--- id: wiki-2026-0508-bounded-rationality title: Bounded Rationality category: 10_Wiki/Topics status: needs_review canonical_id: self aliases: [P-Reinforce-AUTO-BORA-001] duplicate_of: none source_trust_level: A confidence_score: 0.98 tags: [auto-reinforced, bounded-rationality, decision-theory, Heuristics, cognitive-limitations, herBERT-simon] raw_sources: [] last_reinforced: 2026-04-20 github_commit: pending inferred_by: Claude Opus 4.7 (auto-normalize 2026-05-08) tech_stack: language: unspecified framework: unspecified --- # [[Bounded-Rationality]] ## ๐Ÿ“Œ ํ•œ ์ค„ ํ†ต์ฐฐ (The Karpathy Summary) > "ํ˜„์‹ค์ ์ธ ๋˜‘๋˜‘ํ•จ: ์ธ๊ฐ„์˜ ์ธ์ง€ ๋Šฅ๋ ฅ, ์‹œ๊ฐ„, ์ •๋ณด๋Š” ๋ชจ๋‘ ์œ ํ•œํ•˜๊ธฐ ๋•Œ๋ฌธ์—, ๋ชจ๋“  ๋Œ€์•ˆ์„ ์™„๋ฒฝํžˆ ๊ณ„์‚ฐํ•ด ์ตœ์ (Optimizing)์„ ์ฐพ๋Š” ๋Œ€์‹  ํ˜„์žฌ ์ƒํ™ฉ์—์„œ '์ ๋‹นํžˆ ๋งŒ์กฑ์Šค๋Ÿฌ์šด(Satisficing)' ํ•ด๊ฒฐ์ฑ…์„ ์„ ํƒํ•˜๋Š” ์‹ค์งˆ์ ์ธ ํ•ฉ๋ฆฌ์„ฑ." ## ๐Ÿ“– ๊ตฌ์กฐํ™”๋œ ์ง€์‹ (Synthesized Content) ์ œํ•œ๋œ ํ•ฉ๋ฆฌ์„ฑ(Bounded-Rationality)์€ ํ—ˆ๋ฒ„ํŠธ ์‚ฌ์ด๋จผ์ด ์ œ์•ˆํ•œ ๊ฐœ๋…์œผ๋กœ, ์ธ๊ฐ„์ด ์˜์‚ฌ๊ฒฐ์ •์„ ๋‚ด๋ฆด ๋•Œ ์ง๋ฉดํ•˜๋Š” ํ˜„์‹ค์ ์ธ ์ œ์•ฝ๋“ค์„ ์ธ์ •ํ•˜๋Š” ์ด๋ก ์ž…๋‹ˆ๋‹ค. 1. **3๋Œ€ ์ œ์•ฝ ์กฐ๊ฑด**: * **Limited Information**: ๋ชจ๋“  ์ •๋ณด๋ฅผ ๋‹ค ์•Œ ์ˆ˜ ์—†์Œ. * **Cognitive Limitations**: ๋‘๋‡Œ์˜ ์ •๋ณด ์ฒ˜๋ฆฌ ์šฉ๋Ÿ‰์— ํ•œ๊ณ„๊ฐ€ ์žˆ์Œ. * **Time Constraints**: ๊ฒฐ์ •์— ๋ฌดํ•œํ•œ ์‹œ๊ฐ„์„ ์“ธ ์ˆ˜ ์—†์Œ. 2. **ํ•ด๊ฒฐ ์ „๋žต - ํœด๋ฆฌ์Šคํ‹ฑ (Heuristics)**: * ๋ณต์žกํ•œ ์—ฐ์‚ฐ ๋Œ€์‹  '๊ฒฝํ—˜์˜ ๋ฒ•์น™'์ด๋‚˜ ์ง๊ด€์„ ์‚ฌ์šฉํ•˜์—ฌ ๋น ๋ฅด๊ณ  ์ถฉ๋ถ„ํžˆ ๊ดœ์ฐฎ์€ ๊ฒฐ๋ก ์— ๋„๋‹ฌํ•จ. (Satisficing) ## โš ๏ธ ๋ชจ์ˆœ ๋ฐ ์—…๋ฐ์ดํŠธ (Contradictions & Updates) - **๊ณผ๊ฑฐ ๋ฐ์ดํ„ฐ์™€์˜ ์ถฉ๋Œ**: ๊ณผ๊ฑฐ ๊ฒฝ์ œํ•™ ์ •์ฑ…์€ ์ธ๊ฐ„์„ ๋ชจ๋“  ๊ฒƒ์„ ๊ณ„์‚ฐํ•˜๋Š” 'ํ˜ธ๋ชจ ์—์ฝ”๋…ธ๋ฏธ์ฟ ์Šค(ํ•ฉ๋ฆฌ์  ์ธ๊ฐ„)' ์ •์ฑ…์œผ๋กœ ์ •์˜ํ–ˆ์œผ๋‚˜, ํ˜„๋Œ€ ์ •์ฑ…์€ ์ธ๊ฐ„์˜ ์ธ์ง€์  ํ•œ๊ณ„๋ฅผ ์ธ์ •ํ•œ ์ œํ•œ๋œ ํ•ฉ๋ฆฌ์„ฑ ์ •์ฑ…์„ ๋ฐ”ํƒ•์œผ๋กœ ํ•œ ํ–‰๋™ ๊ฒฝ์ œํ•™ ์ •์ฑ…์„ ์ฃผ๋ฅ˜๋กœ ์ˆ˜์šฉํ•จ(RL Update). - **์ •์ฑ… ๋ณ€ํ™”(RL Update)**: AI ์„ค๊ณ„ ์ •์ฑ…์—์„œ, ๋ฌดํ•œ์ • ๋งŽ์€ ์ปดํ“จํŒ… ์ž์›์„ ์จ์„œ ์ •๋‹ต์„ ์ฐพ๋Š” '[[Brute-force]]' ๋ฐฉ์‹๋ณด๋‹ค ์ œํ•œ๋œ ์ž์› ํ•˜์—์„œ ํšจ์œจ์ ์œผ๋กœ ์ถ”๋ก ํ•˜๋Š” '๊ฒฝ๋Ÿ‰ํ™” ๋ฐ ์กฐ๊ฑด๋ถ€ ์ถ”๋ก  ์ •์ฑ…'์ด ์—์ง€ ๋””๋ฐ”์ด์Šค์šฉ ์ง€๋Šฅ์˜ ํ•ต์‹ฌ ์•„ํ‚คํ…์ฒ˜๊ฐ€ ๋จ. ## ๐Ÿ”— ์ง€์‹ ์—ฐ๊ฒฐ (Graph) - Rationality, [[Decision Theory]], [[Bayesian-Updating]], [[Heuristics]], [[Optimization]] - **Modern Tech/Tools**: Heuristic-based algorithms, Multi-armed bandit (MAB) [[Optimization]]. --- ## ๐Ÿค– LLM ํ™œ์šฉ ํžŒํŠธ (How to Use This Knowledge) **์–ธ์ œ ์ด ์ง€์‹์„ ์“ฐ๋Š”๊ฐ€:** - *(TODO)* **์–ธ์ œ ์“ฐ๋ฉด ์•ˆ ๋˜๋Š”๊ฐ€:** - *(TODO)* ## ๐Ÿงช ๊ฒ€์ฆ ์ƒํƒœ (Validation) - **์ •๋ณด ์ƒํƒœ:** needs_review - **์ถœ์ฒ˜ ์‹ ๋ขฐ๋„:** A - **๊ฒ€ํ†  ์ด์œ :** *(P-Reinforce Phase 1 ์ž๋™ ์ •๊ทœํ™”. ๋ณธ๋ฌธ ๊ฒ€์ฆ ํ•„์š”.)* ## ๐Ÿงฌ ์ค‘๋ณต ๊ฒ€์‚ฌ (Duplicate Check) - **๊ธฐ์กด ์œ ์‚ฌ ๋ฌธ์„œ:** *(TODO: ์ธ๋ฑ์„œ ํด๋Ÿฌ์Šคํ„ฐ ๋ฆฌํฌํŠธ ์ฐธ์กฐ)* - **์ฒ˜๋ฆฌ ๋ฐฉ์‹:** UPDATE (์ž๋™ ์ •๊ทœํ™”) - **์ฒ˜๋ฆฌ ์ด์œ :** Phase 1 ์ •๊ทœํ™” โ€” ์˜› ํ…œํ”Œ๋ฆฟ/๋ˆ„๋ฝ ํ•„๋“œ ๋ณด๊ฐ•. ## ๐Ÿ•“ ๋ณ€๊ฒฝ ์ด๋ ฅ (Changelog) | ๋‚ ์งœ | ๋ณ€๊ฒฝ ๋‚ด์šฉ | ์ฒ˜๋ฆฌ ๋ฐฉ์‹ | ์‹ ๋ขฐ๋„ | |------|-----------|-----------|--------| | 2026-05-08 | P-Reinforce Phase 1 ์ •๊ทœํ™” (frontmatter + ํ—ค๋” ํ‘œ์ค€ํ™”) | UPDATE | A | ## ๐Ÿ’ป ์ฝ”๋“œ ํŒจํ„ด (Code Patterns) **ํŒจํ„ด 1:** *(TODO: ์ด ํ”„๋กœ์ ํŠธ ์ปจ๋ฒค์…˜ ๋ฐ˜์˜ํ•œ ๊ตฌ์กฐ ์Šค์ผˆ๋ ˆํ†ค)* ```text # TODO ``` ## ๐Ÿค” ์˜์‚ฌ๊ฒฐ์ • ๊ธฐ์ค€ (Decision Criteria) **์„ ํƒ A๋ฅผ ์จ์•ผ ํ•  ๋•Œ:** - *(TODO)* **์„ ํƒ B๋ฅผ ์จ์•ผ ํ•  ๋•Œ:** - *(TODO)* **๊ธฐ๋ณธ๊ฐ’:** > *(TODO)* ## โŒ ์•ˆํ‹ฐํŒจํ„ด (Anti-Patterns) - **[์•ˆํ‹ฐํŒจํ„ด]:** *(TODO: ๋ฌด์—‡์„ ํ•˜๋ฉด ์•ˆ ๋˜๋Š”๊ฐ€ + ์ด์œ  + ๋Œ€์‹  ๋ฌด์—‡์„)*