Losing Fast or Slow? Preferences for Uncertainty Resolution