yogsoth-ai Counterfactual Reasoning
Reason about what would happen if a variable were different. "If X had not occurred, would Y still have happened?" This is the gold standard for causal identification.
Available SOPs
- causal-chain-query — trace causal paths to identify dependencies
- contradiction-flagging — flag when counterfactual contradicts observed evidence
- confidence-scoring — assess confidence in counterfactual conclusions
Guiding Principles
- Minimal change. Counterfactuals should change only the target variable, holding all else constant.
- Mechanism required. A counterfactual without a mechanism is speculation, not reasoning.
- Multiple paths. If Y can be caused by X through multiple paths, removing X may not remove Y.
- Temporal ordering. Causes precede effects. Counterfactuals that violate temporal order are invalid.
Minimum Yield
Available SOPs
Optional, no fixed order; the final leaf is always a sop.
| SOP | When to use |
|---|---|
| causal-chain-query | SOP for tracing causal chains — follow edges from cause to effect through intermediate variables. |
| confidence-scoring | SOP for assigning calibrated confidence scores to causal claims based on evidence quality and quantity. |
| contradiction-flagging | SOP for flagging contradictions in the causal model — identify conflicting evidence or mechanism claims. |