NEURAL AND COMPUTATIONAL PRINCIPLES UNDERLYING SOCIAL VS NON-SOCIAL DECISION MAKING

大多数战略决策都是在相当大的不确定性下做出的。例如，当投资于股票市场时，交易者可能会使用纯粹的概率模型来估计市场波动的风险。相比之下，当亲自谈判交易时，交易员的风险评估可能会依赖于另一方的可信度。在标准的效用模型中，管理这些决策的规则是相同的，不管不确定性的来源是什么（例如，人类还是在线平台）。然而，社会神经科学的最新进展表明，不同的大脑网络可能会不同地处理概率和社会信息，可能导致不同的结果。到目前为止，还没有统一的框架来整合决策不确定性的社会和非社会来源，因为以前的研究孤立地看待这些因素。这一缺陷主要是由于这项工作的跨学科性质，这需要在实验设计和大脑分析方面取得重大的方法学进展。在这里，我们将结合联合收割机两种流行的脑成像技术（脑电图功能磁共振成像），与新颖的实验设计和计算建模，以获得信息的时候，在哪里以及如何大脑处理社会和非社会信息在决策过程中。我们将研究决策过程的两个不同阶段：（1）选择阶段，评估和比较决策方案以指导行动;（2）结果阶段，计算预期的回报和风险信号以更新未来的预期。我们将在每个阶段对社会和非社会形式的不确定性的整合进行建模，并解释相关神经系统的计算原理。在这样做的过程中，我们将放置新的，神经生物学衍生的，信息集成的决策理论模型的约束。将社会和非社会形式的不确定性结合到一个全面的决策理论中，有望大大提高我们对重要现实生活事件的理解，从政策制定和风险管理到为个人健康行为和储蓄策略的决策提供信息。目前的项目将更仔细地研究人们如何预测和学习他人的行动。尤其是我们如何预测他人的冒险行为，以及负责的大脑机制是什么。根据博弈论，参与者构建了一个其他人行为的模型，然后选择最佳策略，使他们的回报最大化，损失最小化。为了决定他们的最佳行动，每个玩家要么从其他玩家的行动频率中学习，要么模拟其他玩家的当前状态。这种预测在被违反时会产生“状态预测错误”，当环境中出现意外变化时会发生这种错误。当做出预测并观察对手的行为时，参与者根据预测结果和观察结果之间的差异更新他们当前的知识。这被称为“模拟他人行为预测误差”（sAPE）。模拟预测通常基于玩家的偏好和价值观，以及关于其他对手的知识（基于经验和信念学习）。一些突出的问题是：在社会决策过程中，心理模拟与大脑有什么关系？在为自己做决定时，大脑信号是否与预测他人的决定相似？是否有特定的大脑区域和过程来模拟别人的预测错误，或者它们与自己预测错误的区域重叠？目前的博士项目将致力于通过使用计算模型和fMRI，EEG或同时使用两者来回答其中的一些问题。

Most strategic decisions occur under considerable uncertainty. For example, when investing in the stock market, a trader may use purely probabilistic models to estimate risk in the market's fluctuations. In contrast, when negotiating a deal in person, the trader's risk assessment may rely instead on how trustworthy the other party appears. In standard utility models, the rules governing such decisions are the same, regardless of the source of uncertainty (e.g. human vs on-line platform). However, recent advances in social neuroscience suggest that separate brain networks might distinctly process probabilistic and social information, possibly leading to different outcomes. To date, there is no unified framework for integrating social and non-social sources of decision uncertainty as previous studies looked at these factors in isolation. This shortcoming is mainly due to the interdisciplinary nature of the endeavour, which requires major methodological developments in experimental design and brain analytics. Here, we will combine two popular brain imaging techniques (EEG-fMRI), with novel experimental design and computational modelling to obtain information on when, where and how the brain processes social and non-social information during decision-making. We will investigate two different phases of the decision process: (1) the choice-phase, where decision alternatives are evaluated and compared to guide action and (2) the outcome-phase, where expected reward and risk signals are computed to update future expectations. We will model the integration of social and non-social forms of uncertainty at each stage and characterise the computational principles of the relevant neural systems. In doing so, we will place new, neurobiologically-derived, constraints on decision-theoretic models of information integration. The marriage of social and non-social forms of uncertainty into a comprehensive theory of decision-making promises to significantly improve our understanding of important real-life events, ranging from policy making and risk management to informing individual decisions on health behaviours and savings strategies. The current project will look more closely into how people predict and learn from others' actions. And especially how we predict risk-taking in others and what are the responsible brain mechanisms. According to game theory players construct a model of others' behaviours and subsequently choose the best strategy that would maximise their rewards and minimise their losses. To decide on their best action each player either learns from the frequency of the other's actions or simulates the current state of the other player. Such predictions when violated produce the 'state prediction error', which occurs when there are unexpected changes in the environment. When predictions are made, and behaviour from the opponent is observed, participants update their current knowledge according to the difference between the predicted and the observed outcomes. This is termed 'simulated other's action prediction error' (sAPE). The simulated predictions are usually based on a player's preferences and values, with addition to knowledge about the other opponent (based on experiences and belief learning). Some outstanding questions are: what are the brain correlates of mental simulation during social decision making? Are the brain signals responsible for prediction error in decision making for oneself similar to predicting others' decisions? Are there specific brain areas and processes for simulating others' prediction error, or do they overlap with the areas for prediction error for oneself? The current PhD project will aim to answer some of these questions with using computational models and either fMRI, EEG or both simultaneously.