Neurocomputational approaches to individual differences in `virtuous' decision-making

人脑中的哪些运算决定了我们是失败还是成功地做出了“善意”的选择？吃得更健康、为退休储蓄或减少生态足迹是被广泛认为是“好”的选择的几个例子，因为它们对决策者和/或社会有长期好处。然而，经验数据和个人经验都表明，人们在将选择与道德目标相结合的倾向上存在巨大差异，因为这往往要求放弃直接、直接的奖励，转而支持更间接、更抽象或更延迟的奖励(例如，“健康”)。我和我的受训者使用下面的框架，研究了在不同人、不同背景和不同选择领域中，驱动“美德”决策差异的神经基础。大多数选择都需要权衡相互竞争的因素。一个流行的计算模型假设，这是通过计算每个选择选项(例如午餐的披萨或沙拉)的决策值(DV)作为与选择相关的考虑因素(或属性)的加权和(DV=？iwi(Attributei))，并比较DV来实现的。然而，并不是所有属性的权重都是一样的，这导致了选择上的偏见--通常倾向于直接的享乐主义回报。采用这一框架，我的研究计划构建了三个相关的目标，以了解“好的”决策中差异的神经基础。在AIMS上重叠的参与者样本允许他们的关键问题、衡量标准和方法工具之间的协同作用。-目标1.人与人之间的差异。为什么有些人似乎比其他人更容易做出“正确”的选择？我们探索了大脑解剖学和大脑决策网络节点的连通性在不同人的计算和选择中的差异所起的作用。-目标2.延展性。特定属性的计算在大脑中的可塑性有多大？语境和注意力对“美德”属性的操纵能在多大程度上改变选择？大脑数据的分析利用先进的多变量模式分析(MVPA)，这将是HQP培训的关键要素。-目标3.提高普适性和生态有效性。那些在一个选择领域(例如饮食选择)难以将“美德”属性(与抽象的、间接的奖励联系在一起)结合在一起的人，是否也更有可能在他们生活的其他领域(例如为退休储蓄)与此作斗争？在实验室获得的“善意选择”的模型估计在多大程度上适用于现实世界中的行为？我和我的受训者将利用这些发现来建立一个人类决策的神经信息模型，这是这个研究计划的长期目标。这项工作对我们理解选择的基本认知过程具有变革性，并可以为政策制定者提供信息。了解基本的计算、不同的神经基础及其延展性最终可能会为有针对性的、特定于过程的干预提供信息，从而有效地增加我们日常生活各个领域的“美德”行为。

What computations in the human brain determine whether we fail or succeed to make `virtuous' choices? Eat healthier, save for retirement, or reduce the ecological footprint are a few examples of choices widely considered as `good' due to their long-term benefits for decision-makers and/or society. Yet, empirical data and personal experience alike point to dramatic differences in peoples' tendencies to align choices with virtuous goals, as this often requires forgoing direct, immediate rewards in favor of rewards that are more indirect, abstract or delayed (e.g. `health'). My trainees and I study the neural underpinnings that drive differences in `virtuous' decision-making across people, contexts, and choice domains, using the following framework. Most choices require weighing competing considerations. A popular computational model assumes that this is done by computing a decision value (DV) for each choice option (e.g. pizza or salad for lunch) as the weighted sum of choice-relevant considerations (or attributes) (DV=?iwi(Attributei), and comparing DVs. Yet, not all attributes are weighted equally, leading to biases in choices - often in favor of immediate hedonistic rewards. Adopting this framework, my research program structures three related aims to understand the neural basis of variance in `good' decision-making. Overlapping participant samples across aims permit synergistic cross talk among their key questions, measures, and methodological tools. -AIM 1. Differences across people. Why do some people seem to have an easier time making the `right' choice than others? We explore the role of brain anatomy and connectivity of nodes of the brain's decision network for differences in computations and choices across people. -AIM 2. Malleability. How malleable are attribute-specific computations in the brain? To what degree can contextual and attention manipulations towards `virtuous' attributes alter choices? Analyses of brain data utilizes advanced multivariate pattern analyses (MVPA) that will be a crucial element of HQP training. -AIM 3.  Generalisability and ecological validity. Are people that have difficulties incorporating `virtuous' attributes (linked to abstract, indirect rewards) in one choice domain (e.g. dietary choice) also more likely to struggle with this in other areas of their life (e.g. saving for retirement)? To what degree do model estimates of `virtuous choice' obtained in the lab generalize to behaviors in the real world? My trainees and I will use these findings to build a neurally informed model of human decision-making, a long-term goal of this research program. This work is transformational for our understanding of basic cognitive processes of choices and can inform policy makers. Understanding the underlying computations, distinct neural substrates, and their malleability might ultimately inform targeted, process-specific interventions to effectively increase `virtuous' behaviors in various domains of our daily life.