Correspondence Bias Reversal? — LessWrong