跨期选择的认知机制与神经基础

Cognitive and Neural Mechanisms of Intertemporal Choice

跨期选择是现实生活中一个极为普遍的决策现象。本文首先介绍跨期选择的几种经典心理效应,重点分析了跨期选择的认知机制:计算模型和认知成分模型,然后对其神经基础的三种研究取向(双机制加工、单机制加工、自我控制)进行了阐述。最后,对跨期选择的研究进展进行了分析和总结,并提出了该领域将来可能的研究方向,这对于进一步深化跨期选择研究具有重要的参考价值。

Decision makers frequently balance the costs and benefits of decision-making process which should occur at different times. Humans, however, exhibit impatience in such intertemporal choices, especially if immediate rewards are available. As the delay until payoff delivery increases, an individual’s valuation of a future reward declines, a phenomenon which is known as temporal discounting. From an economic point of view, the computing theories of intertemporal decision-making construct mathematical models of temporal discounting, which contains the discounted utility model （DU）, the hyperbolic discounting model family （including the hyperbolic discounting model and the quasi-hyperbolic discounting model）, the subadditive discounting model, and the ＂as soon as possible＂ （ASAP） discounting model. The ＂As soon as possible＂ （ASAP） discounting model proposed recently assumes that human subjects do not necessarily cause the impulsive preference reversals to be predicted by hyperbolic-like discounting, and subjective value declines hyperbolically in relation to the soonest currently available reward. From the perspective of psychology, the cognitive omponent theories study the psychological effects and cognitive components of intertemporal choice, including impulsive and self-control theories, cognitive imagery theories, and the query theory. The query theory has a greater impact in recent years. It assumes that preferences, like all knowledge, are subject to the processes and dynamics of memory encoding and retrieval, and explores whether memory and attention processes can explain observed anomalies in evaluation and choice. There are three recent neural accounts： dual-valuation, single-valuation and self-control. The dual-valuation account assumes two different valuation processes during intertemporal choice. An impatient β-process steeply discounts all non-immediate rewards and is active only in now trials, and a more patient δ-process far less steeply discounts all delayed rewards and is active in both now and not-now trials. The single-valuation account assumes that at least one location in the human brain encodes the hyperbolically discounted value of all rewards, both on now and not-now trials, and that the option with the higher discounted value is then chosen. However, the single- and dual-valuation accounts agree that the choice of an option results directly from the comparison of their valuations, without additional intervening processes such as self-control. In contrast, the self-control account assumes that preference expressed by evaluation judgments and choice can be discrepant, because neural valuations reflected by valuation judgments can be overridden in choice. Choice is determined not by valuation processes alone, but also involves self-control processes, and especially, the areas of DLPFC may involve self-control process. What has been reviewed above suggests that we are just beginning to gain insight into the nature of temporal decision-making. Much of this work has sought to explore the phenomenon of delay discounting; compared with current or recent gains （or losses）, people always give less weight on future gains （or losses）. We have pointed to several places where there is currently ambiguity in the treatment of temporal choice： how to perceive time, the neural mechanisms of intertemporal choice, e.g. correlates of patience （self-control） vs. subjective value signals. With the clinical and public policy implications of temporal decision-making as well as the sheer scientific potential in psychology, neuroscience, and economics, further studies should more closely exam the application of intertemporal choice and give more attention from an evolutionary perspective.