Xiangbin Zhao

Can the fundamental directions of theoretical proof be classified into a complete and non-overlapping system? This paper answers this question for formal theoretical tasks that satisfy explicit admissibility conditions. Using direct theoretical output as the sole classification axis, it develops a decision tree that classifies every admissible task into one of five fundamental output types: structural construction, formal representation, necessary consequence, existential demonstration, and inverse identification. These correspond respectively to the General Form, Mathematical Form, Necessity Form, Possibility Form, and Inverse Form established in On Self-Manifestation. The paper further proves that each formal work has one and only one primary proof direction under single-task normalization, establishes the conditions for strict completeness, specifies explicit falsification criteria, and provides a cross-disciplinary validation protocol. The result upgrades the empirical completeness proposed in Canon Chapter 1 to strict completeness within its stated domain.

Why can the same research object require different proof forms, while different research objects may share the same proof structure? This paper proves that the determining factor is not the research object but the theoretical proposition to be established. Building upon the taxonomy introduced in Canon Chapter 1, three theorems—the Object Insufficiency Theorem, the Proposition Sufficiency Theorem, and the Decision Theorem—demonstrate that the choice of theoretical form, together with its completion criteria and falsification conditions, is determined by the proposition and its proof objective rather than by the object under study. The result provides the theoretical foundation for the five primary forms of On Self-Manifestation and shifts theoretical methodology from an object-centered paradigm to a task-centered paradigm for constructing, organizing, and reviewing large formal theories.

Different theoretical tasks require different forms of proof, yet existing proof theory, mathematical logic, scientific methodology, and formal modeling primarily focus on logical validity, theorem proving, or empirical support rather than on the classification of theoretical forms themselves. This paper proposes a taxonomy of theoretical forms based on the proposition to be established instead of the research object. It introduces five primary forms, fourteen auxiliary forms, comparative meta-forms, governing forms, and formal components, together with unified specifications for their purposes, applicability, organization, completion criteria, falsification, citation, and peer review. Rather than presenting a new logical calculus or axiomatic system, this work establishes a formal specification for constructing, organizing, reviewing, and maintaining large theoretical systems. It provides the common standard for all formal works in On Self-Manifestation while offering a general framework for organizing complex theoretical research.

This paper offers a structural decomposition of the black box of GPT and large language models. It does not explain GPT’s capability by a single parameter, a single attention head, a single feature, or simple next-token prediction. Nor does it treat hallucination, drift, and false attribution as merely a lack of knowledge. Instead, the paper explains how large language models form their capability boundaries through external training, textual sedimentation, path compression, prompt-driven activation, and generative continuation, and how the same structure produces both their strengths and their characteristic failures. The paper argues that a large language model is not a database of stored knowledge entries, but a language sedimentation system shaped by large-scale text training. The training objective supplies direction, loss marks difference, gradient and optimizer adjust paths, RLHF or DPO shapes preference, prompt provides the current task, and system prompt sets the boundary. Large-scale text training repeatedly compresses paths from previous text to later text, question to answer, context to response, format to completion, and task to fulfillment. These repeated paths form deeply familiar generative channels. GPT is powerful because these channels have been deeply sedimented: it can write, translate, program, imitate styles, continue tasks, and transfer across languages and task forms. The same structure also explains its failures. GPT can generate fact-like forms, but it cannot stably distinguish external facts, contextual materials, internal inference, and self-generated completion. It can maintain textual continuity, but this is not the same as stable current state. It can use tools, but this is not the same as forming an internal action loop. It can adjust output according to prompts, but this is not the same as having its own intention. In this framework, hallucination is not an accidental error, but a structural consequence of generation being stronger than source tracking. Drift is not merely a lack of context length, but the result of textual continuity not being equivalent to stable current state. False attribution is not merely a failure of factual retrieval, but a consequence of lacking reliable internal source stratification. The significance of this paper is that it opens the black box of GPT and explains it as a generative system shaped by external training, supported by sedimented language paths, activated by prompts, and not yet internally closed. The framework yields testable implications: interventions on FFN/MLP, cross-layer residual paths, attention boundaries, source tracking training, current state maintenance, and online learning mechanisms should affect capability, hallucination, drift, and task stability in different ways. Therefore, the key to next-generation AI is not only larger models, longer context, and more tools, but the gradual formation of stronger source tracking, current state maintenance, feedback loops, and self-updating capacity.

本文提出对 GPT 和大语言模型黑箱的结构拆解。本文不把 GPT 的能力解释为单个参数、单个 attention head、单个 feature 或简单的下一词概率预测，也不把幻觉、漂移和错归因仅仅归为知识不足。相反，本文试图说明大语言模型如何在外部训练、文本沉积、路径压缩、提示牵引和输出续行中形成其能力边界，并由此产生其特有的强项与缺陷。 本文认为，大语言模型的核心不是存放知识条目的数据库，而是由大规模文本训练塑成的语言沉积系统。训练目标提供方向，loss 标定差异，gradient 和 optimizer 调整路径，RLHF 或 DPO 塑造偏好，prompt 给出当前任务，system prompt 设定边界。大量文本训练反复压入从前文到后文、从问题到回答、从语境到回应、从格式到补全、从任务到完成的路径，使模型形成高度熟化的生成通道。因此，GPT 的强项来自这些通道的深度沉积: 它能写作、翻译、编程、模仿风格、接续任务，并在不同语言和任务形式之间迁移。 但同一结构也解释了它的缺陷。GPT 能生成事实样式，却不能稳定区分外部事实、上下文材料、内部推断和自行补全；能维持文本连续，却不等于拥有稳定当前状态；能调用工具，却不等于形成内在行动回路；能按提示调整输出，却不等于具有自身意图。由此，幻觉不是偶然错误，而是生成能力强于来源追踪时的结构后果；漂移不是单纯上下文长度不足，而是文本连续并不等于当前状态稳定；错归因也不是个别事实检索失败，而是模型内部缺少可靠来源分层的结果。 本文的意义在于，它把 GPT 从神秘黑箱中拆开，解释为一个由外部训练塑成、由语言路径沉积支撑、由提示牵引启动、但尚未完成内部闭合的生成系统。这个框架可产生可检验推论: 改变 FFN/MLP、cross-layer residual paths、attention boundaries、source tracking training、current state maintenance 和 online learning mechanisms，应当以不同方式影响模型能力、幻觉、漂移和任务稳定性。由此，下一代 AI 的关键不只是更大模型、更长上下文和更多工具，而是让模型逐步形成更可靠的来源追踪、当前状态保持、反馈回路和自我更新能力。

This paper proposes a general form of the Observation-Domain, explaining why not all information entering a system can become current experience. The Observation-Domain is not the physical range of the receptors, nor is it a synonym for attention. It is a structural boundary jointly produced by candidacy, admission, and convergence reachability. A content may be physically close to the receptors and highly intense, yet still fail to manifest. Whether it can manifest depends on whether it can be unfolded as a candidate, admitted into finite competition, and pushed under current resources and parameters into a maintainable state of occupancy. The paper divides the Observation-Domain into three gates and three domains. The three gates are the gate of candidacy, the gate of admission, and the gate of convergence reachability. The first decides whether a structure can be unfolded as a candidate. The second decides whether it can enter finite competition. The third decides whether, under current system parameters, it can be pushed by resources into manifestation. The first two gates define the eligibility domain E(t), namely contents that already have eligibility for competition. The conjunction of all three gates defines the manifestable domain W(t), namely contents that can in principle be lit up under current conditions. Their difference, M(t), is the boundary-fringe: content that has eligibility but is not reachable. Boundary-fringe content is not completely ineffective. It may be processed by the system, produce bias, or influence behavior, but it cannot enter current manifestation while the parameters remain unchanged. The paper further shows that the Observation-Domain varies with capacity, channel distribution, familiar routes, and orientation. Increased load contracts the Observation-Domain and biases it toward highly familiar content. Anesthesia or declining convergence capacity first contracts and then empties the Observation-Domain. Training and tools may expand the boundary without increasing the number of observers. Orientation may reshape the boundary in real time. This framework can explain hemispatial neglect, blindsight, peripheral loss under dual-task load, and the temporary recovery produced by cueing or motivational intervention. The core conclusion is: the Observer solves “who occupies the current moment,” while the Observation-Domain solves “what can be lit up.” Without such a boundary, manifestation would require infinite capacity, which violates the basic premise of a finite system.

本文提出观界的一般形式，用以说明为什么并非所有进入系统的信息都能成为当前经验。观界不是感受器的物理范围，也不是注意力的同义词，而是由候选资格、进入限制和收敛可达性共同压出的结构边界。一个内容即使物理上接近感受器、刺激强度很高，也未必能够显现；反之，一个内容能否显现，取决于它能否被展开为候选，能否通过有限竞争的进入限制，并且在当前资源和参数下能否被推至可维持的占据状态。 本文将观界划分为三个门和三个域。三个门是候选门、准入门和收敛可达门。候选门决定某个结构能否被展开为候选；准入门决定它能否进入有限竞争；收敛可达门决定它在当前系统参数下能否被资源推至显现。由前两门得到资格域 E(t)，即已经具有竞争资格的内容；三门交会得到可显域 W(t)，即当前条件下原则上能够被点亮的内容；二者之差 M(t) 是界缘，即有资格但不可达的内容。界缘内容并非完全无效，它可以被系统处理、产生偏置或影响行为，但在参数不变时不能真正进入当前显现。 本文进一步说明，观界会随容量、通道分布、熟悉路径和取向而改变。负荷升高会使观界收缩，并偏向高熟悉度内容；麻醉或收敛能力下降会使观界先收缩后清空；训练和工具可以扩展观界而不增加观察者数量；取向改变可以实时重塑观界。由此，本文可解释半侧空间忽略、盲视、双任务负荷下的外周信息丢失、注意线索的暂时恢复作用等现象。本文的核心结论是: 观察者解决“谁在此刻占据”，观界解决“什么能够被点亮”。没有观界，显现就会要求无限容量，而无限容量违背有限系统的基本前提。

This paper introduces the concept of the Current World, explaining how an observer is lodged within a world that can be experienced. The Current World is not merely the external physical world, nor is it merely an internal representation. It is the world-structure in which a system can cut into sound, light, taste, touch, and other forms of information, maintain them as sensible and actionable, and return them to reality. Only when an I is brought forth within the current and gains its reality can one speak of the world of the current. The paper argues that to understand a world, one must not only ask what objects are, but also how a system cuts into information, what is cut, what kind of trace-response is formed, where objectification takes place, and in what form reality is generated. In this sense, the Current World becomes the target of field-construction and world-construction. An experiential world requires at least three coordinated structures: a law-field that provides stable continuation, a self-manifestation field that provides manifestness and subjectivity, and a world-field that provides lodging and support. The law-field gives operation repeatable paths, the self-manifestation field gives the system the capacity to manifest, and the world-field gives the self-manifesting being a world in which it can reside and act. The significance of this framework is that it moves the problem of consciousness, perception, world modeling, and artificial consciousness beyond pure internal representation, and toward the embedded relation between observer and world. For neuroscience and cognitive science, the Current World can help explain how perceptual channels, the sense of reality, body, and environment jointly support experience. For artificial intelligence, it suggests that artificial consciousness cannot rely only on language models or symbolic space, but requires a world-structure that can be cut into, maintained, and fed back. For physics and complex systems research, it interprets world as a continuable domain formed through information cutting, stable regularity, and supporting structure.

本文提出当前世的概念，用以说明一个观察者如何被安置在可经验的世界中。所谓当前世，并不是单纯的外部物理世界，也不是主观内部表象，而是一个系统能够截取声、光、味、触等信息，并将其维持为可感、可行、可返回现实的世界结构。只有当我在当前中被带出，并获得其现实性时，当前之世才成为可论的世界。 本文指出，理解一个世界，不能只问对象是什么，还要问系统如何截取、截取了什么、迹应具有什么性质、物化发生在何处、现实以何种形式成形。由此，当前世成为造场和造世界的目标。一个可经验世界至少需要三类结构协同: 律场提供可续行的规律，自显场提供显性和主体性，世场提供安置和承载。律场使运行具有可重复的路径，自显场使系统形成显现能力，世场则使自显体获得可以居住和行动的世界。 这一框架的意义在于，它把意识、感知、世界模型和人工意识的问题从纯粹内部表征中移出，转向观察者与世界之间的嵌入关系。对于脑科学和认知科学而言，当前世可用于解释感知通道、现实感、身体和环境如何共同支撑经验。对于人工智能研究而言，它指出人工意识不能只依赖语言模型或符号空间，还需要一个可截取、可维持、可反馈的世界结构。对于物理学和复杂系统研究而言，它把世界理解为系统通过信息截取、稳定规律和承载结构共同形成的可续存域。

How can a system’s history become part of what generates its subsequent behavior? Rather than deriving its definition from an existing neural-network architecture, this paper explicitly constructs a minimal runnable sedimentation field whose later dynamics can be altered by its own history. External action must enter the system and produce a trace-response; that response must persist across time and contribute to relational organization; and the resulting historical structure must genuinely participate in subsequent generation. If a load-bearing sedimentary structure is removed or severed while present re-entry conditions are held fixed, later flow must change detectably. The paper gives an explicit Canonical F* together with four core operators—Trace, Sediment, Decay, and Continuation—and closes the required offices of entry, persistence, relational organization, historical participation, decay, and present flow.

Once a law has acquired formal qualification, does that qualification survive as the system continues to change? If its original bearer ceases to qualify, can another organization continue to bear it? And can a qualification that has been lost later be recovered through the system’s own dynamics? This paper separates two levels of stability: whether the same bearing organization remains a formal solution, and whether the formal solution set remains non-empty when the bearer may change. It thereby develops a framework for fixed-solution stability, solution-set stability, destabilization, substitution, and recovery. What is preserved is not a particular path, parameter configuration, or realization, but the formal qualification defined upstream. The resulting distinction provides a framework that can be further instantiated in learning and cognitive systems with redundancy, structural substitution, or continuing adaptation.

If laws can arise through historical formation, a further question follows: what bears them, whether that organization genuinely enters subsequent generation, and what supports stable recurrence. This paper maps the structural account developed in The General Form of Law into a formal system. It separates formed structural bearing, dynamical participation, and re-entry bearing, and introduces a formal solution set, evaluation contexts, local generation kernels, solving procedures, and rules for reading formal results back to the structural level. The framework explicitly distinguishes genuine dynamical contribution from pure correlation, hot redundancy, structural standby, structural pre-emption, temporal pre-emption, and multi-level participation. Its purpose is to make precise three different questions that are often conflated: what has already formed, what actually contributes to what happens next, and what can stably recur.

What makes a regularity a law? A single matching relation, a present response, a retained trace, or even a structure that temporarily affects subsequent behavior is not by itself sufficient for lawhood. This paper approaches the question generatively and asks what must minimally be present once a law has come to stand in a sedimentation-based system. Under the conditions explicitly stated in the paper, lawhood requires at least one and the same bearing organization that is already formed in the present structure, genuinely participates in generating what follows, and is supported for stable recurrence. The result reframes law not merely as an external description of repeated events, but as a problem of formation, present bearing, and subsequent dynamical organization. “Law” is used here in its ordinary sense of a regularity, law of nature, or rule, rather than as a special object introduced by the theory.

This paper derives a minimal canonical model for the Non-Social Terminal State (NST) within a sedimentation field dynamics framework. Prior work in this framework has established that social fields are cross-field shared-core structures formed through long-term coupling of individual fields, and that social dependency originates from individual capacity gaps. This paper further asks: what sustains social dependency, whether that sustaining structure can hold permanently, and if not, how NST is structurally extracted from the general formula. Defining social dependency through the functional capacity gap CapacityGap_j := max(0, GroupSupport_j − IndividualCapacity_j) as a necessary and sufficient condition, the paper extracts three minimal sufficient conditions through an eight-step interrogation chain. First, a finite need bound (NeedBound_j < ∞): individual functional needs are finite, implying a finite upper bound on group support. Second, a sufficient external extension path (PathSufficient_j): an extension path exists that is not capped by a single field’s internal capacity, and whose ability ceiling exceeds the need bound. Third, a non-permanently-zeroed toolization integral (T_j := ∫₀^∞ ToolizationRate_j dt = ∞): group functions are continuously toolized and absorbed into individual effective capacity, and cannot be permanently blocked on an infinite time horizon. When all three conditions hold jointly, the paper formally proves: lim(t→∞) CapacityGap_j(f,t) = 0 (∀j) ⟹ lim(t→∞) SocialDependency(f,t) = 0 ⟹ NST(f). The paper also proves the minimal sufficiency of the three conditions — NST fails if any one is absent — and provides three counterexamples and two domain boundary cases to complete the penetration argument. Five supplementary formulas are provided. Appendix 1 gives the complete formal definitions of the non-social expression set, including capacity gap, social dependency, non-social condition, and effective capacity decomposition. Appendix 2 gives the non-social limit formula, including time domain setup, limit sufficiency conditions, and limit rate bounds. Appendix 3 gives the non-social dissolution formula, including the three dissolution operators, dissolution dynamics equation, dissolution sufficiency conditions, and dissolution irreversibility conditions. Appendix 4 gives the non-social phase diagram, constructing a four-quadrant phase space with functional dependency (FD) and psychological dependency (PD) as axes, tracing the evolutionary trajectory from full social state (S1) through toolized social state (S2) and functional non-social state (S3) to full non-social state (S4). Appendix 5 gives the boundary absorption formula, defining the absorption-type non-social terminal state (NST_A) in which the individual field is absorbed into a total field and the capacity gap becomes undefined, and argues that the coloring circuit remains fully operative under NST_A — emotional experience remains complete, with only the external-source dependency of coloring imagery dissolved. The paper concludes by defining the complete set of non-social terminal states: NonSocialUltimate(f) ⟺ NST(f) ∨ NST_A(f), with the two pathways independent in structure and mechanism, both sufficient to reach non-social terminal states.

本文在沉积场动力学框架下给出非社终态（NST）的最小范式模型。已有工作证明社会场由多个个体场长期耦合后形成的跨场共核结构构成，社群依赖来源于个体功能缺口；本文进一步追问：社群依赖靠什么维持，这个维持结构是否能永久成立，若不能，非社终态如何从一般式中被逼出。 以功能缺口CapacityGap_j := max(0, GroupSupport_j − IndividualCapacity_j)为社群依赖的充要条件，通过八步追问链逐步逼出三个最小充分条件：其一，有限需求上界（NeedBound_j < ∞）——个体实际功能需求有限，蕴含群体支持存在有限上界；其二，外接能力路径存在且充分（PathSufficient_j）——存在不受单一场容积封顶的外接扩展路径，且路径能力上界超过需求上界；其三，工具化进程长期累积有效量不归零（T_j := ∫₀^∞ ToolizationRate_j dt = ∞）——群体功能持续被工具化并归入个体有效能力，在无限时间尺度上不被永久阻断。三条件联立给出正式证明：lim(t→∞) CapacityGap_j(f,t) = 0（∀j）⟹ lim(t→∞) SocialDependency(f,t) = 0 ⟹ NST(f)。本文同时证明三条件的极小充分性——缺少任意一条则NST不成立——并给出三组反例与两个定义域边界，完成击穿论证。 本文附件覆盖五个方向：附件一给出非社表达式的完整形式定义，包括功能缺口、社群依赖、非社条件与有效能力分解式；附件二给出非社极限式，包括时间域设定、极限成立条件与极限速率界；附件三给出非社消解式，包括消解路径三算子、消解动力方程、消解充分条件与消解不可逆条件；附件四给出非社相图式，以功能依赖度（FD）与情感依赖度（PD）为两轴建立四象限相图，描述从完全社会态（S1）经工具化社会态（S2）、功能非社态（S3）至完全非社态（S4）的演化轨迹；附件五给出边界吸收式，定义当个体场被总场吸收、功能缺口定义失效时的吸收型非社终态（NST_A），并论证着色回路在NST_A下仍完整运作，情感体验完备，消解的仅是着色意象的外部来源依赖。最终定义非社终态完备集：NonSocialUltimate(f) ⟺ NST(f) ∨ NST_A(f)，两条路径独立，结构机制不同，均可达非社终态。

Group dependence does not dissolve because individuals “choose” to leave society — it is systematically eroded by four structural mechanisms: Capacity Expansion (the upper bound of individual effective capacity is not sealed by single field volume), Coupling Cost Escalation (the cost of maintaining social coupling grows superlinearly with scale), Specialization Cage Reversal (the emergence of extended capacity systematically dissolves the necessity of division of labor), and Group Function Toolization (every major function provided by groups tends, over sufficiently long time, to be toolized and absorbed into individual capacity). This chapter proves within a sedimentation field dynamics framework that the four mechanisms in combination cause functional gaps to tend to zero across all key dimensions, driving group dependence toward its limiting value of zero — the Non-Social Terminal State is not a possible state but the dynamic endpoint of the Self-Manifesting Being’s evolutionary path. The “necessity” here is a long-run dynamic tendency, not logical necessity: extreme circumstances can block this trend within finite time, but under sustainable upgrade conditions they do not constitute structural counter-arguments. Three results follow. First, the four laws are mutually independent — even if one law is counteracted in a particular historical period, the remaining three maintain the overall tendency, providing strong robustness. Second, the driving force is not that individual capacity growth rate exceeds group capacity, but that the functional gap tends to zero — a far weaker and harder-to-refute condition. Third, the Non-Social Terminal State does not equal the zeroing of social relationships; functional dependence tending to zero and emotional coupling are independent variables.

Social groups are not the essential destination of intelligent beings, but a compensatory structure for the stage of insufficient capacity. This claim is not a value judgment but a dynamic criterion: group dependence loses its motivating force if and only if the individual’s effective capacity in all key functional dimensions exceeds the functional gains groups can provide, and social groups degrade from prerequisites to options. This chapter provides a complete formalization of non-social conditions within a sedimentation field dynamics framework. Group dependence is fundamentally a compensatory structure for the gap between individual functional capacity and required functions. When this gap trends to zero across all key dimensions, group dependence can no longer be dynamically sustained. Non-social conditions must simultaneously hold across six key functional dimensions — sensory extension, action execution, construction and manufacturing, communication and information acquisition, cognition and computation, and survival and repair — and satisfy four criteria: full-dimensional transcendence, robustness of extended capacity, demand finiteness, and the functional criterion (functional non-sociality does not equal emotional non-sociality). Three results follow. First, the group dependence origin theorem: gaps sustain dependence; gap dissolution eliminates dependence. Second, effective capacity openization: the upper bound of individual capacity is not constrained by single field volume, but tends toward openness through mechanisms of Lodging-Vessel substitution, external tool connection, sensory extension, and cognitive parallelism. Third, the non-social stable state theorem: once established under maintained effective capacity, non-social conditions are conditionally stable and do not spontaneously reverse, though structural decline in effective capacity can break this stability.

History is neither a random walk nor a linear progression, but exhibits structured trajectories: certain civilizations cycle repeatedly between the same institutional forms, certain societies consistently return to similar states after external shocks, and the characteristics of dead civilizations persistently resurface in their successors. What is the common source of these regularities? This chapter provides a unified dynamic equation for social evolution within a sedimentation field dynamics framework: dCore_S/dt = f(current dynamic pressure) + historical residual bias × bias strength + disturbance term. History is not random because the historical residual bias term is nonzero; history is not deterministic because the disturbance term and current dynamics coexist. From this, four laws of social evolution are derived: the Path Dependence Law (historical starting points determine long-run trajectories), the Evolutionary Inertia Law (why core structures resist complete replacement after revolutions), the Evolutionary Convergence Law (why industrialization, bureaucratization, and marketization emerge across civilizations), and the Evolutionary Bifurcation Law (why societies split into multiple trajectories under critical pressure). Three results follow. First, the four laws correspond to four distinct observable historical patterns and are mutually consistent. Second, the laws of social evolution are the social-layer realization of field evolution laws, inheriting the dynamic framework of field theory rather than introducing independent assumptions. Third, evolutionary trajectories carry no directional purpose — the dynamic equation describes only direction and speed, not the value of the destination.

Every generation dies; civilizations can persist for millennia. What mechanism sustains continuity across the death of any single generation? Many apparently “sudden” civilizational collapses are in fact the cumulative result of intergenerational transmission breakdown: not a single external shock that directly destroys civilization, but the transmission efficiency falling below threshold across successive generations until one generation cannot find enough inheritors, and civilization enters irreversible collapse. This chapter provides a complete mechanism for civilizational intergenerational transmission within a sedimentation field dynamics framework. The intergenerational continuity of a civilization’s cross-generational co-nucleus is maintained through the co-nucleus overlap between adjacent generations (GenOverlap) — jointly determined by three factors: transmission integrity, participation rate, and institutional continuity. As long as GenOverlap between adjacent generations remains above the intergenerational survival threshold, civilization retains the capacity to continue across the death of any single generation; when GenOverlap falls below the intergenerational fracture threshold, transmission enters irreversible breakdown, triggering the intergenerational collapse pathway. Three results follow. First, civilizational continuity depends on the collective transmission behavior of the generational cohort — ritual participation, language use, educational institution maintenance — not on the memory of any single individual. Second, the decline of GenOverlap is not linearly lethal; two distinct threshold regions exist: a fragile transmission zone and an irreversible fracture zone. Third, intergenerational fracture is the micro-level realization mechanism of civilizational compression death; external shocks (forced assimilation, religious purges, large-scale displacement) destroy civilizations precisely by driving GenOverlap below the fracture threshold.

每一代人都会死亡，而文明可以延续数千年——这个跨越单代人生命周期的连续性，是通过什么机制实现的？历史上许多”突然”的文明死亡，实际上都是代际传递断裂的累积：不是某次外部冲击直接摧毁文明，而是连续数代的传承效率低于阈值，直到某一代找不到足够多的传承者，文明进入不可逆崩溃。 本文在沉积场动力学框架下给出文明代际传递的完整机制：CivilCore 的代际连续性通过相邻世代之间的共核重合度（GenOverlap）维持——GenOverlap 由传递完整性、参与率和制度连续性三项因子共同决定。只要相邻世代之间的 GenOverlap 持续高于代际存续阈值，文明就具备跨越单代人死亡而延续的条件；当 GenOverlap 跌破代际断裂阈值时，传递进入不可逆断裂，触发文明的代际崩溃通道。 由此推出三个结论：其一，文明存续依赖世代群体的集体传承行为（仪式参与、语言使用、教育制度），而非任何单个个体的记忆；其二，GenOverlap 的下降不是线性致死的，存在两个不同的阈值区间——脆弱传递区与不可逆断裂区；其三，代际断裂是文明压缩死亡的微观实现机制，外部冲击（强制同化、宗教清洗、大规模迁移）之所以能摧毁文明，正是因为它将 GenOverlap 击穿至断裂阈值以下。