
什么是科学?这个看似简单的问题,其实一直存在争论。特别是像精神分析学、进化论、相对论这样的理论,它们在时间和空间尺度上远远超出了人们容易直观想象和直接验证的范围,因此长期以来不断有人质疑,它们究竟算不算科学。
围绕“什么是科学”,哲学界曾发生过非常激烈的论战。科学哲学家卡尔·波普尔在其著作《科学发现的逻辑》(1934)中提出了 falsifiability 这一概念,后来被译成中文“可证伪性”。波普尔强调了普遍规律与基本观察陈述之间所存在的一种不对称关系,并将这一概念与当时逻辑实证主义哲学中流行、在直觉上看似相近的“可证实性”作了对比。他指出,要验证诸如“所有天鹅都是白的”这样的断言,唯一的方式是经验性地观察所有天鹅,而这在原则上是不可能的;相反,只需要观察到一只黑天鹅,就足以否定这一断言。
波普尔的观点在科学界和哲学界都受到了尊重,但并没有形成普遍共识。托马斯·库恩指出,如果从科学史的角度来看,科学实际上并不是靠“证伪”在运作的。在常规科学阶段,反例往往被吸收,异常被暂时搁置,理论并不会因为一次失败就被抛弃。库恩的研究使得“可证伪性等同于科学方法”这一说法,在经验层面难以成立。随后,伊姆雷·拉卡托斯进一步指出,波普尔在原则上是对的,但他的模型过于简化,原始形式并不能直接适用于真实的科学实践。
几乎没有当代主流哲学家认为,一个理论是否科学,可以仅仅依靠可证伪性来判断。
我很早就知道“可证伪性”这个词,不仅在科技领域,在法律领域中也时常遇到。但这个词我一次也没有主动使用过,因为从直觉上就对它非常反感。这种反感并不只是认识层面的,而是实实在在的生理反应,看到这个词时,肠胃会产生一种明显的不适感。当人们用这个词来说明什么是科学时,它在我看来完全是反逻辑、反直觉的。原因并不复杂——“伪”怎么可能被“证”呢?
后来我的阅读逐渐从以中文为主转向以英文为主,才意识到这种反感在很大程度上来自于错误的翻译。Falsifiability 的字面含义,其实更接近于“存在出错的可能性”。随着阅读的深入,我才逐渐认识到,卡尔·波普尔并没有试图给科学下一个正面的定义,他真正做的是为科学划定一条边界。
然而在中文语境中,关于“什么是科学”,却长期被简化为以“可证伪性”为标准。这其中至少存在两重错误:既误解了波普尔的本意,又采用了极其不恰当的翻译。无论怎样分析,falsifiability 这个词里,都很难找到任何与“证”有关的含义。
“证伪”这个词在中文世界里几乎已经成了一个无需解释的哲学术语,但实际上,它在语言结构、逻辑含义和认识论指向上,都存在着难以忽视的问题。“证”在汉语中始终带有确证性的意味。无论是“证实”“证据”还是“证明”,都指向同一件事——通过某种方式把一个原本存疑的判断固定下来,使之成立、坐实、不可动摇。正因为如此,“证”这个动词天然地与“真”“实”“是”相连,而不是与否定性的结果相连。
“伪”则恰恰相反,它表示虚假、不成立、并非如此。把“证”和“伪”强行拼接在一起,构成“证伪”这个词,本身就在汉语内部制造了一种紧张甚至冲突。我们在日常语言中不会说“证假”或“证不真”,因为这样的表达方式明显违背汉语直觉。一个动词如果是为了确立某种状态,那么它就不适合用来确立一种否定状态。
如果可证伪是科学的标准,那么证伪之后就已经是“伪”了,又怎么可能是科学呢?这就好像“我正在说谎”这句悖论一样。
科学理论也并不是封闭的逻辑命题,它们从来不是那种一旦被反驳就整体倒塌的判断结构。任何所谓的“反例”,都只是在一组特定条件、特定辅助假设和特定测量方式之下,使某个模型失效。这种失效指向的是走出了适用范围,而不是对理论的终极裁决。
英文中的 falsifiability 描述的,恰恰不是“被证明为假”,而是“存在出错的可能性”。“证伪”这一翻译制造了一个极具误导性的转向,它把一个关于可能性的特征,悄然转换成了一个关于结论的判断。于是人们开始自然地说出诸如“某理论被证伪了”“一旦被证伪就不再是科学”这样的话,而这些说法无论在逻辑上还是在科学实践中,都难以成立。
事实上,我们完全可以使用“存在被反驳的可能性”“允许出错”“对反例开放”“具有失败条件”等表述,这些说法既保留了原概念的核心含义,又避免了把条件性风险误说成真值裁决。
“证伪”之所以能够长期存活,很大程度上是因为它带有一种专业权威的色彩,仿佛专家故意不说人话。但我还是要补充一点说明,由于专业术语在长期使用中已经形成了相对固定的含义,成为专有名词,确实不能简单地望文生义。遗憾的是,这种荒谬的翻译至今仍在影响一代又一代人,而误解并不会因为它被冠以“专业名词”的身份,就自动获得免疫。
What is science? This seemingly simple question has in fact been debated for a very long time. In particular, theories such as psychoanalysis, evolutionary theory, and relativity operate on temporal and spatial scales that far exceed what people can easily grasp through intuition or verify directly. For this reason, they have repeatedly been questioned over the years as to whether they truly count as science.
Around the question of “what counts as science,” there have been intense debates within philosophy. The philosopher of science Karl Popper, in The Logic of Scientific Discovery (1934), introduced the concept of falsifiability, later translated into Chinese as “证伪性” (literally, “verifiability of falsity”). Popper emphasized an asymmetry between universal laws and basic observational statements, and contrasted falsifiability with the intuitively similar notion of “verifiability” that was popular at the time among logical positivists. He pointed out that to verify a statement such as “all swans are white,” one would have to observe all swans empirically, which is in principle impossible; by contrast, observing just one black swan is sufficient to refute the claim.
Popper’s view earned respect in both scientific and philosophical circles, but it never became a universal consensus. Thomas Kuhn argued that, from the perspective of the history of science, science does not in fact operate through falsification in this straightforward way. During periods of normal science, counterexamples are often absorbed, anomalies are temporarily set aside, and theories are not abandoned simply because of a single failure. Kuhn’s work made it difficult, at the level of actual scientific practice, to maintain the claim that falsifiability is equivalent to the scientific method. Later, Imre Lakatos further argued that while Popper was right in principle, his model was overly simplified and could not, in its original form, be directly applied to real scientific practice.
Almost no contemporary mainstream philosopher holds that whether a theory is scientific can be determined solely by falsifiability.
I encountered the term “falsifiability” quite early on, not only in science and technology but also in legal contexts. Yet I never once chose to use it myself, because I felt an immediate and intense aversion to it. This aversion was not merely intellectual; it was a genuine physiological reaction. When I saw the term, I would feel a distinct discomfort in my stomach. When people used it to explain what science is, it struck me as utterly anti-logical and anti-intuitive. The reason is simple: how can “falsity” possibly be “verified”?
Only later, as my reading gradually shifted from primarily Chinese sources to primarily English ones, did I realize that this aversion stemmed largely from a mistranslation. Taken literally, falsifiability is much closer to “the possibility of being wrong.” With deeper reading, I came to see that Karl Popper was not trying to offer a positive definition of science at all; what he was really doing was drawing a boundary around it.
In the Chinese context, however, the question “what is science” has long been reduced to the criterion of “falsifiability.” This involves at least two layers of error: a misunderstanding of Popper’s original intent, and the adoption of an extremely inappropriate translation. No matter how one analyzes it, there is little in the word falsifiability that bears any relation to the notion of “verification.”
In the Chinese-speaking world, “证伪” has almost become a philosophical term that no longer requires explanation. Yet in terms of linguistic structure, logical meaning, and epistemological orientation, it contains problems that are hard to ignore. In Chinese, the verb “证” always carries a sense of confirmation. Whether in “证实” (to verify), “证据” (evidence), or “证明” (proof), it points to the same thing: fixing a previously doubtful judgment so that it becomes established, affirmed, and no longer open to question. Precisely for this reason, “证” is naturally associated with “true,” “real,” and “is,” rather than with negation.
“伪,” by contrast, signifies falsity, invalidity, or “not the case.” Forcing “证” and “伪” together to form “证伪” creates an internal tension—if not a direct conflict—within the Chinese language itself. In everyday usage, we do not say “证假” (“verify falsity”) or “证不真” (“verify non-truth”), because such expressions blatantly violate Chinese linguistic intuition. If a verb is meant to establish a state of affairs, it is ill-suited to establishing a negative one.
If falsifiability is taken as the criterion of science, then once something has been “falsified,” it is already “false”—so how could it still be scientific? This resembles the paradox of the statement “I am lying.”
Scientific theories are not closed logical propositions. They are never the kind of structures that collapse wholesale the moment they are refuted. Any so-called “counterexample” merely renders a model ineffective under a specific set of conditions, auxiliary assumptions, and measurement methods. Such failure indicates that the theory has moved beyond its domain of applicability, not that it has received a final verdict.
What falsifiability actually describes in English is not “being proven false,” but “the possibility of being wrong.” The translation “证伪” introduces a deeply misleading shift: it quietly turns a feature about possibility into a judgment about conclusions. As a result, people begin to say things like “this theory has been falsified” or “once falsified, it is no longer scientific,” statements that are difficult to sustain both logically and in actual scientific practice.
In fact, we could perfectly well use expressions such as “the possibility of being refuted,” “allowing for error,” “openness to counterexamples,” or “having conditions of failure.” These formulations preserve the core meaning of the original concept while avoiding the mistake of presenting conditional risk as a verdict on truth.
The reason “证伪” has survived for so long is largely that it carries an aura of professional authority, as if experts were deliberately refusing to speak plain language. I should add, however, that once technical terms have been used for a long time, they do acquire relatively fixed meanings as proper terms, and cannot simply be interpreted literally. Unfortunately, this absurd translation continues to influence generation after generation, and a misunderstanding does not become immune to criticism merely by being labeled a “technical term.”
了解 Geoffrey Chen 的更多信息
订阅后即可通过电子邮件收到最新文章。