Dept. of Psychology, Royal Holloway, University of London, Egham, Surrey, TW20 0EX, UK.
Neuroimage. 2013 Jan 1;64:1-9. doi: 10.1016/j.neuroimage.2012.09.010. Epub 2012 Sep 13.
The ability to recognise that another's belief is false is a hallmark of our capacity to understand others' mental states. It has been suggested that the computational and neural mechanisms that underpin learning about others' mental states may be similar to those that underpin first-person Reinforcement Learning (RL). In RL, unexpected decision-making outcomes constitute prediction errors (PE), which are coded for by neurons in the Anterior Cingulate Cortex (ACC). Does the ACC signal the PEs (false beliefs) of others about the outcomes of their decisions? We scanned subjects using fMRI while they monitored a third-person's decisions and similar responses made by a computer. The outcomes of the trials were manipulated, such that the actual outcome was unexpectedly different from the predicted outcome on 1/3 of trials. We examined activity time-locked to privileged information which indicated the actual outcomes only to subjects. Activity in the gyral ACC was found when the outcomes of the third-person's decisions were unexpectedly positive. Activity in the sulcal ACC was found when the third-person's or computer's outcomes were unexpectedly positive. We suggest that a property of the ACC is that it codes PEs, with a portion of the gyral ACC specialised for processing the PEs of others.
能够识别他人的信念是错误的，这是我们理解他人心理状态的能力的标志。有人认为，支持学习他人心理状态的计算和神经机制可能与支持第一人称强化学习（RL）的机制相似。在 RL 中，出乎意料的决策结果构成预测误差（PE），由前扣带皮层（ACC）中的神经元编码。ACC 是否会对他人关于其决策结果的 PEs（错误信念）进行信号传递？我们使用 fMRI 对受试者进行扫描，同时他们监测第三个人的决策和计算机做出的类似反应。试验的结果被操纵，使得在三分之一的试验中，实际结果出乎意料地不同于预测结果。我们检查了仅向受试者显示实际结果的特权信息的活动。当第三人的决策结果出人意料地为正时，发现了脑回 ACC 的活动。当第三人或计算机的结果出人意料地为正时，发现了脑沟 ACC 的活动。我们认为，ACC 的一个特性是它对 PE 进行编码，其中一部分脑回 ACC 专门用于处理他人的 PE。
