TMM可视化流程图：三层结构、判定标准与典型反例

摘要：
本流程图以TMM（真理层–模型层–方法层）为核心，构建了科学主张的层级判定与约束反馈机制。真理层拥有公理主权与一票否决权，模型层须具备数学硬度与边界条件，方法层仅作为检验工具，严禁僭越。图中明确示例与反例，并展示自上而下的硬约束（红箭头）与自下而上的信息反馈（绿箭头）。凡无法归入三层或声称方法层可否定真理层者，一律归为伪科学拒收。该图为TMM从理论到操作的可视化工程蓝图。

TMM 可视化流程图：三层结构、判定标准与典型反例

以下 Mermaid 流程图完整呈现了 TMM（真理层–模型层–方法层）的逻辑架构，包含：

• 三个主层的核心定位与判定标准

• 各层的典型反例（即层级僭越的案例）

• 自上而下的约束关系（真理层→模型层→方法层）

• 自下而上的反馈路径（方法层→模型层→真理层）

• 最终的“接收”与“拒收”判定门

直接复制以下代码到支持 Mermaid 的编辑器（如 Typora、GitHub Markdown、Notion 等）中渲染。

图表

代码

flowchart TD
    Start([科学主张 / 研究实践]) --> LayerCheck{属于哪个层级？}

    %% 真理层
    LayerCheck -->|公理体系内可证明<br>域内绝对正确| TruthLayer[真理层<br>Truth Layer]
    TruthLayer --> TruthCrit[判定标准<br>• 给定公理体系内严格可证<br>• 边界显式定义 D<br>• ∀x∈D: P(x) 必然真<br>• 反例只能说明“越界”]
    TruthLayer --> TruthExample[✅ 典型示例<br>1+1=2, F=ma (宏观低速),<br>能量守恒 (封闭系统)]
    TruthLayer --> TruthAnti[❌ 典型反例<br>• “水在100°C沸腾”<br>  （属模型层/经验规律）<br>• “所有天鹅白色”<br>  （归纳概括，无公理证明）]
    TruthLayer --> TruthOut[输出: 作为科学基石<br>授予“真理主权”]

    %% 模型层
    LayerCheck -->|数学/因果架构<br>逼近真理层| ModelLayer[模型层<br>Model Layer]
    ModelLayer --> ModelCrit[判定标准<br>• 明确的因果/数理机制<br>• 对真理层硬度的拟合度可量化<br>• 有显式边界条件<br>• 可被方法层检验]
    ModelLayer --> ModelExample[✅ 典型示例<br>牛顿力学（宏观低速域）<br>开普勒椭圆轨道<br>气象数值模型]
    ModelLayer --> ModelAnti[❌ 典型反例（层级僭越）<br>• 心理学“启动效应”<br>  （定性箭头，无数学硬度）<br>• 经济学过拟合回归模型<br>  （样本内R²高，样本外失效）<br>• 营养学观察性研究直接声称因果<br>  （统计关联→因果模型）]
    ModelLayer --> ModelOut[输出: 科学理论体系<br>接受模型审查与退化清退]

    %% 方法层
    LayerCheck -->|实验/统计/观测工具| MethodLayer[方法层<br>Method Layer]
    MethodLayer --> MethodCrit[判定标准<br>• 可重复、可靠<br>• 与上层模型/真理保持一致<br>• 不能单独定义科学<br>• 永远背负自证清白的举证责任]
    MethodLayer --> MethodExample[✅ 典型示例<br>双盲实验, P值计算,<br>证伪检验, 问卷量表,<br>回归分析, 交叉验证]
    MethodLayer --> MethodAnti[❌ 典型反例（僭越）<br>• “P<0.05 所以理论正确”<br>• “可证伪 所以是科学”<br>• “样本内拟合优度高 所以模型真理”<br>• “观察性研究HR显著 所以因果成立”]
    MethodLayer --> MethodOut[输出: 探测器信息<br>用于修正模型/精化边界]

    %% 拒收分支：不属于任何层或声称方法层僭越
    LayerCheck -->|无法归入三层<br>或宣称方法层可以<br>否定真理层| Reject[🚫 伪科学 / 非科学<br>拒收]
    Reject --> RejectReason[典型情况<br>• 怎么都行的相对主义<br>• 宣称“真理可被证伪推翻”<br>• 纯粹数据挖掘无因果模型<br>• 波普尔式证伪主义滥用]

    %% 自上而下的约束（红箭头）
    TruthLayer -.->|主权干预 / 硬约束| ModelLayer
    ModelLayer -.->|提供检验目标| MethodLayer

    %% 自下而上的反馈（绿箭头）
    MethodLayer -.->|反馈：方法结果| ModelLayer
    ModelLayer -.->|反馈：模型拟合度| TruthLayer

    %% 整体闭环说明
    subgraph Legend [图例]
        direction LR
        A[🔴 红色虚线: 自上而下约束] 
        B[🟢 绿色虚线: 自下而上反馈]
        C[✅ 示例]
        D[❌ 反例/僭越]
    end

    %% 样式定义
    classDef truth fill:#f9f,stroke:#333,stroke-width:2px;
    classDef model fill:#bbf,stroke:#333;
    classDef method fill:#bfb,stroke:#333;
    classDef reject fill:#fbb,stroke:#c00,stroke-width:2px;
    class TruthLayer,TruthCrit,TruthExample,TruthAnti,TruthOut truth;
    class ModelLayer,ModelCrit,ModelExample,ModelAnti,ModelOut model;
    class MethodLayer,MethodCrit,MethodExample,MethodAnti,MethodOut method;
    class Reject,RejectReason reject;

---

流程图节点说明（便于学术引用）

节点	内容	备注
真理层	公理体系内可证明，域内绝对正确	拥有“一票否决权”，反例只能越界解释
模型层	数学/因果架构，对真理层硬度的拟合	不允许定性箭头或无边界声明；过拟合、启动效应等为反例
方法层	实验、统计、证伪等操作工具	P值、可证伪性等被剥夺定义权；观察性研究HR不能僭越因果
伪科学拒收	无法归入三层，或声称方法层否定真理层	波普尔式证伪主义、相对主义、纯粹数据挖掘
约束关系	真理层 → 模型层 → 方法层（红虚线）	上层逻辑限制下层
反馈关系	方法层 → 模型层 → 真理层（绿虚线）	下层提供信息用于精化上层

---

使用建议

1. 学术论文/PPT：直接复制 Mermaid 代码块，在支持 Mermaid 的渲染器中生成矢量图，可导出为 PNG/SVG。

2. 讲解顺序：从顶部“科学主张”开始，沿三个分支讲解判定标准与反例，最后强调“拒收”门与双向箭头所代表的 TMM 运作机制。

3. 配合案例：在展示“典型反例”时，可对照前文心理学重复性危机、经济学过拟合、医学营养学反转三个案例进行具体说明。

此流程图将 TMM 从抽象理论转化为可操作、可传播的视觉工具，是“真理主权宣言”的工程蓝图。

---

---

TMM Visual Flowchart: Three-Layer Structure, Judgment Criteria, and Typical Counterexamples

Abstract

This flowchart centers on TMM (Truth–Model–Method Framework) and establishes a hierarchical judgment and constraint-feedback mechanism for scientific claims. The Truth Layer holds axiomatic sovereignty and one-vote veto power; the Model Layer must possess mathematical hardness and explicit boundary conditions; the Method Layer serves only as a testing tool, with usurpation strictly prohibited. The diagram provides clear examples and counterexamples, and displays top-down hard constraints (red arrows) and bottom-up information feedback (green arrows). Any claim that cannot be classified into the three layers, or asserts that the Method Layer may negate the Truth Layer, is rejected as pseudoscience. This diagram serves as a visual engineering blueprint for translating TMM from theory into practice.

TMM Visual Flowchart: Three-Layer Structure, Judgment Criteria, and Typical Counterexamples

The complete Mermaid flowchart below presents the logical architecture of TMM (Truth–Model–Method Framework), including:

Copy the code directly into a Mermaid-supported editor (such as Typora, GitHub Markdown, Notion, etc.) for rendering.

Code

flowchart TD
    Start([Scientific Claim / Research Practice]) --> LayerCheck{Which layer does it belong to?}

    %% Truth Layer
    LayerCheck -->|Provable within an axiomatic system<br>Absolutely correct within its domain| TruthLayer[Truth Layer]
    TruthLayer --> TruthCrit[Judgment Criteria<br>• Strictly provable within a given axiomatic system<br>• Explicitly defined domain D<br>• ∀x∈D: P(x) is necessarily true<br>• Counterexamples only indicate “transgression of boundaries”]
    TruthLayer --> TruthExample[✅ Typical Examples<br>1+1=2, F=ma (macroscopic & low-speed),<br>Conservation of energy (closed system)]
    TruthLayer --> TruthAnti[❌ Typical Counterexamples<br>• “Water boils at 100°C”<br>  (belongs to Model Layer / empirical regularity)<br>• “All swans are white”<br>  (inductive generalization without axiomatic proof)]
    TruthLayer --> TruthOut[Output: Scientific cornerstone<br>Granted “Truth Sovereignty”]

    %% Model Layer
    LayerCheck -->|Mathematical / causal framework<br>Approximating the Truth Layer| ModelLayer[Model Layer]
    ModelLayer --> ModelCrit[Judgment Criteria<br>• Explicit causal / mathematical mechanism<br>• Quantifiable fit to the hardness of the Truth Layer<br>• Explicit boundary conditions<br>• Testable by the Method Layer]
    ModelLayer --> ModelExample[✅ Typical Examples<br>Newtonian mechanics (macroscopic low-speed domain)<br>Kepler’s elliptical orbits<br>Numerical weather models]
    ModelLayer --> ModelAnti[❌ Typical Counterexamples (Hierarchical Usurpation)<br>• Psychological “priming effects”<br>  (qualitative arrows, no mathematical hardness)<br>• Overfitted regression models in economics<br>  (high in-sample R², out-of-sample failure)<br>• Direct causal claims from observational nutrition studies<br>  (statistical correlation → causal model)]
    ModelLayer --> ModelOut[Output: Scientific theoretical system<br>Subject to model review and degenerate elimination]

    %% Method Layer
    LayerCheck -->|Experimental / statistical / observational tools| MethodLayer[Method Layer]
    MethodLayer --> MethodCrit[Judgment Criteria<br>• Reproducible and reliable<br>• Consistent with upper-layer models / truth<br>• Cannot define science on its own<br>• Always bears the burden of proving its validity]
    MethodLayer --> MethodExample[✅ Typical Examples<br>Double-blind experiments, p-value calculation,<br>Falsification tests, questionnaire scales,<br>Regression analysis, cross-validation]
    MethodLayer --> MethodAnti[❌ Typical Counterexamples (Usurpation)<br>• “p<0.05 therefore the theory is true”<br>• “Falsifiable therefore it is science”<br>• “High in-sample goodness of fit therefore model truth”<br>• “Significant HR in observational studies therefore causation”]
    MethodLayer --> MethodOut[Output: Detector information<br>For model revision / boundary refinement]

    %% Rejection branch
    LayerCheck -->|Cannot be classified into three layers<br>or claims that the Method Layer may<br>negate the Truth Layer| Reject[🚫 Pseudoscience / Non-science<br>Rejected]
    Reject --> RejectReason[Typical Cases<br>• “Anything goes” relativism<br>• Claims “truth can be overthrown by falsification”<br>• Pure data mining without causal models<br>• Abuse of Popperian falsificationism]

    %% Top-down constraints (red dashed)
    TruthLayer -.->|Sovereign intervention / Hard constraint| ModelLayer
    ModelLayer -.->|Provides testing objectives| MethodLayer

    %% Bottom-up feedback (green dashed)
    MethodLayer -.->|Feedback: Method results| ModelLayer
    ModelLayer -.->|Feedback: Model fit| TruthLayer

    %% Legend
    subgraph Legend [Legend]
        direction LR
        A[🔴 Red dashed: Top-down constraints] 
        B[🟢 Green dashed: Bottom-up feedback]
        C[✅ Example]
        D[❌ Counterexample / Usurpation]
    end

    %% Style definitions
    classDef truth fill:#f9f,stroke:#333,stroke-width:2px;
    classDef model fill:#bbf,stroke:#333;
    classDef method fill:#bfb,stroke:#333;
    classDef reject fill:#fbb,stroke:#c00,stroke-width:2px;
    class TruthLayer,TruthCrit,TruthExample,TruthAnti,TruthOut truth;
    class ModelLayer,ModelCrit,ModelExample,ModelAnti,ModelOut model;
    class MethodLayer,MethodCrit,MethodExample,MethodAnti,MethodOut method;
    class Reject,RejectReason reject;

Explanation of Flowchart Nodes (for academic citation)

表格

Node	Content	Notes
Truth Layer	Provable within an axiomatic system, absolutely correct within its domain	Holds “one-vote veto power”; counterexamples only interpreted as transgression
Model Layer	Mathematical / causal framework, fit to the hardness of the Truth Layer	Qualitative arrows or undefined boundaries prohibited; overfitting, priming effects, etc., are counterexamples
Method Layer	Operational tools including experiment, statistics, falsification	p-values, falsifiability, etc., stripped of definitional power; observational HR cannot usurp causation
Pseudoscience Rejection	Cannot be classified into three layers, or claims Method Layer negates Truth Layer	Popperian falsificationism, relativism, pure data mining
Constraint Relation	Truth Layer → Model Layer → Method Layer (red dashed)	Upper-layer logic restricts lower layers
Feedback Relation	Method Layer → Model Layer → Truth Layer (green dashed)	Lower layers provide information for upper-layer refinement

Usage Recommendations

This flowchart transforms TMM from abstract theory into an operable and communicable visual tool, serving as the engineering blueprint for the “Declaration of Truth Sovereignty”.