Reasoning Kingdom Map
"Boundaries are not endpoints; they are the starting point of design." — Professor Manul
Welcome to the Reasoning Kingdom. This is not an ordinary book, but a thought laboratory. This map will help you understand the structure, narrative threads, and core ideas of the entire work.
Overall Structure: A Trilogy
The Reasoning Kingdom is now a complete trilogy:
- Prequel: To the Future Reasoning Scientist — An introduction to reasoning science for young readers and amateur enthusiasts
- Volume I: The Historical Evolution of Reasoning — From entropy to boundaries, pursuing the nature of reasoning along historical threads
- Volume II: The Formal Deduction of Reasoning — Starting from formal systems, rebuilding the Reasoning Kingdom through logical deduction
Prequel: Introduction to Reasoning Science (For Beginners)
Target Readers
- Young readers: Beginners curious about scientific reasoning, algorithmic thinking, and AI
- Amateur enthusiasts: No specialized background needed — just curiosity and a willingness to "get your hands dirty"
- Pre-university students: Build foundations in mathematics and programming through Manul Academy
Four-Part Structure
| Part | Chapters | Core Content |
|---|---|---|
| Part I The Deterministic Universe | Ch1-7 | From Boolean logic to dynamic programming, establishing an algorithmic thinking framework |
| Part II Crossing the Fault Lines of Logic | Ch8-9 | The mathematical leap from discrete symbols to continuous vector spaces |
| Part III The Emergence of Neural Networks | Ch10-15 | Complete understanding from neurons to Transformers |
| Part IV Toward the Reasoning Kingdom | Ch16-18 | The main trilogy awaits; letters from Professor Manul |
Manul Academy (Supplementary Content)
- Comprehensive Mathematical Foundations: 7 chapters from natural numbers to fixed-point theory
- AI Mathematical Foundations: 5 chapters from probability & statistics to linear models
- Calculus: 12 entries from functions to differential equations
- Linear Algebra: 12 entries from vectors to Jacobian matrices
- Philosophy: 25 entries from ancient Greece to 1840
- Python Programming: 10 entries from syntax to data structures
Overall Goal: Pursuing the Nature of Reasoning
Core question: When AI "reasons," is it really reasoning? What is reasoning?
The book's central proposition: How reasoning evolved from a survival strategy against entropy, to an algorithmic process constrained by computational complexity, rigorously reconstructed within formal systems, and finally halted at the boundary of self-reference — and how we must continue forward in uncertainty.
Volume I: The Historical Evolution of Reasoning (Question-Driven, Intuition First)
Part I: The Origins of Reasoning (Chapters 1-6)
Core question: Where does reasoning come from? Why do we need reasoning?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch1 Entropy and Survival | Understand the survival necessity of reasoning | Reasoning as a survival strategy against the Second Law of Thermodynamics | Information theory, thermodynamics |
| Ch2 The Dawn of Symbolism | The first formalization of causality | If-Then rules as the logical foundation | Symbolic logic, expert systems |
| Ch3 Vector Space | From discrete to continuous representation | The Word2Vec revolution: geometrization of semantics | Word embeddings, distributional hypothesis |
| Ch4 Manifold | Discovering hidden order in high-dimensional data | The manifold hypothesis: data lies on low-dimensional manifolds | Dimensionality reduction, locally linear embedding |
| Ch5 The Fitting Trap | Distinguishing statistics from understanding | Statistical correlation ≠ genuine understanding | Overfitting, generalization theory |
| Ch6 The Causal Boundary | Understanding the inaccessibility of causality | Observational data is never enough | Pearl's causal ladder |
Part II: The Mechanisms of Reasoning (Chapters 7-11)
Core question: How does AI implement reasoning? What are its limitations?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch7 Complexity | Understanding computational asymmetry | P vs NP: the fundamental asymmetry between verification and search | Complexity theory, SAT problem |
| Ch8 Heuristics | Accepting the contract of "good enough" | Heuristics as pragmatic compromise | A* algorithm, admissibility |
| Ch9 Transformer | Reconstructing the infrastructure of reasoning | Attention mechanism: parallelizing global relationships | Self-attention, multi-head attention |
| Ch10 Search | Navigating uncertainty | Monte Carlo Tree Search and AlphaGo | MCTS, policy networks |
| Ch11 Efficiency | The economics of reasoning | Touching the physical limits of compression | Mamba, linear attention |
Part III: The Boundaries of Reasoning (Chapters 12-13)
Core question: Where are the limits of reasoning?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch12 Implicit Reasoning | Understanding the internal processes of neural networks | Reasoning processes in the hidden layers of neural networks | Activation pattern analysis |
| Ch13 The Boundary | Recognizing the essential incompleteness of reasoning | A computational version of Godel's theorem | Yonglin Formula, meta-layer rupture |
Volume II: The Formal Deduction of Reasoning (Logical Deduction, Solid Foundation)
Formal System Foundation (Chapters 14-15)
Core question: How do we give reasoning an unambiguous foundation?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch14 Formal Systems | Building the axiomatic foundation of reasoning | Propositions, inference rules, axioms, proofs | First-order logic, syntax-semantics separation |
| Ch15 Godel Incompleteness | Understanding the hard boundaries of formal systems | Any sufficiently strong system contains undecidable true propositions | Godel numbering, self-reference construction |
Logical Extensions (Chapters 16-18)
Core question: How do we expand the expressive power of reasoning?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch16 Linear Logic | Introducing resource awareness | Each assumption used exactly once | Linear logic, resource management |
| Ch17 Probabilistic Reasoning | From Boolean to probability | Truth values from {0,1} to [0,1] | Probability theory, Bayesian inference |
| Ch18 Causal Formalization | Distinguishing observation from intervention | Formalizing Pearl's causal ladder | do-calculus, structural causal models |
Computation Theory (Chapters 19-21)
Core question: What is the relationship between reasoning and computation?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch19 Complexity Theory | Understanding the geometry of reasoning | Strict definitions of P, NP, PSPACE | Complexity classes, reductions |
| Ch20 The Heuristic Contract | Giving precise definitions to "good enough" | Admissibility, consistency, PAC learning | Approximation algorithms, theoretical guarantees |
| Ch21 Learning Theory | Understanding the nature of generalization | Learning as inverse inference | VC dimension, PAC learning |
Self-Reference and Emergence (Chapters 22-24)
Core question: What happens when a reasoning system begins to reason about itself?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch22 Self-Reference and Emergence | Exploring the meta-layer of reasoning | When reasoning systems reason about themselves | Curry-Howard correspondence, fixed-point theorems |
| Ch23 Yonglin-Lyapunov Correspondence | Describing the dynamics of reasoning systems | Reasoning as a dynamical system | Stability analysis, convergence boundaries |
| Ch24 Reasoning Convergence Through the Lens of Category Theory | Explaining the structural necessity of reasoning convergence with category theory | Ghost pointer corresponds to terminal objects; absent adjoint functors cause meta-layer rupture | Category theory, adjoint functors, Lyapunov functors, categorical interpretation of self-attention |
Conclusion (Chapter 25)
Core question: How do the eight boundaries unify? What is the nature of reasoning?
| Chapter | Goal | Key Idea | Methods |
|---|---|---|---|
| Ch25 The Unification of Boundaries | The culmination of the entire book — unifying the eight boundaries, proposing the Impossible Triangle, returning to the meta-question | All boundaries are the same boundary projected in different directions; AI's reasoning is a layered, bounded, conditional spectrum of capabilities | Boundary Unification Framework, Impossible Triangle Conjecture, Four-Level Answer |
Three Major Intellectual Pillars
1. The Pragmatic Thread
- Origin: Reasoning is a survival tool (Ch1)
- Development: The need for heuristic compromise (Ch8)
- Constraint: Bounded by economic cost (Ch11)
2. The Representation Thread
- Symbols: Discrete logical representation (Ch2)
- Vectors: Continuous geometric representation (Ch3)
- Manifold: Low-dimensional hidden structure (Ch4)
- Attention: Dynamic associations (Ch9)
- Formal Systems: Rigorous axiomatization (Ch14)
3. The Cognitive Limits Thread
- The Fitting Trap: Statistics ≠ understanding (Ch5)
- Causal Inaccessibility: Observation ≠ causation (Ch6)
- The Iron Law of Complexity: The structural nature of P ≠ NP (Ch7)
- The Godel Boundary: The limits of formal systems (Ch15)
- The Self-Reference Dilemma: The inaccessibility of the meta-layer (Ch22)
Meta-Questions Throughout the Book
Progressive Questions
- Early: AI is fitting patterns (Ch5)
- Middle: AI is using heuristics to search the solution space (Ch8, Ch10)
- Later: AI's hidden-layer computations are equivalent to reasoning under certain conditions (Ch12)
- Formalization: Reasoning can be strictly defined as proof in a formal system (Ch14-17)
- Finally: But every reasoning system has insurmountable boundaries (Ch13, Ch15, Ch22)
Levels of Answer
- Technical layer: Algorithms, architectures, optimization
- Theoretical layer: Complexity, computability, learning theory
- Philosophical layer: Understanding, causality, self-reference, boundaries
Original Research Anchors
This book's narrative is built around 6 original research contributions:
- QMCB / OpenXOR (Ch7): Phase diagram of NP problems in continuous space
- Yonglin Formula (Ch12, Ch13): The essential incompleteness of AI reasoning
- ADS (Ch8, Ch10): Information-theoretic formulation of heuristic weights
- Collins Optimizer (Ch11): Touching the physical limits of compression
- Attention Causal Topology (Ch9): Transformer as an implicit causal inference machine
- CocDo (Ch18 Extra): Implementing Pearl's causal calculus as lambda calculus
Reading Path Suggestions
By Reader Type
- Beginners / Young readers: Start from the prequel -> Volume I -> Volume II (step by step)
- Engineers / Practitioners: Prioritize Volume I (Ch1-13), focus on "do it yourself" sections
- Researchers / Theorists: Prioritize Volume II (Ch14-24), focus on original research
- Students / Self-learners: Read sequentially: prequel -> Volume I -> Volume II
By Interest
- To understand the current state of AI reasoning: Prequel Ch10-15 -> Volume I Ch9 -> Volume I Ch12
- To understand the theoretical foundations: Volume II Ch14 -> Volume II Ch15 -> Volume II Ch19
- To explore open questions: Volume II Ch22 -> Ch25 -> each chapter's "Unresolved" sections
- To build algorithmic thinking: Prequel Ch1-7 -> Volume I Ch7-8
By Time Available
- Quick start: Prequel Ch1 -> Prequel Ch10 -> Volume I Ch9
- Systematic study: Follow the trilogy order: prequel -> Volume I -> Volume II -> Ch25 (Conclusion)
- Topic study: Follow the three major intellectual pillar threads
- Quick reference: Map -> Professor Manul's Mini-Dictionary -> Relevant chapters -> Ch25 Conceptual Map
Professor Manul's Advice
"This book will not give you answers, but it will give you better questions. The boundary of reasoning is not the endpoint of knowledge, but the starting point of wisdom. When you understand why some questions will never have answers, you begin to truly understand reasoning."
Remember: Confusion is normal. If you feel more confused after finishing a chapter, you are reading correctly. The expedition through the Reasoning Kingdom begins with admitting your own ignorance.