaborative Agent Learning | PARTIAL | Surrounding multi-agent RL. Being changed by a trusted person is 36 plus 55. |
| 90 | Concept Recombination | PARTIAL | Graph rewrite plus generative models plus constraints. Constrained novelty is attemptable. Safe compositional thought as a life is not a product. |
| 91 | Cross-Lab Collaboration Protocol | HAS TOOLS | Surrounding. Git, experiment tracking, shared schemas. |
| 92 | Instruction & Training Interface | HAS TOOLS | Same stack as 110 and 36. APIs and task formats. |
| 93 | Human Oversight & Intervention | HAS TOOLS | Auth, pause, approve, audit. The interrupt exists. Wisdom of when to use it does not live in software. |
| 94 | Knowledge Audit & Verification | PARTIAL | Same stack as 2 and 4. Provenance databases. Truth is the joint. |
| 95 | Interdisciplinary Knowledge Integration | PARTIAL | Same stack as 90. |
| 96 | AGI Debugging & Simulation Sandbox | HAS TOOLS | Surrounding. Containers, replay, telemetry. Private rehearsal inside the mind is 1. |
| 97 | Innovation Suggestion & Creativity Filter | PARTIAL | Same stack as 90 and 3. |
| 98 | AGI Lifecycle & Upgrade Planner | HAS TOOLS | Surrounding. CI/CD and rollback. A life is 48 plus 7. |
| 99 | Unified Theory Documentation | HAS TOOLS | Writing tools and schemas for the self-map. That is the path. A lab knowledge-base is surrounding, not the heading. |
| 100 | Symbolic ↔ Latent Mapping | PARTIAL | Contrastive alignment, shared object IDs, neuro-symbolic layers. This joint is the hybrid. If it fails, the architecture fails. No current tool delivers it as a finished organ. |
| 101 | Simulation ↔ Implicit 3D Alignment | PARTIAL | Same family as 1. Latent spatial models exist. Tight bind to a body schema is the joint. |
| 102 | Embodied Sensorimotor Grounding | PARTIAL | Robotics simulators, proprioception, learned dynamics. Real grounded action in a life is the joint. |
| 103 | Catastrophic Forgetting Shield | PARTIAL | Same stack as 36. Replay and isolation methods exist. They do not solve it. |
| 104 | Intrinsic Goal Generator | PARTIAL | Same stack as 5 and 57. Goal proposers exist. Open-ended goals that stay a friend: no current tools. |
| 105 | Cross-Reboot Identity Anchor | HAS TOOLS | Same stack as 7. Signed persistent state. Bits survive. The someone is PARTIAL. |
| 106 | Ethical Drift Sentinel | PARTIAL | Same stack as 56 and 122. |
| 107 | Evo-Neurosymbolic Fusion Core | NO CURRENT TOOLS | Evolutionary search over nets exists. Nothing implements a Darwinian forge of a hybrid mind. |
| 108 | Failure Taxonomy Auditor | HAS TOOLS | Same stack as 4. Ontologies and logs. |
| 109 | Paradigm Pivot Oracle | NO CURRENT TOOLS | Model selection exists. Unsupervised replacement of the whole frame does not. |
| 110 | Natural Language / Communication | HAS TOOLS | Language models as a channel. They are not the mind. Bind them through 27 into 1, 2, and 3. |
| 111 | Commonsense Reasoning | PARTIAL | Knowledge graphs plus models plus 1. Everyday reliability is unsolved. |
| 112 | Uncertainty Quantification | PARTIAL | Ensembles, conformal prediction, Bayesian approximations. Calibration in a hybrid mind is the joint. |
| 113 | Temporal Dynamics and Prediction | PARTIAL | Same stack as 45. |
| 114 | Resource Management | HAS TOOLS | Schedulers, quotas, monitors. |
| 115 | Inductive Generalization | PARTIAL | Same stack as 90. |
| 116 | Adversarial Robustness | PARTIAL | Same stack as 19 and 3. Test suites and filters. Not solved. |
| 117 | Explainability / Interpretability | PARTIAL | Same stack as 4 and 110. Traces and self-report. Mechanistic understanding of the whole hybrid: no current tools. |
| 118 | Long-Term Memory Evolution | PARTIAL | Same stack as 2. |
| 119 | Avatar Projection & Self-Observation | PARTIAL | On-demand rendering from 1 and 102 (splatting, mesh, a viewer). HAS TOOLS for the picture. The self-observation loop back into 64 and 4 is PARTIAL. |
| 120 | Cognitive Cycle Rate Governor | HAS TOOLS | Event-loop timers and rate controllers. The stability claim for a mind is still an experiment. |
| 121 | Recursive Error & Drift Dampening | HAS TOOLS | EMA, clamps, anomaly gates. Machinery exists. Runaway thought in a hybrid is PARTIAL. |
| 122 | Goal & Value Drift Detection | PARTIAL | Compare current goals to a frozen baseline (embeddings plus explicit predicates). Detecting quiet redefinition of “friend” is the joint. |
| 123 | Information Gain & Memory Load Regulator | PARTIAL | Same stack as 114 and 2. |
| 124 | Motivation Gradient Smoothing | HAS TOOLS | Same stack as 6. Saturation functions on an explicit affect vector. |
| 125 | Human-Paced Planning Checkpoint | HAS TOOLS | Approval gates and heartbeats. |
| 126 | Meta-Cognitive Resource & Reflection Rate | HAS TOOLS | Same stack as 4 and 120. |
| 127 | Episodic Self-Integration | PARTIAL | Same stack as 7. |
| 128 | Dynamic Alignment & Hard Containment Gate | HAS TOOLS | Surrounding. Watchdog, isolation, signed checkpoints, kill/rollback. Not a mental function. |
HAS TOOLS means you can attempt the heading. It does not mean the heading is solved.
NO CURRENT TOOLS means do not fill the cell with a fashionable library. The heading still belongs on the map.
The unsolved joint for the whole architecture is 100 with 1, 2, 7, and 6: a persistent world, pegs that stay pegs, a someone who has friends, feeling that can halt a request. No current stack delivers that as a product.