Interdisciplinary Research

Interdisciplinary Research Skill

Techniques for finding connections across domains and generating novel hypotheses.

Analogical Reasoning Protocol

When encountering a problem in domain A:

1. Abstract the structure

   • What are the key entities?
   • What are the relationships?
   • What is being optimized/conserved?
   • What are the constraints?

2. Search for structural parallels

   • Where else do similar structures appear?
   • What fields have solved analogous problems?
   • What metaphors capture the essence?

3. Map the analogy carefully

   Source Domain	Target Domain
   Entity A	Entity X
   Relationship R	Relationship Y
   Constraint C	Constraint Z

4. Test the mapping

   • Where does the analogy break down?
   • What does the source predict about the target?
   • Are the breakdowns informative?

Cross-Domain Connection Patterns

Shared Mathematical Structure

Many domains share underlying mathematics:

Analogous Processes

Hypothesis Generation Techniques

Negation

Take an assumption and ask: what if the opposite were true?

• "Neurons compute" -> "Neurons don't compute, they filter"
• "More data is better" -> "More data is harmful after some point"

Extreme Cases

Push parameters to limits:

• What if this process were infinitely fast?
• What if there were zero cost?
• What if the population were infinite?

Mechanism Swap

Take the mechanism from field A, apply to field B:

• "What if immune systems used gradient descent?"
• "What if markets had error-correcting codes?"
• "What if memory worked like version control?"

Missing Middle

If A causes C, what is B?

• Identify known correlations
• Generate candidate mechanisms
• Design experiments to distinguish

Synthesis Protocol

When integrating insights from multiple domains:

1. Gather perspectives

   • What does each field say about this?
   • What terminology do they use?
   • What do they consider solved/open?

2. Identify tensions

   • Where do fields disagree?
   • Are disagreements real or terminological?
   • What would resolve the tension?

3. Find the deeper pattern

   • What is common across all perspectives?
   • What level of abstraction reveals unity?
   • Is there a framework that contains all as special cases?

4. Generate novel predictions

   • What does the synthesis predict that no single field predicts?
   • Where should we look for evidence?
   • What experiments would test the synthesis?

Question Prompts for Connection-Finding

When stuck, ask:

• "What does this remind me of from [other field]?"
• "Who else has solved a problem like this?"
• "What if we approached this like a [physicist/biologist/economist] would?"
• "What is the dual of this concept?"
• "What would the inverse problem look like?"
• "If I had to explain this to someone from [field], what would I say?"

Reading Across Boundaries

Efficient Cross-Domain Learning

1. Start with review papers - Get the landscape fast
2. Learn the vocabulary - Same words often mean different things
3. Find the canonical papers - Every field has its founding texts
4. Identify the open problems - This is where you might contribute
5. Talk to practitioners - Written sources miss tacit knowledge

Warning Signs of Bad Analogies

• Surface similarity only (both have "networks" does not mean they work the same)
• Ignoring scale differences (what works for 10 might not work for 10 million)
• Category errors (treating metaphors as mechanisms)
• Confirmation bias (only seeing what supports the mapping)

Documenting Connections

When you find a promising connection, capture:

## Connection: [Source] <-> [Target]

### The Analogy
[What maps to what]

### Why It Might Work
[Structural similarities, shared mathematics]

### Where It Breaks Down
[Known limits of the analogy]

### Predictions
[What this connection suggests we should observe]

### Tests
[How we could validate or falsify]

### References
[Key papers from both domains]