① 这与你熟悉的"读没有文档的遗留代码"同构——你无法直接看懂实现，只能从结构反推"它当初要满足的需求是什么"。对人脑用同样的手法：一个看似非理性的情绪反应，先问"它在祖先环境里解决了什么"。② 对比 AI：LLM 接近"通用计算"，人脑是"模块拼装"——这恰恰解释了为何人类有系统性偏差而强通用模型没有，也提示人机协同的真正价值在于互补，而非让人去模仿通用计算。③ 育儿：孩子怕黑、怕陌生大型动物，远比怕汽车、怕插座来得"自然"——因为前者是祖先环境里的真实威胁。理解这点，就不会用"讲道理"去对抗一个根本不在讲道理层面的恐惧回路。
English Summary
Adaptationist Perspective — the mind is not a blank slate or a general-purpose CPU, but a bundle of domain-specific modules (face recognition, cheater detection, mate choice, predator avoidance), each an "adaptation" shaped by selection to solve a recurrent ancestral problem. Key moves: ask the functional question ("what problem did this evolve to solve?") and reverse-engineer the mechanism; separate proximate causes (blood sugar) from ultimate causes (reproductive success). Crucial caveat: not everything is an adaptation — some traits are byproducts (spandrels) or drift. Avoid "just-so stories"; generate falsifiable functional hypotheses instead. The Wason selection task shows it vividly: people fail abstract logic but excel at the identical logic dressed as a social contract — the cheater-detection module switches on.
AI Prompts
中文提示词
我观察到一个看似非理性的行为/情绪：[描述行为]。请用适应器视角分析：
① 提出 2 个可证伪的"功能假说"——这个反应在祖先环境里可能解决了什么问题？
② 区分它的"近因"（当下触发的生理/心理机制）和"远因"（演化上的功能）；
③ 警惕"just-so story"：指出哪个假说证据更薄弱，以及它可能其实是副产品而非适应器。
English Prompt
I've observed a seemingly irrational behavior/emotion: [describe]. Analyze it through the adaptationist lens:
1. Propose 2 falsifiable "functional hypotheses" — what ancestral problem might this have solved?
2. Distinguish its proximate cause (the immediate physiological/psychological trigger) from its ultimate cause (evolutionary function).
3. Guard against just-so stories: flag which hypothesis is weakest, and whether it might be a byproduct rather than a true adaptation.
① 这在你眼里其实就是机器学习里的分布漂移（distribution shift）：模型在训练分布（祖先环境）上表现优异，部署到新分布（现代环境）上性能崩塌——人脑就是一个在旧分布上训练、却被迫在新分布上运行的模型。② 与多巴胺奖赏预测误差（见 Day 35）直接咬合：超常刺激不断制造"超预期奖赏"，把奖赏回路调校到现实永远满足不了的基线。③ 育儿最实用：孩子对短视频/游戏的"上瘾"，本质是超常刺激劫持了为真实社交与探索而生的回路。对策不是天天和意志力拔河，而是设计环境——默认不可得、提高摩擦、用真实的高质量刺激（运动、面对面玩耍）去占位。改环境，远比改孩子省力。
English Summary
Evolutionary Mismatch — adaptations were built for the ancestral environment, but culture and technology evolve far faster than genes. Circuits that were once adaptive now systematically misfire; the bug isn't in us, it's in the speed gap. The key mechanism is the supernormal stimulus: exaggerated versions of ancestral cues that hijack reward circuits — junk food (concentrated sugar/fat), social-media likes (digitized reputation), short-form video. Mismatch is specific, not "everything modern is bad." The cure is to design your environment, not fight willpower: change the stimuli you're exposed to rather than battling a circuit engineered to hijack you. Engineering analogue: distribution shift — a model trained on one distribution failing on another.
AI Prompts
中文提示词
我有一个想戒掉/想理解的现代习惯：[描述习惯，如刷短视频/暴食/查手机]。请用进化失配框架分析：
① 这个习惯劫持的是哪个为祖先环境而生的回路？它原本要解决什么问题？
② 指出其中的"超常刺激"——它把哪个真实信号做了夸张放大？
③ 给出 3 个"改环境而非靠意志力"的具体干预（提高摩擦/默认不可得/真实刺激占位）。
English Prompt
I have a modern habit I want to understand/quit: [describe, e.g. doomscrolling/overeating/phone-checking]. Analyze it via evolutionary mismatch:
1. Which ancestral circuit does this hijack, and what problem did that circuit originally solve?
2. Identify the "supernormal stimulus" — which real signal is being exaggerated?
3. Give 3 concrete "design the environment, not willpower" interventions (add friction / make it the non-default / crowd it out with a real stimulus).
① 代价高昂的信号在现代职场无处不在：一个技术上"过度精雕"的开源副项目，本质是一条孔雀尾巴——它昂贵、难伪造，向雇主诚实地广播"我有余力做到这个程度"。看懂这点，你就能有意识地选择信号：把精力投在难以伪造、目标受众真正在看的地方，而非廉价的自我标榜。② 炫耀性消费、刷存在感式的过度加班，都是同一逻辑——理解其信号功能，才能不被卷入失控循环。③ 这也是一个典型的正反馈系统（呼应你熟悉的反馈循环）：Fisherian runaway 与社交媒体上的"地位竞赛"结构同构，都会自我加速到失去理性边界。
English Summary
Sexual Selection — Darwin's "second selection": not optimizing for survival but for reproduction, and the two often conflict. The peacock's tail hurts survival yet boosts mating success — selection optimizes gene propagation, not individual welfare. Two modes: intrasexual competition (combat) and intersexual choice (display). The key idea is the handicap principle: a signal is honest precisely because it is costly — only genuinely fit individuals can afford the burden, so unfakeable handicaps carry real information. Fisherian runaway is a positive-feedback loop in which preference and trait co-amplify to absurd extremes. Lens for humans: much "wasteful" or exaggerated behavior reveals its function once seen as costly signaling to a target audience.
AI Prompts
中文提示词
我想理解/优化某个"信号"行为：[描述场景，如个人品牌/作品集/职场表现]。请用性选择与代价高昂信号分析：
① 我当前发出的信号里，哪些是"廉价可伪造"的、哪些是"昂贵难伪造"的？后者才有信息量；
② 我的目标"受众"真正在看的是什么信号？我有没有在错的地方花力气？
③ 是否存在"失控循环"风险（我在和别人比拼一个无止境放大的指标）？如何退出。
English Prompt
I want to understand/optimize a "signaling" behavior: [describe, e.g. personal brand / portfolio / workplace performance]. Analyze via sexual selection & costly signaling:
1. Among the signals I currently send, which are "cheap and fakeable" vs "costly and hard to fake"? Only the latter carry information.
2. What signal is my target "audience" actually watching? Am I spending effort in the wrong place?
3. Is there a "runaway loop" risk (am I in an ever-escalating status contest)? How do I exit it?
亲缘选择 · Kin Selection
"我愿为两个兄弟或八个表亲献出生命。" — 用基因之眼看利他
中文详解
为什么生物会为亲属牺牲？汉密尔顿法则给出冷峻的答案：当 rB > C 时，利他基因就会扩散——r 是亲缘关系系数，B 是受益者得到的繁殖收益，C 是利他者付出的代价。换句话说，帮亲属传播的，正是你自己也携带的那份基因。
① 基因之眼视角对你这样的分布式背景格外亲切：把基因看作自私的"数据副本"，生物体只是复制和传播它的节点——这与分布式系统里"数据为中心、节点为载具"的视角惊人地同构。② 现代最实用的推论是看穿拟亲触发：公司用"我们是一家人"、品牌用"家人们"来调用你的亲缘忠诚回路，本质是在非亲属身上借用 r 值。识破它，才能分清哪些忠诚值得给、哪些是被劫持。③ 育儿：父母对孩子近乎无条件的巨大投入（r = 0.5）正是亲缘选择最强烈的表达——理解这点不会让爱贬值，反而提醒我们：这份回路如此强大，值得被有意识地、也向血缘之外的人慷慨地扩展。
English Summary
Kin Selection — Hamilton's rule: an altruistic gene spreads when rB > C, where r is relatedness, B the recipient's reproductive benefit, C the altruist's cost. Helping relatives propagates the very genes you carry. This requires shifting the unit of selection from the organism to the gene (the "gene's-eye view"): the organism is a vehicle genes build to copy themselves. It explains family bonds — and their misfires: ancestrally, nearby people were usually kin, so the brain approximates relatedness via cues like familiarity, similarity, and co-residence. Modern groups (shared language, nationality, a flag) hijack these cues into fierce loyalty toward "fictive kin" — the root of tribalism. Demystifying altruism doesn't cheapen it; understanding the circuit lets us consciously extend it beyond blood.
AI Prompts
中文提示词
我在分析一段忠诚/利他/牺牲行为：[描述场景，如对公司/团队/某群体的投入]。请用亲缘选择与基因之眼视角分析：
① 这里的"利他"对象是真亲属，还是被"拟亲线索"（共同语言/身份/旗帜）激活的非亲属？
② 套用 rB > C 的直觉：我付出的代价 C 与对方收益 B 是否值得？谁在从这份忠诚里获益？
③ 哪些忠诚是我想有意识保留并扩展的，哪些是被他人劫持的"拟亲触发"，该如何抽身。
English Prompt
I'm analyzing an act of loyalty/altruism/sacrifice: [describe, e.g. devotion to a company/team/group]. Analyze via kin selection & the gene's-eye view:
1. Is the "recipient" actual kin, or a non-relative activated by "fictive-kin cues" (shared language/identity/flag)?
2. Apply the rB > C intuition: is my cost C worth the recipient's benefit B? Who actually profits from this loyalty?
3. Which loyalties do I want to consciously keep and extend, and which are hijacked fictive-kin triggers I should step back from?