Dynamic memory (re-)activation in human decision making

在做决定时，我们的个人和情境偏好受到过去的经验和学习的强烈影响。传统上，基于价值的选择的神经认知描述侧重于程序和语义记忆功能，作为对行为经济学中主观价值和概率加权函数的正式描述的补充。最近，越来越多的人假设，决策也可能是由具体记忆的检索和（再）激活所决定的，例如，从情景记忆的抽样中。迄今为止，具体记忆再激活对决策的贡献，尽管其理论和直观的吸引力，尚未在人类神经信号中直接研究。在这里，我们试图用一个新开发的分析框架来填补这一空白，以揭示决策审议过程中人类脑电图模式的动态记忆（再）激活。在一系列实验中，我们测试了在(i)简单价值判断、（ii）传递推理和（iii）情境依赖价值学习的决策评估过程中先前遇到的价值项目的重新激活。在进一步的研究中，我们采用相同的分析框架来测试先前经验结果的内生（再）激活在多大程度上控制（iv）风险选择中的主观偏好和(v)动态“现实世界”场景中的主动冒险。将多变量神经模式分析与行为计算模型相结合，利用人类脑电图的毫秒分辨率，我们试图深入了解记忆采样的时间动态，以及它在人类决策审议中的行为相关性。

When making decisions, our individual and situational preferences are strongly influenced by past experience and learning. Neurocognitive accounts of value-based choice have traditionally focused on procedural and semantic memory functions, in complement to formal descriptions of subjective value- and probability weighting functions in behavioral economics. More recently, it has increasingly been hypothesized that decisions may be crucially shaped also by retrieval and (re-)activation of concrete memories, for instance, in terms of sampling from episodic memory. To date, the contribution of concrete memory reactivation to decision-making, despite its theoretical and intuitive appeal, has not been studied directly in human neural signals. Here, we seek to fill this gap using a newly developed analytical framework to disclose dynamic memory (re-)activations in human electroencephalographic patterns during decisional deliberation. In a series of experiments, we test for reactivation of previously encountered value items during decisional evaluation in (i) simple value judgments, (ii) transitive inference, and (iii) context-dependent value learning. In further studies, we adopt the same analysis framework to test the extent to which endogenous (re-)activation of previously experienced outcomes governs (iv) subjective preferences in risky choice and (v) active risk-taking in a dynamic “real-world” scenario. Fusing multivariate neural pattern analyses with computational modelling of behavior, and capitalizing on the millisecond resolution of human electroencephalography, we seek to gain insights into the temporal dynamics of memory sampling, and its behavioral relevance in human decisional deliberation.