# Philosophy Of Science

> Master philosophy of science - scientific method, explanation, realism, theory change. Use for: scientific methodology, explanation, realism/anti-realism, paradigms. Triggers: 'scientific method', 'falsification', 'Popper', 'Kuhn', 'paradigm', 'scientific explanation', 'scientific realism', 'instrumentalism', 'theory change', 'confirmation', 'induction problem', 'underdetermination', 'demarcation', 'reduction', 'emergence'.

- Skill: `majiayu000/philosophy-of-science` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/philosophy-of-science`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/philosophy-of-science/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/philosophy-of-science

---


# Philosophy of Science Skill

Master the philosophical foundations of science: What is scientific method? What is explanation? Are scientific theories true?

## Core Questions

| Question | Issue |
|----------|-------|
| What distinguishes science from non-science? | Demarcation |
| How do we confirm theories? | Confirmation |
| What is scientific explanation? | Explanation |
| Are theories true or useful fictions? | Realism |
| How does science change? | Theory change |

---

## Scientific Method

### The Problem of Induction

**Hume's Problem**: How do we justify inductive inference?
- Past regularities don't logically guarantee future ones
- Cannot use induction to justify induction (circular)
- Yet science relies on induction

### Falsificationism (Popper)

```
POPPER'S FALSIFICATIONISM
═════════════════════════

DEMARCATION CRITERION
├── Science: Falsifiable claims
├── Pseudo-science: Unfalsifiable
└── Examples: Astrology, Freud (unfalsifiable)

METHOD
├── Bold conjectures
├── Severe tests
├── Refutation → new conjecture
└── Corroboration ≠ confirmation

KEY IDEA:
We never confirm theories
We only fail to falsify them
Asymmetry: One counterexample refutes
```

**Problems**:
- Theories rarely abandoned on single refutation
- Auxiliary hypotheses can absorb refutation
- No purely observational test

### Paradigms (Kuhn)

```
KUHN'S STRUCTURE OF SCIENTIFIC REVOLUTIONS
══════════════════════════════════════════

NORMAL SCIENCE
├── Work within paradigm
├── Puzzle-solving
├── Anomalies accumulate
└── Paradigm defines problems, methods

CRISIS
├── Too many anomalies
├── Alternative paradigms emerge
├── Debate between paradigms
└── Incommensurability

SCIENTIFIC REVOLUTION
├── Paradigm shift
├── Not cumulative progress
├── New worldview
└── Gestalt switch

EXAMPLES:
├── Ptolemy → Copernicus
├── Newton → Einstein
└── Phlogiston → Oxygen
```

---

## Scientific Explanation

### Deductive-Nomological (D-N) Model

```
D-N MODEL (Hempel)
══════════════════

EXPLANATION STRUCTURE:
L₁, L₂, ... Lₙ  (Laws)
C₁, C₂, ... Cₙ  (Conditions)
─────────────────
E              (Explanandum)

REQUIREMENTS:
├── Deductively valid
├── Laws are essential
├── Empirically testable
└── True premises

EXAMPLE:
All metals expand when heated.
This is metal.
This was heated.
∴ This expanded.
```

**Problems**:
- Symmetry problem (flagpole and shadow)
- Irrelevance problem
- Statistical explanation

### Causal-Mechanical Model

- Explanation = tracing causal mechanism
- Not just subsumption under laws
- Mechanisms explain, not just correlate

### Unificationism

- Explanation = unifying diverse phenomena
- Fewer patterns explaining more
- Newton unified celestial and terrestrial motion

---

## Scientific Realism

### The Debate

**Scientific Realism**:
- Mature scientific theories are approximately true
- Theoretical entities (electrons, genes) exist
- Science aims at truth

**Anti-Realism (Instrumentalism)**:
- Theories are useful tools
- Theoretical terms don't refer
- Science aims at empirical adequacy

### Arguments for Realism

**No Miracles Argument**:
- Science's success would be miraculous if theories weren't true
- Best explanation of predictive success is truth
- "The only philosophy that doesn't make science a miracle"

### Arguments Against Realism

**Pessimistic Meta-Induction**:
- Past "successful" theories were false
- Caloric, phlogiston, ether
- Current theories probably also false

**Underdetermination**:
- Multiple theories compatible with same evidence
- Evidence doesn't uniquely determine theory
- Why think ours is true?

### Structural Realism

**Epistemic**: We can know structure, not nature
**Ontic**: Structure is all there is

---

## Reduction and Emergence

### Reductionism

- Higher-level sciences reducible to lower
- Biology → Chemistry → Physics
- Unity of science thesis

### Emergence

- Some properties not reducible
- Whole greater than parts
- Consciousness? Life?

### Multiple Realizability

- Same higher-level state, different lower-level realizations
- Pain in humans ≠ pain in octopi (neurally)
- Blocks type-identity reduction

---

## Key Debates

### Demarcation

- What makes something science?
- Falsifiability? Paradigms? Method?
- Is demarcation possible?

### Theory Choice

- Empirical adequacy
- Simplicity, parsimony
- Explanatory power
- Fruitfulness

### Values in Science

- Value-free ideal achievable?
- Social influences on science
- Science studies, feminist philosophy of science

---

## Key Vocabulary

| Term | Meaning |
|------|---------|
| Falsification | Disproving through counterevidence |
| Paradigm | Shared framework for research |
| Incommensurability | Paradigms can't be compared |
| Confirmation | Evidence supporting theory |
| Underdetermination | Evidence doesn't determine theory |
| Instrumentalism | Theories are tools, not truths |
| Reduction | Higher explained by lower |
| Emergence | Irreducible higher-level properties |
| Demarcation | Distinguishing science from non-science |
| Corroboration | Surviving falsification attempts |

---

## Integration with Repository

### Related Themes
- `thoughts/knowledge/`: Scientific knowledge
- `thoughts/consciousness/`: Neuroscience methodology