deductive-hypothesis-generation
$
npx mdskill add yogsoth-ai/de-anthropocentric-research-engine/deductive-hypothesis-generation从现有理论演绎推导假设:在理论成熟领域,通过显式推理链将理论命题转化为具体的可测试预测。
SKILL.md
.github/skills/deductive-hypothesis-generationView on GitHub ↗
---
name: deductive-hypothesis-generation
description: "Strategy: 从现有理论演绎推导假设"
version: 1.0.0
category: hypothesis-formation
type: strategy
campaign: hypothesis-formulation
tactics:
- theory-mechanism-extraction
- falsifiability-audit
sops:
- theory-identification
- mechanism-extraction
- variable-identification
- relationship-specification
- boundary-condition-specification
- falsifiability-check
- operationalization
dependencies:
skills:
- context-management
- subagent-spawning
- literature-engine
---
# Deductive Hypothesis Generation
从现有理论演绎推导假设:在理论成熟领域,通过显式推理链将理论命题转化为具体的可测试预测。
## 适用场景
- 领域拥有成熟的基础理论(有 named theories、正式模型、或公认机制)
- 研究 gap 表现为"理论预测与现实观察之间的偏差"
- 目标是检验、扩展或限定已有理论的适用范围
- 需要高度可辩护的假设(审稿人会追问理论依据)
不适用:数据丰富但理论空白的新兴领域 → 改用 inductive-hypothesis-generation。
## 思维框架
**Theory → Mechanism → Variable Relationship → Testable Prediction**
演绎的核心逻辑:
1. **Theory**:识别支撑研究问题的基础理论(命名理论、正式模型)
2. **Mechanism**:从理论中提取因果机制("X 通过 Z 影响 Y"的中间过程)
3. **Variable Relationship**:将机制转化为变量间的方向性关系(正向/负向/调节/中介)
4. **Testable Prediction**:将变量关系具体化为在特定条件下的可观测预测
每一步都必须可追溯:每个 prediction 能回溯到机制,每个机制能回溯到理论。这是演绎假设区别于猜测的核心。
**常见陷阱**:
- 理论引用流于表面(只提名字,不提具体命题)→ 必须引用理论的核心命题
- 跳过机制直接从理论跳到预测 → 机制是演绎链的关键节点,不可省略
- 假设范围过广("在所有情境下")→ 演绎必须说明边界条件
## Budget Gate
| Tier | 理论覆盖 | 机制提取 | 假设产出 | 可证伪性 |
|------|---------|---------|---------|---------|
| S | ≥2 个具名理论 | ≥3 个因果机制 | ≥2 个结构化假设 | 每个假设 1 个 falsification scenario |
| M | ≥3 个具名理论 | ≥5 个因果机制 | ≥3 个结构化假设 | 每个假设 ≥1 scenario + boundary conditions |
| L | ≥5 个具名理论 | ≥8 个因果机制 | ≥5 个结构化假设 | 完整 falsifiability audit + 竞争理论比较 |
## 默认参考流
1. 调用 `theory-identification` SOP:扫描领域文献,列出与 gap 相关的具名理论及其核心命题
2. 调用 `mechanism-extraction` SOP(via `theory-mechanism-extraction` tactic):从每个理论中提取因果机制链
3. 调用 `variable-identification` SOP:将机制中的构念(constructs)转化为可操作变量
4. 调用 `relationship-specification` SOP:明确变量间方向性关系(含调节/中介结构)
5. 调用 `boundary-condition-specification` SOP:识别理论适用的前提条件(群体、情境、时间范围等)
6. 调用 `falsifiability-check` SOP(via `falsifiability-audit` tactic):为每个假设生成 falsification scenario
7. 调用 `operationalization` SOP:为关键变量提供测量方法草案
## context-checkpoint
每轮结束后记录:
- 已识别理论清单(名称、核心命题、来源)
- 提取的机制列表(每条机制标注来源理论)
- 当前假设草稿集(含变量关系 + 边界条件)
- 可证伪性状态(已通过 / 待审查 / 不可证伪需修改)
More from yogsoth-ai/de-anthropocentric-research-engine
- abductive-hypothesis-generationStrategy: 面对异常的最佳解释推理
- ablation-brainstormRemove components one by one, observe system changes to reveal hidden dependencies and generate ideas from structural gaps.
- ablation-component-mappingMap system architecture to ablatable units for ablation studies
- ablation-designDesign ablation studies to isolate component contributions in ML systems
- ablation-executionRemove components one by one from a system, record the response/impact of each removal.
- abp-vulnerability-classificationClassify assumptions on 2 axes — load-bearing (how much conclusion depends on it) × vulnerable (how likely to be false). Focuses attention on High-Load × High-Vulnerable quadrant.
- abstraction-extractionExtract abstract principles from concrete domain cases. Strips domain-specific details to reveal transferable mechanisms.
- abstraction-ladderPerform bisociation at multiple abstraction levels
- abstraction-ladderingMove between concrete and abstract framings — 3 levels up (Why?) and 3 levels down (How?) to find the most productive research level.
- abstraction-to-designAbstract biological principle to design principle. Bridge from biology to engineering.