Introverted Intuition

What does it mean to see a pattern before one can explain it?

Carl Jung called one version of this experience introverted intuition: perception turns away from the immediately given object and follows the image or process that the object releases within. Gestalt psychology, founded in the same historical period, asked a neighboring question in the laboratory: why does an organized whole appear before we have consciously assembled its parts? A century of work on implicit learning, insight, global precedence, predictive processing, and expert judgment now lets us separate several mechanisms that older theories gathered under the word intuition.

The thesis of this note is deliberately narrow:

Introverted intuition is best treated as a phenomenological description of whole-forming cognition, not as an experimentally established mental organ. Its characteristic experience can be reconstructed as weak cues becoming an inner configuration, that configuration crossing into consciousness as a felt whole, and the whole generating hypotheses about what is not yet visible.

This is not clairvoyance, and it is not proof that Jungian function stacks are psychometric facts. It is a bridge between a historically important description and mechanisms that psychology can test. For the broader typological vocabulary, see Analytical Psychology.

在能够解释一个模式之前，就已经看见它，究竟意味着什么？

卡尔·荣格把这类经验的一种形式称为内倾直觉（introverted intuition）：知觉不再停留于眼前给定的对象，而是追随对象在内心释放出来的图像或过程。诞生于同一历史时期的格式塔心理学，则在实验室中提出了一个相邻问题：为什么一个有组织的整体会先于我们对局部的有意识拼装而出现？此后一百年关于隐性学习、洞察、整体优先、预测加工和专家判断的研究，使我们能够把旧理论笼统放进“直觉”一词中的若干机制拆开。

本文的论点有意保持克制：

内倾直觉最好被视为对“形成整体的认知”所作的现象学描述，而不是一个已经被实验确认的心智器官。 它的典型经验可以被重构为：分散的弱线索形成一个内在构型；构型以“完整感”越过意识阈值；整体再对尚未显现之物生成假设。

这不是预知，也不能证明荣格功能栈是心理测量学事实。本文要做的是，在一个有历史解释力的描述与心理学可检验的机制之间搭桥。更完整的类型学词汇见 Analytical Psychology。

Before discussing Ni, we need to prevent several ordinary words from doing too many jobs. Consciousness, the unconscious, and meaning each name multiple research problems. Treating them as if they referred to single substances makes intuition sound more mysterious than it is.

1.1 Consciousness is access, experience, and monitoring

There is no single accepted scientific theory of consciousness. Contemporary theories disagree about recurrent processing, global broadcasting, higher-order representation, predictive processing, and information integration. One useful distinction is between phenomenal consciousness—there is something it feels like to have an experience—and access consciousness—information is available to memory, report, reasoning, and intentional control. Dehaene, Lau, and Kouider further distinguish global availability from metacognitive monitoring, the capacity to estimate confidence and error.

This essay does not settle those debates. Here, conscious means that a content has become part of lived experience and/or can be flexibly used and reported. Crucially, consciousness of a result does not entail consciousness of the processing that produced it. I can recognize a face, understand a sentence, or feel that a pattern is wrong without inspecting every computation underneath.

1.2 “The unconscious” names different things

Jung’s unconscious belongs to depth psychology: it includes personally excluded material, autonomous complexes, and—within his theory—collective forms expressed in symbols and images. The cognitive unconscious of experimental psychology is defined operationally: learning, priming, memory, evaluation, or perceptual organization can influence performance without complete awareness of their source or rule. A third use refers simply to subpersonal neural processing that never becomes a candidate for report.

These senses overlap only partially. Evidence for implicit statistical learning does not establish a collective unconscious; a Jungian dream interpretation does not demonstrate an unconscious decision algorithm. Even within cognitive psychology, researchers disagree about how powerful unconscious deliberation is. Bargh and Morsella review evidence for flexible unconscious guidance, while Newell and Shanks argue that many claims about complex unconscious decision making exceed the evidence. The safe claim is narrower: much processing lacks introspective access; its sophistication and reliability must be established task by task.

“Subconscious” is common and intuitively useful, but too imprecise for this distinction. Below, I use unconscious processing for experimentally non-reportable operations and the Jungian unconscious when discussing Jung’s historical model.

1.3 Meaning has at least four layers

When an intuition “finds meaning,” the phrase may refer to four different achievements:

1. Semantic meaning: what a sign, word, or image refers to.
2. Structural or causal meaning: what role an element plays in a pattern and what it implies.
3. Motivational meaning: why something matters to the organism—its relevance, value, or threat.
4. Narrative or existential meaning: how an event fits into a model of self, world, purpose, and time.

Semantic cognition research mainly studies the first layer; Gestalt psychology is strongest on the second; affect and decision research address the third; clinical and personality traditions often emphasize the fourth. Park’s meaning-making framework, for example, distinguishes a person’s relatively global orienting beliefs and goals from the situational meaning assigned to an event. Distress can arise when the two conflict, and “making meaning” is an attempt to revise one or both.

Jung’s inner images often bind all four layers at once. That density is part of their force, but also a source of confusion: a symbol can feel personally and narratively profound without being a correct causal explanation. Significance is not the same relation as truth.

1.4 The minimal process model

With these distinctions in place, the phenomenon can be drawn without positing a hidden oracle:

world → sensation → perceptual organization → latent integration → conscious hunch → explicit judgment → action and feedback

This is not a rigid one-way pipeline; attention, expectations, action, and feedback loop backward at every level. Its purpose is to separate two questions:

• How did a configuration become available?—the problem of perception, learning, and consciousness.
• Should I believe it?—the problem of evidence, judgment, and calibration.

Introverted intuition concerns the first question before it can responsibly answer the second.

在讨论 Ni 之前，需要先避免几个日常词汇承担过多工作。意识、无意识／潜意识与意义，各自都指向多个不同的研究问题。若把它们当成单一实体，直觉就会显得比实际更加神秘。

1.1 意识：经验、通达与监控

心理学至今没有一个被普遍接受的意识理论。当代理论分别强调循环加工、全局广播、高阶表征、预测加工或信息整合。一个有用的区分是：现象意识指某种经验“体验起来是什么样”；通达意识指信息能够进入记忆、报告、推理和有意控制。Dehaene、Lau 与 Kouider 又把信息的全局可用性同元认知监控区分开；后者是估计确信度与错误的能力。

本文不试图裁决这些争论。在这里，有意识表示一项内容已经进入生活经验，和／或能够被灵活使用与报告。关键是：意识到结果，不等于意识到产生结果的加工。我可以认出一张脸、理解一句话，或感到某个模式不对，却无法检查底下的每一步计算。

1.2 “无意识”指向不同的对象

荣格的无意识属于深度心理学：它包含被个人排除的材料、自主情结，以及——在他的理论内部——通过象征与图像表达的集体形式。实验心理学中的认知无意识采用操作性定义：学习、启动、记忆、评价或知觉组织能够在主体无法完整报告其来源或规则时影响行为。第三种用法则仅指从未成为可报告候选的次人格层神经加工。

这些含义只部分重叠。隐性统计学习的证据不能证明集体无意识；对荣格式梦境的解释也不能证明存在无意识决策算法。即使在认知心理学内部，研究者对无意识深思究竟有多强也存在争论。Bargh 与 Morsella 总结了灵活的无意识引导证据，而 Newell 与 Shanks 则认为，许多关于复杂无意识决策的主张超过了证据。更安全的结论较窄：大量加工缺少内省通路；它究竟多复杂、多可靠，必须逐任务证明。

“潜意识”在日常语言中常用而直观，但不足以承担上述区分。下文用无意识加工指实验意义上不可报告的操作；在讨论荣格的历史模型时，则明确说荣格式无意识。

1.3 意义至少有四个层次

当我们说直觉“发现了意义”，这句话至少可能指四种不同成果：

1. 语义意义：一个符号、词语或图像指向什么。
2. 结构或因果意义：一个元素在模式中扮演什么角色，又蕴含什么后果。
3. 动机意义：某物为什么对一个有机体重要——它的相关性、价值或威胁。
4. 叙事或存在意义：一件事如何进入关于自我、世界、目的与时间的模型。

语义认知研究主要处理第一层；格式塔心理学最擅长第二层；情感与决策研究处理第三层；临床和人格传统则常强调第四层。例如 Park 的意义建构框架，把一个人相对全局的定向信念与目标，同一件具体事件被赋予的情境意义区分开。当二者冲突时，痛苦可能出现；“建构意义”就是修改其中一方或双方的尝试。

荣格的内在图像常把四层意义一次绑定。这种密度是其力量的一部分，也是一种混淆来源：一个象征可以在个人和叙事上极其深刻，却仍不是正确的因果解释。重要性与真实性不是同一种关系。

1.4 最小过程模型

有了这些区分，我们无需假设一个隐藏的神谕，也可以画出直觉现象：

世界 → 感觉 → 知觉组织 → 潜在整合 → 有意识的预感 → 显式判断 → 行动与反馈

这不是僵硬的单向流水线；注意、期待、行动和反馈都会在每一层向后回流。这个图的目的，是分开两个问题：

• 一个构型如何进入意识？——这是知觉、学习与意识的问题。
• 我是否应该相信它？——这是证据、判断与校准的问题。

内倾直觉首先属于第一个问题；在它能够负责任地回答第二个问题之前尤其如此。

In Psychological Types (1921), Jung classifies intuition with sensation as an “irrational” function. In his technical usage, irrational does not mean foolish. It means perceptive rather than judging: intuition presents something; thinking and feeling evaluate it.

Jung’s decisive phrase is that introverted intuition is “directed upon the inner object.” An external event may trigger it, but attention remains with the subjective image and watches how that image changes and unfolds. This gives the construct seven linked properties:

1. Indirect access. Its intermediate processing is not available as a chain of reasons. Consciousness receives a result before it receives an account.
2. Inner reference. The target is not the external object alone, but the internal transformation produced by contact with it.
3. Configurational form. The content often appears as an image, scene, metaphor, or relation among elements—not initially as a proposition.
4. Temporal depth. The image is dynamic. It suggests where a process came from, how it is developing, and what it might become.
5. Convergence. Many cues are compressed into one organizing interpretation rather than expanded into many external possibilities.
6. Intrinsic conviction. Because the whole arrives before its derivation, it can feel self-authenticating.
7. Weak contact with correction. The same inwardness that produces synthesis can detach the image from bodily sensation, base rates, and the recalcitrance of the object.

The seventh property is not an accidental defect; it follows from Jung’s own architecture. A perceptive mode that privileges the inner image is powerful precisely where literal observation is incomplete, but dangerous when the image is allowed to replace observation. Mature intuition therefore needs a judging function to formulate the vision and renewed sensation to test it.

There is also a historical distinction worth preserving. Jung described a function-attitude within a theory of consciousness. The later MBTI turned aspects of that theory into a questionnaire and discrete type code. Those are not the same claim. Psychometric work has found that MBTI scales overlap with continuous personality dimensions, while support for sharply separated types and the proposed “type dynamics” is much weaker. A 2025 synthesis of 193 studies of MBTI Form M found strong internal consistency in the studies that reported it, but no structural-validity or test–retest studies in its 1999–2024 sample. So this essay uses Ni as a disciplined descriptive hypothesis, not as a diagnosis.

在《心理类型》（1921）中，荣格把直觉与感觉一起归为“非理性（irrational）”功能。在他的术语里，非理性并不是愚蠢，而是指知觉性的、尚未作判断的：直觉呈现某物；思维和情感才对它作出评价。

荣格的关键说法是，内倾直觉“指向内在对象”。外部事件可以触发它，但注意力停留在主观图像上，并观察图像如何变化、展开。由此可以推出七个相连的性质：

1. 间接通达。 中间加工过程并不以理由链的形式进入意识；意识先收到结果，后收到解释。
2. 内在指向。 对象不只是外部事物本身，而是心智与它接触之后所产生的内部变换。
3. 构型形式。 内容常以图像、场景、隐喻或元素之间的关系出现，而不是一开始就成为命题。
4. 时间纵深。 图像是动态的；它暗示一个过程从哪里来、正在怎样发展、可能变成什么。
5. 收敛。 大量线索被压缩为一个组织性解释，而不是向外展开为许多并列可能。
6. 内生确信。 因为整体先于推导抵达，它容易给人一种能够自我担保的感觉。
7. 纠错接触较弱。 产生综合的同一种内向性，也可能让图像脱离身体感觉、基础率和对象本身的顽固事实。

第七点不是偶然缺陷，而是荣格架构自身的推论。一个优先处理内在图像的知觉模式，恰恰在字面观察不完整时最有力量；但当图像开始替代观察时，它也最危险。因此，成熟的直觉需要判断功能把愿景表达出来，并重新接入感觉，让现实检验它。

还应保留一个历史区分。荣格描述的是意识理论中的一种功能—态度；后来的 MBTI 把其中一些部分变成问卷和离散类型码。这不是同一个主张。心理测量研究发现，MBTI 各量表与连续人格维度有所重合，但对清晰分离的类型以及所谓“类型动力学”的支持弱得多。2025 年对 MBTI Form M 的 193 项研究所作的综合发现：在报告内部一致性的研究中，其数值较高；但 1999–2024 年的样本里没有结构效度或重测信度研究。因此，本文把 Ni 当成一个受到约束的描述性假设，而不是诊断。

Gestalt is often translated as “whole,” but the important idea is organized form. In Wertheimer’s apparent-motion experiments, two stationary stimuli could be experienced as one movement. The experience therefore contained a relation that was not present as a property of either isolated stimulus. Later work studied figure–ground organization and grouping by proximity, similarity, closure, good continuation, common fate, and other constraints.

Three principles matter for intuition:

• Relations have causal force in perception. A part is not interpreted independently; its role depends on the configuration in which it appears.
• Organization can precede inventory. Navon’s global-precedence experiments found that observers can process the “forest” before the local “trees,” although later research shows that stimulus scale, attention, and task can modify the effect.
• A solution may require restructuring, not more search in the same representation. From Köhler’s ape studies through Duncker’s work on fixation to modern insight experiments, the critical event is often a change in what the elements are for the solver.

This is the genuine affinity between Gestalt and introverted intuition. Both describe a transition in which scattered elements become intelligible because a whole reorganizes their meaning. Neither says that the mind simply adds one fact to another.

But they should not be identified:

Gestalt psychology therefore does not validate Ni. It supplies a rigorous precedent for one of its deepest intuitions: perception is constructive, relational, and sometimes whole-first.

Gestalt 常被译作“整体”，但真正重要的是有组织的形式。在韦特海默的似动实验中，两个静止刺激可以被体验为一次运动；经验中因此出现了一种关系，而这种关系并不是任何单个刺激的属性。后来的研究进一步考察了图形—背景组织，以及邻近、相似、闭合、良好连续、共同命运等分组约束。

其中三个原则与直觉尤其相关：

• 关系在知觉中具有因果力量。 局部不是被独立解释的；它在整体构型中的位置决定了它是什么。
• 组织可以先于清点。 Navon 的整体优先实验发现，观察者可以先处理“森林”，再处理局部的“树”；不过后续研究也表明，刺激尺度、注意和任务会改变这一效应。
• 解题可能需要重构，而不是在同一表征里搜索得更久。 从 Köhler 的猩猩研究、Duncker 对固着的研究，到现代洞察实验，关键事件往往是组成元素对解题者而言突然“变成了别的东西”。

这正是格式塔与内倾直觉真正相通之处：两者都描述了这样一种转变——分散元素之所以变得可理解，是因为一个整体重新组织了它们的意义。两者都不是把认知描述成事实的线性相加。

但二者不能被等同：

因此，格式塔心理学并不“验证” Ni。它提供了一个严格的先例，支持 Ni 最深的一条直觉：知觉具有建构性、关系性，并且有时是整体优先的。

No single modern theory is “the science of Ni.” Several well-studied processes can, however, jointly produce an experience with its shape.

Two results sharpen the picture.

First, unconscious does not have to mean inactive. Reber’s artificial-grammar studies and Saffran, Aslin, and Newport’s infant work show that nervous systems extract distributional structure without being taught an explicit symbolic rule. Bowers and colleagues then provided a particularly close model of a hunch: partial clues activate related memory and semantic networks; a patterned activation can guide behavior before its common theme can be named.

Second, sudden consciousness does not imply sudden computation. Metcalfe and Wiebe found that reported “warmth” rose gradually for analytic problems but remained relatively flat before insight solutions. Recent work refines rather than abolishes this distinction. Wiley and Danek’s 2024 review separates restructuring, a change in the problem representation, from the Aha experience, its subjective suddenness and affect; the two often coincide, but not always. In 2025, Chao and colleagues used remote-associate tasks and simulation to show that insight was associated with longer-distance exploration in a solution space. The whole can arrive suddenly because a search or reorganization that was not introspectively accessible has finally crossed a reporting threshold.

This suggests a useful formula:

Intuition is not reasoning with zero steps. It is cognition whose decisive steps are compressed, distributed, or unavailable to introspection.

现代心理学里不存在一个统一的“Ni 科学”。但是，几种研究较充分的过程可以共同产生与它形状相似的经验。

两个结果使这幅图更清晰。

第一，无意识并不等于没有加工。Reber 的人工语法实验，以及 Saffran、Aslin 与 Newport 的婴儿研究都表明，神经系统无需被教授显式符号规则，也会抽取分布结构。随后 Bowers 等人给出了一个尤其接近“预感”的模型：局部线索激活相关记忆和语义网络；在共同主题能够被命名之前，带有模式的激活已经可以引导行为。

第二，突然进入意识，并不意味着计算也是突然发生的。Metcalfe 与 Wiebe 发现，分析题的“接近答案感”会逐渐上升，而洞察题在解答出现之前通常保持较平。近年研究不是取消这个区分，而是把它做得更细。Wiley 与 Danek 在 2024 年的综述中区分了重构（问题表征发生改变）与 Aha 体验（主观上的突然性与情绪）；两者常常同时出现，却并非必然重合。2025 年，Chao 等人利用远距联想任务和模拟发现，洞察与解空间中的长距离探索相关。整体之所以可以突然抵达，是因为一段无法被内省访问的搜索或重组终于越过了报告阈值。

由此得到一个有用的公式：

直觉不是“零步骤推理”，而是决定性步骤被压缩、分布式完成，或无法被内省访问的认知。

Jungian analyst John Beebe summarizes introverted intuition with three verbs: imagining, knowing, divining. They are not three experimentally validated stages, but they form an unusually good phenomenological decomposition.

5.1 Imagining: construct the inner configuration

Here imagining does not mean decorative fantasy. It means releasing the current representation from literal appearance so that elements can acquire new roles. In Gestalt language, the field is reorganized; in predictive-processing language, an internal generative model proposes a hidden cause that could make the observations cohere.

The essential operation is counterfactual configuration: if these cues were expressions of one process, what would that process look like?

5.2 Knowing: let coherence cross the threshold

Knowing is the moment at which the configuration becomes compelling enough to guide attention. Bowers’s tacit-coherence results are a close experimental analogue: a person can discriminate the more coherent set before identifying why. The output is not yet a proof. It is a weighted signal that one representation organizes more of the evidence than its competitors.

The phenomenology is certainty; the epistemic status is candidate hypothesis.

5.3 Divining: run the configuration forward

Beebe’s divining need not mean supernatural prophecy. Read naturalistically, it is temporal extrapolation: if the inferred process is real, what should appear next, what hidden variable should exist, or what intervention should change the trajectory? This combines prediction with mental simulation.

That final move converts a private gestalt into public consequences:

distributed traces → inner configuration → felt coherence → forward simulation → risky prediction → external test

The arrows after “felt coherence” are crucial. Without them, imagining can harden into fantasy and knowing into dogma. With them, divining becomes accountable forecasting.

荣格分析师 John Beebe 用三个动词概括内倾直觉：imagining、knowing、divining——想象、知晓、推演。它们不是三个已经被实验验证的阶段，却构成了一个异常清楚的现象学拆解。

5.1 想象：建构内在构型

这里的想象不是装饰性的幻想，而是让当前表征摆脱字面外观，使组成元素能够获得新的角色。用格式塔语言说，问题场被重新组织；用预测加工语言说，内部生成模型提出了一个可能使观察彼此连贯的隐藏原因。

其核心操作是反事实构型：如果这些线索是同一个过程的表现，那个过程会是什么形状？

5.2 知晓：让连贯性越过阈值

知晓，是一个构型变得足够有说服力、因而开始引导注意的时刻。Bowers 的默会连贯性结果是一个很接近的实验类比：人可以先区分出更连贯的一组材料，之后才说出原因。这个输出还不是证明；它只是一个加权信号，表示某个表征比竞争者组织了更多证据。

它的现象学形式是确信；它的认识论地位只是候选假设。

5.3 推演：让构型沿时间向前运行

Beebe 的 divining 不必被理解为超自然预言。以自然主义方式理解，它是时间外推：如果推测出的过程真实存在，下一步应该出现什么？哪个隐藏变量也应当存在？何种干预会改变轨迹？这一步把预测与心理模拟结合起来。

最后这一步，把私人格式塔转化成公共后果：

分布痕迹 → 内在构型 → 连贯感 → 前向模拟 → 有风险的预测 → 外部检验

“连贯感”之后的箭头至关重要。没有它们，想象会硬化成幻想，知晓会硬化成教条；有了它们，推演才成为需要负责的预测。

The feeling of rightness is data about a cognitive event, not a certificate about the world.

Insight solutions were more accurate than analytic solutions across four task families in Salvi and colleagues’ 2016 experiments, so the Aha signal is not meaningless. Yet Danek and Wiley demonstrated that incorrect solutions can also produce Aha experiences. Their 2024 review concludes that restructuring and Aha phenomenology dissociate in both directions. The evidence through 2026 makes the caution sharper: Cason and Tidikis measured physiological arousal during insight problems and found that arousal peaks did not generally distinguish correct from incorrect solutions in their sample. A bodily “click” is no oracle.

Intuition becomes reliable under identifiable conditions. Kahneman and Klein—representing the heuristics-and-biases and naturalistic-decision traditions—agreed on two prerequisites for skilled intuition:

1. The environment contains learnable, sufficiently stable regularities.
2. The learner receives enough practice and reasonably rapid, unambiguous feedback.

Chess and firefighting can sometimes satisfy these conditions. Long-range political forecasting, stock selection, interpersonal mind-reading, and one-off scientific bets often do not. Expertise is therefore local: years of feedback in one cue ecology do not create a universal faculty for hidden truth.

The complementary failure mode is apophenia, finding organization in noise. Initial experiments linked lack of control to illusory pattern perception; later preregistered replication attempts did not reproduce the proposed causal effect consistently. The broader lesson is more important than either single result: pattern perception must itself be studied under replication and calibration. A beautiful pattern deserves a better test, not automatic belief.

For practical use, an intuitive judgment should pass six gates:

1. Name the observation separately from the interpretation.
2. Record the prediction before seeing the outcome.
3. State what would disconfirm it.
4. Generate at least one competing gestalt.
5. Check base rates and seek concrete sensory evidence.
6. Track calibration over repeated cases, not memorable anecdotes.

This protocol does not suppress intuition. It completes it. The inner whole performs discovery; explicit judgment and contact with the object perform justification.

正确感是一次认知事件的数据，而不是世界给出的证书。

在 Salvi 等人 2016 年覆盖四类任务的实验中，洞察式答案平均比分析式答案更准确，所以 Aha 信号并非毫无信息。但 Danek 与 Wiley 也证明，错误答案同样会产生 Aha 体验。他们 2024 年的综述认为，表征重构与 Aha 现象可以双向分离。直到 2026 年的证据让这一警告更加清楚：Cason 与 Tidikis 测量洞察解题时的生理唤醒，在其样本中，唤醒峰值通常不能区分正确与错误答案。身体里的“咔哒”并不是神谕。

直觉只有在可识别的条件下才会变得可靠。分别代表“启发式与偏差”传统和“自然决策”传统的 Kahneman 与 Klein，最终同意熟练直觉需要两个前提：

1. 环境包含可以学习、足够稳定的规律。
2. 学习者拥有充分练习，并得到相对迅速、明确的反馈。

国际象棋和消防判断有时能够满足这些条件；长期政治预测、选股、揣测他人内心，以及一次性的科学赌注往往不能。因此，专长是局部的：在一种线索生态中接受多年反馈，并不会生成一种通向所有隐藏真理的普遍能力。

与之互补的失败模式是空想性错视（apophenia），即在噪声中发现组织。早期实验把控制感缺失与虚假模式知觉联系起来；后续预注册复现实验并未稳定复现这个因果效应。比任何一项单独结果都更重要的教训是：模式知觉本身也必须接受复现和校准。一个漂亮的模式值得的是更好的检验，而不是自动相信。

在实践中，一项直觉判断至少应通过六道门：

1. 把观察与解释分开命名。
2. 在看到结果之前记录预测。
3. 提前说明什么会推翻它。
4. 至少生成一个竞争性的整体解释。
5. 检查基础率，并寻找具体感官证据。
6. 在重复案例上追踪校准，而不是依赖令人难忘的故事。

这套协议不是压制直觉，而是完成直觉。内在整体负责发现；显式判断以及与对象重新接触负责证成。

The historical and experimental lines can now be joined without pretending that they are identical:

Introverted intuition is a mode of perception in which distributed cues are compressed into an internally generated configuration; the emergent whole becomes available before its derivation and supports hypotheses about hidden structure and future development.

Every clause matters:

• mode, not a permanent identity;
• perception, not yet judgment;
• distributed cues, not revelation from nothing;
• internally generated, but triggered and constrained by contact with a world;
• configuration, because relations matter more than isolated features;
• before its derivation, explaining opacity and suddenness;
• hypotheses, preserving fallibility;
• hidden structure and future development, distinguishing it from mere familiarity.

As of 2026, the evidence has three levels. Perceptual grouping, implicit/statistical learning, global–local interaction, restructuring, predictive influence, and domain-specific expert recognition are established research programs. The proposal that their conjunction reconstructs an experience resembling Jung’s Ni is a plausible synthesis. The claim that Ni is a unitary neural module, a sharply bounded personality type, or a source of knowledge independent of evidence remains unsupported.

The enduring value of Jung’s concept is therefore not that it solved intuition. It named a problem that remains scientifically alive: how does a mind turn traces into a whole, and how can that whole become knowledge without becoming illusion?

现在，我们可以把历史线索与实验线索接起来，同时不假装它们彼此等同：

内倾直觉是一种知觉模式：分布的线索被压缩为一个由内部生成的构型；涌现的整体先于其推导进入意识，并支持对隐藏结构与未来发展的假设。

这一定义中的每个词都很重要：

• 是一种模式，不是永久身份；
• 是知觉，尚未成为判断；
• 来自分布线索，不是无中生有的启示；
• 由内部生成，但仍由同世界的接触所触发和约束；
• 是一个构型，因为关系比孤立特征更重要；
• 先于推导，从而解释不透明性和突然性；
• 只产生假设，因此保留可错性；
• 面向隐藏结构与未来发展，从而区别于单纯熟悉感。

截至 2026 年，证据可以分成三层。知觉分组、隐性／统计学习、整体—局部交互、表征重构、预测影响以及领域特定的专家识别，都是已经成立的研究项目。“这些机制的结合能够重构一种与荣格 Ni 相似的经验”，是一个合理的综合假设。而“Ni 是单一神经模块、边界清晰的人格类型，或不依赖证据的知识来源”，仍然没有得到支持。

因此，荣格概念持久的价值，不在于它已经解决了直觉，而在于它命名了一个至今仍有科学生命的问题：心智如何把痕迹变成整体？这个整体又如何成为知识，而不沦为幻觉？

• Wertheimer, M. (1912). Experimentelle Studien über das Sehen von Bewegung. Zeitschrift für Psychologie, 61, 161–265.
• Jung, C. G. (1921/1971). Psychological Types, Chapter X. Collected Works, Vol. 6. Princeton University Press.
• Köhler, W. (1925). The Mentality of Apes. Kegan Paul.
• Duncker, K. (1945). On problem-solving. Psychological Monographs, 58(5), i–113.
• Reber, A. S. (1967). Implicit learning of artificial grammars. Journal of Verbal Learning and Verbal Behavior, 6(6), 855–863.
• Navon, D. (1977). Forest before trees: The precedence of global features in visual perception. Cognitive Psychology, 9(3), 353–383.
• Metcalfe, J., & Wiebe, D. (1987). Intuition in insight and noninsight problem solving. Memory & Cognition, 15(3), 238–246.
• McCrae, R. R., & Costa, P. T. (1989). Reinterpreting the Myers–Briggs Type Indicator from the perspective of the five-factor model of personality. Journal of Personality, 57(1), 17–40.
• Bowers, K. S., Regehr, G., Balthazard, C., & Parker, K. (1990). Intuition in the context of discovery. Cognitive Psychology, 22(1), 72–110.
• Block, N. (1995). On a confusion about a function of consciousness. Behavioral and Brain Sciences, 18(2), 227–247.
• Saffran, J. R., Aslin, R. N., & Newport, E. L. (1996). Statistical learning by 8-month-old infants. Science, 274(5294), 1926–1928.
• Rao, R. P. N., & Ballard, D. H. (1999). Predictive coding in the visual cortex. Nature Neuroscience, 2(1), 79–87.
• Bar, M., et al. (2006). Top-down facilitation of visual recognition. PNAS, 103(2), 449–454.
• Bargh, J. A., & Morsella, E. (2008). The unconscious mind. Perspectives on Psychological Science, 3(1), 73–79.
• Whitson, J. A., & Galinsky, A. D. (2008). Lacking control increases illusory pattern perception. Science, 322(5898), 115–117.
• Kahneman, D., & Klein, G. (2009). Conditions for intuitive expertise: A failure to disagree. American Psychologist, 64(6), 515–526.
• Sio, U. N., & Ormerod, T. C. (2009). Does incubation enhance problem solving? A meta-analytic review. Psychological Bulletin, 135(1), 94–120.
• Park, C. L. (2010). Making sense of the meaning literature. Psychological Bulletin, 136(2), 257–301.
• Wagemans, J., et al. (2012). A century of Gestalt psychology in visual perception I. Psychological Bulletin, 138(6), 1172–1217.
• Clark, A. (2013). Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behavioral and Brain Sciences, 36(3), 181–204.
• Kounios, J., & Beeman, M. (2014). The cognitive neuroscience of insight. Annual Review of Psychology, 65, 71–93.
• Newell, B. R., & Shanks, D. R. (2014). Unconscious influences on decision making: A critical review. Behavioral and Brain Sciences, 37(1), 1–19.
• Salvi, C., Bricolo, E., Kounios, J., Bowden, E., & Beeman, M. (2016). Insight solutions are correct more often than analytic solutions. Thinking & Reasoning, 22(4), 443–460.
• Dehaene, S., Lau, H., & Kouider, S. (2017). What is consciousness, and could machines have it?. Science, 358(6362), 486–492.
• Danek, A. H., & Wiley, J. (2017). What about false insights?. Frontiers in Psychology, 7, 2077.
• Lambon Ralph, M. A., Jefferies, E., Patterson, K., & Rogers, T. T. (2017). The neural and computational bases of semantic cognition. Nature Reviews Neuroscience, 18(1), 42–55.
• de Lange, F. P., Heilbron, M., & Kok, P. (2018). How do expectations shape perception?. Trends in Cognitive Sciences, 22(9), 764–779.
• van Elk, M., & Lodder, P. (2018). Experimental manipulations of personal control do not increase illusory pattern perception. Collabra: Psychology, 4(1), 19.
• Beebe, J. (2021). The Eight Function-Attitudes Unpacked.
• Seth, A. K., & Bayne, T. (2022). Theories of consciousness. Nature Reviews Neuroscience, 23, 439–452.
• Wiley, J., & Danek, A. H. (2024). Restructuring processes and Aha! experiences in insight problem solving. Nature Reviews Psychology, 3, 42–55.
• Chao, Z. C., Hsieh, F.-Y., & Wu, C.-T. (2025). Long-distance exploration in insightful problem-solving. Communications Psychology, 3, 53.
• Erford, B. T., Zhang, X., Sweeting, E. L., et al. (2025). A 25-year review and psychometric synthesis of the MBTI Form M. Journal of Counseling & Development, 103(4), 403–417.
• Cason, B., & Tidikis, V. (2026). Exploring the relationship between physiological arousal and insight problem solving. Imagination, Cognition and Personality, 45(3), 293–312. (Online first 2025; issue published 2026.)