Researchers developed a theoretical model showing how attention to unusually positive or negative experiences can systematically distort long-term beliefs about probability. The model suggests that rare events may remain overweighted even after extensive learning, while ordinary observations receive less influence.