Agent Studies the System
- How to have your agent systematically study documentation
- What the agent should extract from each document
- How to verify the agent has internalized the methodology
- Give your agent the study prompt
- Wait while it reads the key documents
- Review its summary to verify understanding
This Is the Key Step
This is where the magic happens. By having your agent systematically study the Brenner documentation, it builds a working understanding of the methodology. It won't just follow instructions — it will think with the cognitive operators.
The Study Prompt
Copy and paste this prompt to your agent. It will read the key documents and build its understanding:
What to Expect
Your agent will spend a few minutes reading the documents. When it responds, its summary should demonstrate understanding of:
Level Split (Σ)
Breaking problems into appropriate levels of analysis
Exclusion Test (⊘)
Designing experiments that can falsify hypotheses
Object Transpose (⟳)
Considering reversed causation and third variables
Scale Check (⊙)
Verifying effect sizes make physical/biological sense
Verification Questions
If the agent's summary seems shallow, ask these follow-up questions:
“What makes an experiment 'discriminative' rather than 'confirmation-seeking'?”
“When should I use Level Split vs Object Transpose?”
“What is a 'third alternative' and why is it important?”
“How do I know if my hypothesis is at the right level of abstraction?”
Success Criteria
You're ready to proceed when your agent can:
- Explain each of the four operators in its own words
- Distinguish discriminative experiments from confirmation-seeking
- Explain why “third alternatives” matter
- Reference specific concepts from the Brenner corpus
Next up: In Step 5, you'll define your research problem and have the agent help refine it using Brenner's criteria for good questions.