all three interventions worked primarily by reducing noise. As the researchers put it, “Whenever an intervention boosted accuracy, it worked mainly by suppressing random errors in judgment. Curiously, the original intent of the training intervention was to reduce bias.”
