Variability in Decision Strategies Across Description-based and Experience-based Decision Making

摘要

Individuals are known to make systematically different decisions when the probabilities in risky choice problems are described or experienced. This difference, known as the description-experience gap, has been reliably replicated across several studies using binary choice gambles. Yet little is known whether these differences exist in more complex gambles in the absence of rare outcomes, and whether they are associated with systematic differences in the use of decision heuristics and strategies across formats. Using three-outcome mixed gambles, we found that participants showed a strong preference for alternatives that maximized the overall probability of winning when such an option was available in the description condition, and chose more randomly otherwise. In the experience condition, preferences were more homogenous across trials types, with participants choosing the alternative with extreme values more often relative to the description condition. However, when we controlled for the experienced outcomes, both natural mean heuristic (choosing the alternative with highest sampled mean or expected value) and overall probability of winning heuristic reliably predicted choice on each trial. In fact, expected value was the strongest predictor of preferences in a conditional logistic regression model that included extreme values, expected value, and overall probability of winning variables simultaneously. Yet expected value did not predict preferences in decisions from description. Together, these findings provide evidence for an explicit dissociation in decision strategies across description and experience formats. Copyright (c) 2017 John Wiley & Sons, Ltd.