
Note sull'episodio
According to the uncorrected output of an fMRI scanner, a dead, ice-packed Atlantic salmon was actively reading human emotions. This week, we dive into Craig Bennett’s legendary 2009 Ig Nobel-winning study to explore how running 140,000 simultaneous statistical tests on pure thermal background noise can manufacture phantom brain activity out of thin air. We break down the Multiple Comparisons Problem, spatial autocorrelation, and why modern high-dimensional data analytics needs a lot more than a prayer to separate true signals from dead fish.
Parole chiave
science
statistics
popular science
publishing
academic publishing
neuroscience
fMRI
