Cognitive effort and decision-making: integrating computational, behavioral, and psychophysiological approaches

从决定午餐吃什么到做出职业选择，我们的日常经历都是由决定决定的，但并不是所有的决定都是以同样的方式做出的：有些决定需要付出努力，而且很慢，而另一些决定则很容易，很快就能做出。本研究旨在阐明努力过程在何种情况下被部署为最大化回报的服务。这一目标将通过解决三个基本问题来实现：1)人们如何计算和平衡认知努力的相对成本和收益，2)这些成本和收益的心理生理标记是什么，这些信号如何影响行为所证明的认知努力的支出，以及3)认知能力的个体差异如何影响这种成本-收益计算？******虽然人们何时以及为什么决定消耗或保留认知努力的问题最近得到了理论研究，但在理解任务环境或个人层面的因素如何影响这种努力过程的分配方面，实验进展甚微。该研究利用了强化学习（RL）这一计算框架来理解我们如何学习选择行为以最大化回报，以精确计算支持我们的努力-回报权衡谈判。******一个有影响力的基于强化学习的账户认为，时间的机会成本应该决定响应速度或“活力”：当延迟行动的成本更高时，行动应该更快。我们对认知要求策略的依赖应该相应地改变：当时间昂贵时，我们应该采用认知成本低廉的策略来做决定。为了测试这个想法，我将使用一个实验设计，实现机会成本的波动，将其作为每秒平均奖励率来操作，以仔细隔离成本。这种操作将用于几个具有良好特征的认知任务，从而了解成本和收益如何决定认知努力支出的试验间调整。接下来的一系列实验将表征认知努力调节，我们将使用瞳孔测量（加工负荷的生理标记）来阐明这些假定的成本和收益是如何在内部表示的。与认知能力的行为基线测量相结合，这些生物标志物将提供实时认知努力的量化以及对认知成本敏感性的个体差异。

Everyday experiences are defined by decisions, from deciding what to eat for lunch to making career choices, but not all of our decisions come about the same way: some decisions are made with effort and are slow, while other decisions are easier and made quickly. The proposed research aims to elucidate when effortful processes are deployed the service of maximizing rewards. This aim will be met by addressing three fundamental questions: 1) how do people calculate and balance the relative costs and benefits of cognitive effort, 2) what are the psychophysiological markers of these costs and benefits and how do these signals influence the expenditure of cognitive effort as evidenced by behavior, and 3) how do individual differences in cognitive ability influence this cost-benefit calculus?******While the question of when and why people decide to expendor withholdcognitive effort has received recent theoretical treatment, little experimental progress has been made in understanding how either the task environment or factors at the level of the individual affects the allocation of such effortful processes. The proposed research leverages Reinforcement Learning (RL)a computational framework for understanding how we learn to select actions in the service of maximizing rewardsto pinpoint computations that underpin our negotiation of the effort-reward tradeoff.******One influential RL-based account posits that the opportunity cost of time should dictate response speed or “vigor”: when delayed action is more expensive, actions should be made more quickly. Our reliance upon cognitively demanding strategies should shift accordingly: when time is expensive, we should employ cognitively inexpensive' strategies to make decisions. To test this idea, I will use an experimental design that implements fluctuations in opportunity costsoperationalized as the average rate of reward per secondto carefully isolate the cost' . This manipulation will be used with several well-characterized cognitive tasks, yielding an understanding of how costs and benefits dictate trial-to-trial adjustments of cognitive effort expenditure. Following this set of experiments that will characterize cognitive effort modulation, we will use pupillometrya physiological marker of processing loadto elucidate how these putative costs and benefits are represented internally. In conjunction with behavioral baseline measures of cognitive ability, these biomarkers will afford quantification of moment-to-moment cognitive effort as well as individual variation in sensitivity to cognitive costs.