摘要：老化伴随着广泛的神经生物学及认知衰退，例如奖励学习的下降(Eppinger, Schuck, Nystrom, & Cohen, 2013)，价值编码能力的降低(Halfmann, Hedgcock, Kable, & Denburg, 2016)，这些改变在许多风险决策任务中都影响了老年人的决策能力(Best & Charness, 2015; Mata, Josef, Samanez-Larkin, & Hertwig, 2011)。但是，不同的风险倾向测量方式是否会显示相似的年龄差异，这些老化效应背后的神经机制是什么？对此，我们进行了三项研究，以调查年轻人和老年人的风险决策倾向。在我们的第一项工作中，我们采用自我报告方式来衡量冒险倾向和风险规避倾向，并发现自我报告的冒险倾向随着年龄的增长而降低，而风险规避倾向在老年人和年轻人之间没有发现显著差异。在我们的第二个研究中，我们采用了没有学习成分的描述性决策——风险赌博范式(Rieskamp, 2008; Tymula, Rosenberg Belmaker, Ruderman, Glimcher, & Levy, 2013)来测量在包括肯定选择时（肯定的收益或肯定的损失），老年人和年轻人风险倾向的年龄差异。结果显示，在收益领域中，相同的预期值和更好的预期值两种不同的实验设定条件下，与年轻人相比，在相比冒险获得更大的收益，老年​​人都更倾向于肯定选择。因此在收益领域老年人表现出较低的风险倾向。然而，这种现象并没有在金钱损失领域中发现。在金钱损失的决策领域中，老年人和年轻人表现出相似的冒险倾向。在另一项研究中，我们采用BART范式（(Lejuez et al., 2002)作为包含学习的经验性决策范式进行行为研究和神经影像学研究，从而探究风险承担中的年龄差异和潜在的脑网络机制。结果显示，老化导致扣带-鳃盖网络（CON）内的激活增加，同时伴随皮质下网络（SN）的改变。值得注意的是，CON-SN之间的连接存在老化失连效应。具体而言，SN网络内与丘脑相关区域和CON网络之间的连通性在年轻人中增强，并且该连接与决策任务表现结果呈正相关。而老年人表现出SN内纹状体相关区域和CON内的连接增强，并且该连接强度与决策任务中的表现结果呈正相关。这一研究进一步深入探究了决策行为与大尺度脑网络之间的关系，并且说明了大脑与行为之间关联的复杂性以及该连接是如何随着年龄的增长而变化。

Abstract：Aging is accompanied by a wide range of neurobiological and cognitive decline, such as declines in reward learning (Eppinger, Schuck, Nystrom, & Cohen, 2013), reduced value coding ability (Halfmann, Hedgcock, Kable, & Denburg, 2016), which affects decision making abilities in a variety of risky choice tasks (Best & Charness, 2015; Mata, Josef, Samanez-Larkin, & Hertwig, 2011). But do different measures converge in showing similar age-related patterns and which underlining neural mechanism contribute to the difference? We conducted three studies to investigate the risk taking propensity across young and older adults. In our first work, we employed self-report assessments to measure risk taking and risk aversion propensity and find self-reported risk taking to decrease with age while no significant difference in risk aversion propensity between older adults and young counterparts. In our second study, we adopted a gambles task involving described monetary lotteries(Rieskamp, 2008; Tymula, Rosenberg Belmaker, Ruderman, Glimcher, & Levy, 2013) as an example measure withouta learning component to examine age differences in risk seeking when choices include a certain option (a sure gain or a sure loss). Compared with younger adults, in both equal expected values and better expected values setting, older adults preferred a certain gain over a chance to win a larger gain and thus, exhibited less risk taking in the domain of gains. However, this phenomenon can not be found in the domain loss and older and young adults show similar risk taking tendency. In another study, we introduced the BART (Lejuez et al., 2002) as an example of a behavioral measure and neuroimage measure with a learning component to contribute to the understanding of age differences in risk taking and underlying possible brain network mechanisms.The results showed that advanced age led to an increasing activation within the cingulo-opercular network (CON), accompanied by its coupling alterations with the subcortical network (SN). It is noteworthy that there was a segregation age effect on the subdivision of CON-SN connection. Specifically, connectivity between thalamus-related regions within the SN and CON enhanced in younger adults and positively associated with task performance (i.e., total win), while the striatum-related regions within the SN and CON enhanced in older adults during risk taking and associated with task performance. This provides further insight into the relationship between the risk taking behavior and large-scale network interactions and, in general, illustrates the intricacy of brain-behavior associations and how they can change with advanced age.