When System 1 cannot answer a hard question, it answers an easier one instead — and does not announce the swap. Kahneman calls this substitution: the heuristic answer is not necessarily simpler, only more accessible.
Examples
- "How happy are you with your life?" → "How is my dating life right now?" (students asked about dates first)
- "What will Julie's college GPA be?" → "How impressive is early reading?" (same percentile assigned to both)
- "Is this startup likely to succeed?" → "How good was the demo?"
The substituted answer is often directionally useful. It fails when the target question requires uncertainty adjustment — intuitive predictions stay as extreme as the evidence because evaluation is substituted for forecasting, ignoring regression-to-the-mean.
Substitution is the mechanism behind availability (how easily examples come to mind substitutes for frequency) and representativeness (similarity to a stereotype substitutes for probability).