People Behave as if they Anticipate Regret Conditional on Experiencing a Bad Outcome

Decision-makers often must decide whether to invest in prospects to reduce risk. Existing models assume that decision-makers consider the absolute improvement in probabilistic chances (e.g., increasing a 10% chance of winning 10 dollars to a 20% chance is roughly similar to increasing an 80% chance of winning 10 dollars to a 90% chance). We present evidence that people instead behave as if they consider the relative reduction in bad outcomes (increasing a 10% chance to a 20% chance eliminates 1/9th of all bad outcomes, while increasing an 80% chance to a 90% chance eliminates 1/2 of all bad outcomes). This bias in the anticipation of preventable bad outcomes drives risk preferences that violate normative standards and results in the same individuals behaving both risk-seeking and risk-averse within the same decision-making task. We discuss how to adjust regret theory to accommodate these results.