Cognitive and reward signals for choices under ambiguity

摘要 糟糕的决策可能会产生严重的后果，决策行为受损是许多疾病的症状 神经和精神障碍。决策理论认识到两种形式的不确定性：风险--其中 潜在的概率分布是已知的，以及模棱两可的-其中潜在的概率分布 都是未知的。风险是预期效用理论的概念基础。尽管它的影响是持久的 在经济学中，这一理论未能充分描述现实世界的选择行为。一个原因是，纯粹的风险是 罕见：只有在赌场、掷硬币和决策实验中才会遇到这种情况。在现实世界中，不完整 信息、稀疏数据和认知限制造成了不确定状态，这些状态存在于 纯粹的冒险和盲目的无知。因此，模糊性，而不是风险，描述了大多数情况下 决策就会发生。在模棱两可的情况下，不可能计算预期效用。相反， 决策者(有意识或下意识地)用信念取代预期效用估计，这已经 对健康的人以及那些有神经和精神障碍的人的后果。健康的决策 制造商表现出对模棱两可的厌恶：他们避免模糊的前景，即使这样做代价高昂。个人 前额叶功能障碍在模棱两可的选择中表现不佳，即使风险下的选择仍然存在 相对完好无损。尽管歧义及其后果很普遍，但神经元是如何编码的还不得而知 与模糊性相关的不确定性，或者这些神经估计如何用于决策。至 了解神经电路如何实现经济选择，以及这种实现可能如何随 前额叶功能障碍，我们将使用非人类灵长类动物模型来表征认知和 奖励系统对模糊情况下的决策，并提供了对模糊厌恶的机械性理解。 本研究的中心假设是歧义厌恶是元认知评价的结果。 概率判断。换句话说，这不是信仰，而是决策者对这些信仰的信心， 这导致了人们对模棱两可的厌恶。我们将使用行为分析、神经生理学和遗传编码 用消音器探测背外侧前额叶皮质和多巴胺神经元的功能并回答三 基本问题。(1)歧义如何编码并转化为信念？(2)歧义的作用是什么？ 认知系统在估计歧义和做出好的选择中的作用？(3)概率的信心如何？ 判断会影响决策吗？我们预测，背外侧前额叶皮质的不确定性编码将会减少 在不明确的情况下可靠，与有风险的情况相比，会导致更低的确定性，更低的信心，以及， 归根结底，是对模棱两可的厌恶。我们的结果将增加对决策的神经基础的理解 在模棱两可的情况下，并提供对一系列人类疾病状态的诊断和治疗选择的见解。

Abstract Poor decisions can have severe consequences, and impaired decision-making behavior is a symptom of many neurological and psychiatric disorders. Decision theory recognizes two forms of uncertainty: risk – in which the underlying probability distributions are known, and ambiguity – in which the underlying probability distributions are not known. Risk is the conceptual foundation for the Expected Utility theory. Despite its enduring influence in economics, this theory fails to adequately describe real-world choice behavior. One reason is that pure risk is rare: it is only encountered in casinos, coin-flips, and decision-making experiments. In the real world, incomplete information, sparse data, and cognitive limitations create states of uncertainty that exist on a spectrum between pure risk and blind ignorance. Therefore, ambiguity, rather than risk, describes the conditions under which most decision-making occurs. In ambiguous circumstances, it is impossible to calculate expected utilities. Instead, decision makers (consciously or subconsciously) substitute beliefs for expected utility estimations, and this has consequences for healthy individuals and those with neurological and psychiatric disorders. Healthy decision makers exhibit ambiguity aversion: they avoid ambiguous prospects, even when doing so is costly. Individuals with prefrontal dysfunctions perform poorly during choices under ambiguity, even when choices under risk remain relatively intact. Despite the prevalence of ambiguity and its consequences, it is not known how neurons code for the uncertainty associated with ambiguity, or how these neural estimates are used for decision making. To understand how neural circuits implement economic choices and how this implementation may be altered with prefrontal dysfunction, we will use a non-human primate model to characterize the contributions of cognitive and reward systems to decisions under ambiguity, and to provide a mechanistic understanding of ambiguity aversion. The central hypothesis of this project is that ambiguity aversion results from the metacognitive appraisal of probability judgements. In other words, it is not beliefs, but rather decision makers’ confidence in those beliefs, that causes ambiguity aversion. We will use behavioral analysis, neurophysiology, and genetically coded silencers to probe the functions of dorsolateral prefrontal cortex and dopamine neurons and answer three fundamental questions. (1) How is ambiguity coded and transformed into beliefs? (2) What are the roles of cognitive systems in estimating ambiguity and making good choices? (3) How does confidence in probability judgements affect decisions? We predict that uncertainty coding in dorsolateral prefrontal cortex will be less reliable in ambiguous, when compared to risky, conditions and lead to lower certainty, lower confidence, and, ultimately, ambiguity aversion. Our results will increase understanding of the neural basis of decision-making under ambiguity and provide insights into diagnosis and treatment options for an array of human disease states.