Data from: The neural basis of risky choice with affective outcomes

Both normative and many descriptive theories of decision making under risk are based on the notion that outcomes are weighted by their probability, with subsequent maximization of the (subjective) expected outcome. Numerous investigations from psychology, economics, and neuroscience have produced evidence consistent with this notion. However, this research has typically investigated choices involving relatively affect-poor, monetary outcomes. We compared choice in relatively affect-poor, monetary lottery problems with choice in relatively affect-rich medical decision problems. Computational modeling of behavioral data and model-based neuroimaging analyses provide converging evidence for substantial differences in the respective decision mechanisms. Relative to affect-poor choices, affect-rich choices yielded a more strongly curved probability weighting function of cumulative prospect theory, thus signaling that the psychological impact of probabilities is strongly diminished for affect-rich outcomes. Examining task-dependent brain activation, we identified a region-by-condition interaction indicating qualitative differences of activation between affect-rich and affect-poor choices. Moreover, brain activation in regions that were more active during affect-poor choices (e.g., the supramarginal gyrus) correlated with individual trial-by-trial decision weights, indicating that these regions reflect processing of probabilities. Formal reverse inference Neurosynth meta-analyses suggested that whereas affect-poor choices seem to be based on brain mechanisms for calculative processes, affect-rich choices are driven by the representation of outcomes’ emotional value and autobiographical memories associated with them. These results provide evidence that the traditional notion of expectation maximization may not apply in the context of outcomes laden with affective responses, and that understanding the brain mechanisms of decision making requires the domain of the decision to be taken into account.

风险决策领域的规范性理论与诸多描述性理论均基于这一核心假设：决策结果会依据其发生概率进行加权，并在此基础上最大化（主观）期望结果。心理学、经济学与神经科学领域的大量研究均为这一假设提供了一致性证据支持。然而此类研究通常仅考察涉及低情感负载（affect-poor）金钱结果的决策选择。本研究将低情感负载（affect-poor）金钱彩票决策任务与高情感负载（affect-rich）医疗决策任务中的选择行为进行了对比分析。行为数据的计算建模与基于模型的神经影像学分析，均提供了一致性证据，表明两类决策机制存在显著差异。相较于低情感负载决策，高情感负载决策的累积前景理论（cumulative prospect theory）概率加权函数曲率更强，这表明对于高情感负载的结果而言，概率的心理影响被显著削弱。通过分析任务依赖性脑激活模式，我们发现了脑区-任务条件交互效应，表明高、低情感负载决策对应的脑激活模式存在质性差异。此外，低情感负载决策中激活更强的脑区（如缘上回（supramarginal gyrus）），其激活水平与个体逐试次的决策权重呈显著相关，表明这些脑区参与了概率加工过程。基于正式反向推断的Neurosynth元分析结果显示：低情感负载决策似乎基于计算加工的脑机制，而高情感负载决策则由结果的情感价值表征及其相关的自传体记忆所驱动。本研究结果表明，传统的期望最大化假设或许并不适用于伴随强烈情感反应的决策场景，同时也提示，要完整理解决策的脑机制，必须将决策所属的领域纳入考量范畴。