未知文明已知理论：未来知识的结构可在性与文明现实化——一种关于发现与文明的计算认识论

中文摘要

一个尚未被发现、但未来可能成为科学知识的理论，在被发现以前究竟处于什么状态？传统认识论多讨论已经形成的信念、证据、真理与知识，却很少追问：未来知识在进入现实认识以前，与既有的对象、概念、符号和规律之间，是否已经具有某种结构上的关系？本文提出”未知文明已知理论”。它的核心主张不是未来知识已经以现实知识的形式存在，也不是未来已经被决定，而是提出一种”结构可在性”：对于某项未来可以被有限表征系统充分表达的知识，如果形成它所需要的对象、状态、关系、过程和基本规律已经进入某一认知系统的表征范围，那么它在被现实发现以前，就已经作为可构造、可组合或可推导的结构，位于该系统的表征可能空间之中。这一理论的认识论根基在于”反映论”——思维的本质是生命物质对世界的”反映”（感官感觉在大脑中的内化），文字则是思维的最高级物化形式（参见《思维论》第一章、第七章）。符号组合的深层基础不是字符本身，而是符号所指向的思维素材及其关系。基于这一根基，本文提出四个核心命题。第一，符号不是思想本身，而是思维素材的公共表征地址。第二，某些未来知识能够先于显式认识而获得结构可在性。第三，在给定表征基础与推理规则的条件下，可以合法推出的后果属于相应的条件性表征闭包，但”隐含可导”不等于主体已经显式知道。第四，一个结构只有进一步经过真实性、文明相关性、可验证性与可转化性的筛选，才可能进入现实文明。由此，本文区分”结构可在”“认识现实化”与”文明现实化”三个层次，并区分”结构性未知”与”搜索性未知”。本文以费马大定理的证明史和孟德尔遗传定律的发现史为思想实验，说明一个后来完成的知识结构，如何随着概念、工具和关系桥梁的形成而逐渐进入可搜索空间；又通过博尔赫斯、博登、波普尔、西蒙及计算认识论等既有思想传统，厘清本理论的学术位置。同时，本文主动回应反实在论与社会建构论的潜在反驳，进一步划定理论的适用范围。哥德尔不完备性、图灵不可判定性和物理计算极限则表明，本理论不能被理解为”存在一个可以穷尽全部未来真理的知识仓库”。它更严格的含义是：未知不必等于结构上的不存在；一部分未知可能已经结构可在，却尚未被计算、识别、验证和现实化。当机器大幅降低候选假设的生成成本以后，科学认识的瓶颈可能逐渐从”如何产生一个候选”转向”如何判断哪个候选值得相信和检验”。本文将这一问题称为”认识判别器问题”。人工智能由此不是本理论的理论前提，而是显著扩大文明探索可能知识空间”搜索带宽”的新技术条件。本文最终提出：文明既不是从绝对的”无”中持续创造一切，也不是从一部预先写好的未来剧本中逐页读取结果，而是一个不断扩展思维素材与表征边界、形成新的可能空间、搜索其中的候选结构，并通过世界与人的双重筛选，将极少数结构转化为现实知识与实践的历史过程

Abstract

What state is an as-yet-undiscovered theory in, before it is discovered, when it may become scientific knowledge in the future? Traditional epistemology mostly discusses already-formed beliefs, evidence, truth, and knowledge, but rarely asks: before future knowledge enters actual cognition, does it already stand in some structural relation to existing objects, concepts, symbols, and laws? This paper proposes "the theory of unknown civilization and known theory." Its core claim is not that future knowledge already exists in the form of actual knowledge, nor that the future is already determined, but rather a thesis of "structural availability": for a piece of future knowledge sufficiently expressible by a finite representational system, if the objects, states, relations, processes, and basic laws required to form it have already entered the representational scope of some cognitive system, then before its actual discovery it already exists, as a constructible, combinable, or derivable structure, within the representational possibility space of that system. The epistemological root of this theory lies in the "reflection theory": thinking is the "reflection" of living matter upon the world (the internalization of sensory feeling in the brain), and writing is the highest materialized form of thinking (see chapter 1 and chapter 7 of "On Thinking"). The deep basis of symbol combination is not the characters themselves, but the thinking materials and their relations to which symbols point. On this foundation, the paper proposes four core propositions. First, symbols are not thought itself, but public representational addresses of thinking materials. Second, some future knowledge can obtain structural availability prior to explicit cognition. Third, under given representational bases and inference rules, legally derivable consequences belong to the corresponding conditional representational closure, but "implicitly derivable" does not mean the subject already explicitly knows. Fourth, a structure can enter actual civilization only after further screening by truthfulness, civilizational relevance, verifiability, and transformability. The paper accordingly distinguishes three levels — "structural availability," "cognitive actualization," and "civilizational actualization" — and distinguishes "structural unknowns" from "search unknowns." Using the proof history of Fermat's Last Theorem and the discovery history of Mendel's laws of inheritance as thought experiments, it shows how a later-completed knowledge structure gradually enters the searchable space as concepts, tools, and relational bridges form; and through existing intellectual traditions including Borges, Boden, Popper, Simon, and computational epistemology, it clarifies the theory's academic position. The paper also proactively responds to potential objections from anti-realism and social constructivism, further delimiting the scope of application of the theory. Godel's incompleteness, Turing undecidability, and physical computational limits show that this theory cannot be understood as "a knowledge warehouse that exhausts all future truths." Its stricter meaning is: the unknown need not equal structural non-existence; part of the unknown may already be structurally available, yet not yet computed, recognized, verified, and actualized. When machines drastically lower the cost of generating candidate hypotheses, the bottleneck of scientific cognition may gradually shift from "how to produce a candidate" to "how to judge which candidate is worth believing and testing." The paper calls this the "epistemic discriminator problem." Artificial intelligence is thus not a theoretical premise of this theory, but a new technological condition that significantly expands the "search bandwidth" of the knowledge space civilization can explore. The paper ultimately proposes: civilization is neither continuously creating everything from absolute "nothing," nor reading results page by page from a pre-written future script, but a historical process of continuously expanding thinking materials and representational boundaries, forming new possibility spaces, searching candidate structures within them, and, through the double screening of the world and of human beings, converting a very small number of structures into actual knowledge and practice.

关键词：未知文明已知理论；结构可在性；表征可能空间；未来知识；假设空间

Keywords: Theory of unknown civilization and known theory; structural availability; representational possibility space; future knowledge; hypothesis space