Interactive Paradoxes

Psychological Effects

Not mathematical paradoxes, but cognitive and memory biases that distort how we judge track records and evidence.

The Jeane Dixon Effect

The Jeane Dixon Effect is named for a mid-20th-century American psychic whose occasional lucky-sounding guesses stuck in people's minds long after her many wrong guesses were forgotten. It describes a common habit of mind. People (and the media covering them) tend to remember the predictions that came true much better than the ones that didn't. The predictions themselves never change, but this kind of selective memory alone can make a track record look far more impressive in hindsight than it really was. Adjust the sliders below to see how big a gap selective memory can open up between what actually happened and what people remember happening.

Change the underlying record

These controls describe what actually happens, before memory or reporting enters the picture.

Change the memory filter

A hit and a non-hit don't have to have the same chance of being remembered, repeated, or retold.

Optional: reinterpreting after the fact

This is separate from selective memory. It's about remembered non-hits later getting relabeled as hits because the original claim was vague or open to interpretation.

Actual success rate
11%
Apparent remembered rate
23%
The remembered record makes an 11% hit rate look like 23%.
What actually happened 11 hits · 89 non-hits
Hits Non-hits
selective memory / reporting
What remains in the remembered record 6.6 hits · 22 non-hits
Remembered as hits Remembered as non-hits
6.6
hits surviving
22
non-hits surviving
71
events forgotten
What this visualization shows

The actual hit rate never changes here. What changes is only how likely hits and non-hits are to survive into the remembered record. When hits are much easier to remember than non-hits, the record people recall can end up looking dramatically better than the full, honest record.

Why "non-hit" instead of "miss"?

The large study used to set the default hit rate didn't just sort every unsuccessful prediction into "wrong." Some predictions were too vague to judge one way or the other, some were obvious or expected, and some were impossible to check at all.

In the Great Australian Psychic Prediction Project, the published predictions broke down as:

  • 11% correct
  • 53% wrong
  • 15% expected or obvious
  • 19% too vague
  • 2% unknown

This visualization uses non-hit for the other 89% because of that mix. Lumping vague, obvious, or uncheckable statements in with definite misses would be misleading, so "non-hit" just means anything that wasn't a clear, scored correct prediction.

Further reading: Jeane Dixon effect on Wikipedia · The Great Australian Psychic Prediction Project.