Probabilistic Reversal Learning (Cools 2002, Hampshire 2008) — classic orbitofrontal/vmPFC test. Two circles: blue (A) and orange (B). One pays +10¢ with 80% probability, the other with 20%. After several correct in a row, rules SILENTLY swap — good became bad. You must detect from feedback and switch. Key biomarkers: reversal_errors (slow relearning), perseverative_errors (sticking to old rule despite punishment), win_stay_rate, lose_shift_rate. Direct correlate of financial decision-making: vmPFC updates value from feedback; PRL measures how fast. Third leg of financial combo (Iowa + BART + PRL).