A bilingual digital institution dedicated to the brain, thinking methods, mental models, learning science, decision-making, creativity, consciousness, and the future of intelligence. The most uncharted territory we have left to explore is between our own ears.
献给大脑、思维方式、思维模型、学习科学、决策、创造力、意识与智能未来的双语数字机构。我们尚未勘察的最重要疆域，就在我们两耳之间。
For ten thousand years humans have learned to bend matter — to fire, to seed, to steel, to silicon, to satellite. We have learned almost nothing about the substance that does the bending. The brain remains the most complex object in the known universe, and the methods by which it thinks remain mostly invisible to the thinker. The Atlas of Mind is an attempt to render that invisibility legible.
一万年来，人类学会了塑造物质——以火、以种子、以钢、以硅、以卫星。但我们几乎不了解那位「塑造者」自身。大脑仍是已知宇宙中最复杂的物体，而它思考所用的方法，对思考者本身大多不可见。《思维之图》是把这种不可见性化为可读的尝试。
It is not a productivity site. It is not a self-help platform. It is a digital monument — drawing on neuroscience, cognitive science, philosophy, learning theory, and the long history of human inquiry into mind — built around a simple premise: that the next great frontier of civilization is not space, not silicon, not even artificial intelligence. It is understanding the substrate that produces all three.
这不是一个效率工具站，也不是一个自我提升平台。这是一座数字纪念碑——汲取神经科学、认知科学、哲学、学习理论与人类心智探究的悠久历史——围绕一个简单前提构建：文明的下一个伟大边疆不是太空，不是硅芯，甚至不是人工智能。是理解产生这一切的基底。
"What lies behind us and what lies before us are tiny matters compared to what lies within us."
「我们身后之事与身前之事，与我们内里所藏相比，都微不足道。」— RALPH WALDO EMERSON · 爱默生
A handful of figures that begin to convey the magnitude of what every human carries inside their skull. Each one understated; each one will be extended, refined, or replaced as neuroscience progresses through the 21st century.
几个开始呈现「每个人颅骨内所承载之物」量级的数字。每一个都仍然保守；每一个都会随 21 世纪神经科学的推进而延展、精炼或被替代。
Three pounds of fat and water. Ten percent of your body's energy budget. The only structure in the universe known to model itself. The brain is not a computer (the metaphor breaks down quickly), but it is a substrate — biological, electrochemical, plastic — through which everything you have ever experienced has passed.
三磅重的脂肪与水。占据身体能量预算的 10%。宇宙中已知唯一能为自身建模的结构。大脑不是计算机（这个隐喻很快就站不住脚），但它是一个基底——生物的、电化学的、可塑的——你所经历过的一切都曾从中穿过。
Modern neuroscience is closer to a 17th-century atlas than to a finished map. We have named the regions; we still cannot read them. Connectomics, cellular imaging, optogenetics, and predictive coding are the four telescopes through which the brain is currently coming into focus.
现代神经科学更像 17 世纪的地图集，而非已完成的地图。我们为各个区域命过名；尚不能读懂它们。连接组学、细胞成像、光遗传学、预测编码是当下让大脑逐渐聚焦的四台望远镜。
Most people default to one or two thinking modes and use them across all situations — even when those situations call for a different tool. The deepest cognitive skill is not "having the right style" but being able to switch deliberately when context changes.
多数人默认使用一两种思维模式，并把它们应用于所有情境——即便情境呼唤不同的工具。最深的认知技能不是「拥有正确的方式」，而是能在情境变化时刻意切换。
Decompose a complex problem into bounded sub-parts. Solve each. Reassemble. The default mode of engineering, accounting, science. Reliable but misses emergent properties.
把复杂问题分解为边界清晰的子部分。逐一解决。再组装。工程、会计、科学的默认模式。可靠但忽略涌现性质。
Test claims against evidence, logic, and source quality. Ask "how do you know?" before "what do you mean?" Catches motivated reasoning; can become reflexive negation.
用证据、逻辑与信息源质量检验主张。先问「你怎么知道？」再问「你是什么意思？」识破动机性推理；易变为反射性否定。
Behavior emerges from structure: stocks, flows, feedback loops. Map the loop before reaching for the lever. Reveals counter-intuitive interventions; can paralyze under scale.
行为由结构涌现：存量、流量、反馈回路。在按下杠杆之前先映射回路。揭示反直觉干预；规模过大时易陷瘫痪。
Strip the problem down to physical or logical primitives. Refuse to reason by analogy. Rebuild upward. Slow but breaks priors. Unlocks 10× moves.
把问题剥离到物理或逻辑基元。拒绝以类比推理。自下而上重建。慢但能击穿成见。解锁十倍突破。
Edward de Bono. Approach the problem from oblique angles — provocations, random word, reversal. Specifically designed to escape the dominant frame. Useless when the frame is already right.
由 de Bono 提出。从倾斜角度切入问题——挑衅性命题、随机词、反转。专为逃离主导框架而设计。当框架本就正确时无用。
Treat outcomes as distributions, not single values. Bayes; expected value; calibrated uncertainty. Robust under uncertainty; counter-intuitive to most humans.
把结果视为分布而非单一值。贝叶斯；期望值；校准的不确定性。在不确定下稳健；对多数人反直觉。
A mental model is a compressed pattern of cause and effect that can be reapplied across domains. Charlie Munger collected hundreds; we present eight here that punch above their weight. Master one a year; in a decade you have a different mind.
思维模型是一种因果关系的压缩模式，可跨领域复用。Charlie Munger 收集了数百个；我们在此呈现八个最具效用的。每年掌握一个；十年后你将拥有一颗不同的心智。
"What would guarantee failure?" Then avoid that. Reverse-engineer the disaster.
「什么能保证失败？」然后避开它。逆向工程灾难。
"And then what happens?" Most decisions are evaluated only on first-order effects. The interesting consequences are downstream.
「然后呢？」多数决策只在一阶效应层评估。有趣的后果在下游。
The true cost of any choice is the next-best alternative you forfeit. Money is rarely the real cost; time and attention almost always are.
任何选择的真正代价是你放弃的次优选项。金钱很少是真正的代价；时间与注意力几乎总是。
Build for failure modes you cannot predict. Engineering, finance, and life all benefit from buffer beyond what the model suggests.
为你无法预测的失败模式留余地。工程、金融、人生皆受益于「超出模型所示」的缓冲。
Most outcomes are not bell-curve. A few of anything dominate; the rest cluster near zero. Bestselling books, startups, friendships, ideas.
多数结果不服从正态分布。少数事物主导；其余聚集近零。畅销书、创业公司、友谊、想法皆如此。
Tiny gains, repeated, beat large gains, sporadic. The most under-appreciated force in human affairs — financial, intellectual, relational.
微小但持续的收益胜过偶发的巨大收益。人类事务中最被低估的力量——财务、智识、关系皆如是。
Hold beliefs as probabilities, not certainties. Update them in proportion to evidence. The hardest thing about this model is doing it on yourself.
把信念视为概率而非确定性。按证据比例更新。这个模型最难之处是对自己使用它。
A peak you can see from where you stand may not be the highest peak. Sometimes you must descend before you can ascend further.
从你立足之处所见的山峰未必是最高峰。有时你必须先下降才能继续上升。
Six rungs lie between matter and "what-it-is-like-to-be-you." The first three are well understood; the last three are under active scientific siege. Between rung five and rung six lies the hard problem of consciousness — why any of this should feel like anything.
在物质与「身为你之感」之间，有六阶之距。前三阶已被良好理解；后三阶仍在科学的积极围攻中。第五与第六阶之间，是意识的难题——为何这一切「应该有任何感受」。
Three contemporary theories compete: Integrated Information Theory (Tononi) — consciousness is integrated information, measurable as Φ. Global Workspace Theory (Dehaene) — consciousness is the broadcast of a contender across cortical regions. Higher-Order Theory — consciousness is having a thought about a thought. None has been falsified; none is fully accepted.
三种当代理论竞争：整合信息论（Tononi）——意识即整合信息，可由 Φ 量化。全局工作空间理论（Dehaene）——意识是某竞争者跨皮层区域的广播。高阶理论——意识是「关于想法的想法」。无一被证伪；无一被完全接受。
"The question is not whether machines think but whether men do."
「问题不在于机器是否能思考，而在于人是否在思考。」— B.F. SKINNER · 斯金纳
Daniel Kahneman's framing (Thinking, Fast and Slow, 2011): two systems run in parallel. System 1 is fast, automatic, effortless, often right in familiar contexts and confidently wrong outside them. System 2 is slow, deliberate, effortful, accurate but expensive. Most cognitive errors are System 1 misfires that System 2 failed to catch.
Daniel Kahneman 的框架（《思考，快与慢》，2011）：两套系统并行。系统 1 快、自动、轻松，在熟悉情境中常对，在外则自信而错。系统 2 慢、刻意、费力，准确但昂贵。多数认知错误是系统 2 没能拦截的系统 1 失火。
The mature decision-maker has not eliminated System 1 — that is impossible and would be undesirable. They have learned to recognize when each is appropriate and to invoke System 2 deliberately for high-stakes, unfamiliar, or counter-intuitive choices. The recognition itself is a System 2 skill.
成熟的决策者并未消除系统 1——那不可能也不可取。他们学会识别两者各自合适的时刻，并在高赌注、陌生或反直觉选择中刻意调用系统 2。这种识别本身即是一项系统 2 技能。
Creativity is not a single act. It is a two-stroke engine. Stroke one — divergent thinking — generates options. Many, wide, weird, deliberately not-yet-judged. Stroke two — convergent thinking — narrows, evaluates, eliminates, ships. People who can do only one make either piles of unmade decisions or polished versions of the obvious.
创造力不是单一动作。它是一台双冲程发动机。冲程一——发散性思维——产生选项。多、广、怪、刻意不予评判。冲程二——收敛性思维——筛选、评估、剔除、交付。只能做其一者，要么得到一堆「未做的决定」，要么得到「显然之物」的精修版。
The schedule matters. Insights almost never come during forced ideation. They come in the shower, on a walk, in the moment between one task and the next. The mind requires the pressure of the question and then the release of attention to deliver the answer.
节律重要。洞见几乎从不来自强制构思。它们来自淋浴、散步、一项任务与下一项之间。心智需要先承受问题的压力，再释放注意力，才能交付答案。
For 6 million years, only one species could think. Now there are two systems on Earth that can manipulate symbols at scale — biological brains and artificial neural networks. They are not the same kind of intelligence. The brain is embodied, evolved, finite, slow, and grounded in a life. AI is disembodied, designed, vast, fast, and grounded in text. Each is strong where the other is weak.
六百万年来，只有一个物种能思考。如今地球上有两种系统能在规模上操纵符号——生物大脑与人工神经网络。它们不是同一类智能。大脑具身、演化而成、有限、缓慢、扎根于生命。AI 非具身、被设计、广博、迅疾、扎根于文本。各自之长，正是对方之短。
The most likely shape of the next decade of cognition is neither pure human nor pure AI but a third thing — small teams of humans wielding fleets of AI agents. Humans choose what is worth knowing; AI explores the space of possible knowing; humans ratify what counts. The unit of cognition is becoming hybrid.
未来十年认知最可能的形态既非纯人类亦非纯 AI，而是第三者——由 AI 智能体集群武装的小型人类团队。人类选择「什么值得被认知」；AI 探索「可能认知」的空间；人类确认「什么算数」。认知的基本单位正在变为混合体。
Click any node to learn what it is and how it connects. The map below is a starting cartography — eight major regions of the mind, eight bridges between them, eight gateways out. The map is not the territory; this map exists to make the territory askable.
点击任一节点以了解其为何物、如何连接。下方地图是一份起点制图——心智的八个主要区域，八座连接它们的桥，八个通向外部的关口。地图非疆域；此图之存在，是为了让疆域变得「可询问」。
A short, deliberately incomplete history of when we figured out what about ourselves. Each entry names a discovery; each understates the work behind it.
关于「我们何时弄懂了关于自己什么」的简短而有意不完整的历史。每一条命名一项发现；每一条都低估了其背后的工作。
Socrates establishes that introspective inquiry — "know thyself" — is the foundation of any meaningful philosophy.
苏格拉底确立了内省探究——「认识你自己」——是任何有意义哲学的基础。
First systematic treatise on the mind. Distinguishes nutritive, sensitive, and rational souls — the first cognitive taxonomy.
关于心智的首部系统论著。区分营养性、感知性与理性灵魂——首个认知分类法。
Cogito, ergo sum. Mind-body dualism. The frame Western philosophy will spend the next 380 years trying to escape.
我思故我在。心物二元论。西方哲学之后 380 年都在试图逃离的框架。
Leipzig, Germany. Psychology becomes a science by becoming experimental. Until this date it was philosophy.
德国莱比锡。心理学因转向实验而成为科学。在此之前，它属于哲学。
The unconscious enters mainstream Western thought. Most of Freud's specifics will be overturned; the framing of "depth psychology" will not.
无意识进入西方主流思想。弗洛伊德的多数具体观点将被推翻；「深度心理学」的框架则不会。
Information becomes a measurable quantity. Cognitive science becomes possible because thought becomes formally tractable.
信息成为可度量的量。认知科学因「思考可被形式化处理」而成为可能。
McCarthy, Minsky, Shannon, Rochester. Coining of "artificial intelligence." The first explicit attempt to model thought in machines.
麦卡锡、明斯基、香农、罗切斯特。「人工智能」一词诞生。首次明确尝试在机器中建模思考。
Language as an innate cognitive faculty. The cognitive revolution begins; behaviorism begins its decline.
语言作为先天认知能力。认知革命开始；行为主义开始衰落。
For prospect theory. The work that becomes Thinking, Fast and Slow. Cognitive bias enters public vocabulary.
凭前景理论获奖。该工作后来汇成《思考，快与慢》。认知偏误进入公共词汇。
Artificial neural networks begin to perceive. The "AI winter" ends. The path to GPT, AlphaFold, and beyond opens.
人工神经网络开始感知。「AI 寒冬」结束。通往 GPT、AlphaFold 及更远之路打开。
Frontier AI models reason about novel situations in ways no non-human animal ever has — but still cannot reliably introspect, want, or feel. The brain remains the only known instance of consciousness. The atlas continues.
前沿 AI 模型对新情境进行任何非人动物所未及的推理——但仍不能可靠地内省、欲望或感受。大脑仍是已知意识的唯一实例。图谱仍在延展。