Everything Hertz 171: The easiest person to fool is yourself (with Daniel Simons and Christopher Chabris)
12 snips
Jul 20, 2023 AI Snips
Chapters
Transcript
Episode notes
Scientists Lose Skepticism When Results Please Them
- Scientists often stop thinking like scientists when results confirm their expectations, reducing skepticism.
- Dan Simons notes that confirmed predictions feel pleasing or novel, which leads researchers to drop critical scrutiny and adopt non-scientific thinking.
Use A Possibility Grid To See What Success Really Means
- The possibility grid (a 2x2 contingency view) reveals why success stories mislead: you must consider successes with and without the action and failures with and without the action.
- Focusing only on positive cases (upper-left cell) ignores how many others did the same and failed, so cause and luck are confounded.
Pre-Register Edge Cases And Exclusions
- Pre-register detailed analysis rules and exclusions to reduce flexibility and post-hoc tinkering.
- Simons describes pre-specifying rules like excluding inattentive participants and logging those decisions before seeing outcomes.