Department of Biological and Clinical Psychology, Friedrich Schiller University, D-07743 Jena, Germany; Institute of Medical Psychology and Systems Neuroscience, University Hospital Muenster, D-48149 Muenster, Germany.
Department of Biological and Clinical Psychology, Friedrich Schiller University, D-07743 Jena, Germany.
Neuroimage. 2016 Dec;143:50-57. doi: 10.1016/j.neuroimage.2016.09.022. Epub 2016 Sep 10.
Several regions of the frontal cortex interact with striatal and amygdala regions to mediate the evaluation of reward-related information and subsequent adjustment of response choices. Recent theories discuss the particular relevance of dorsal anterior cingulate cortex (dACC) for switching behavior; consecutively, ventromedial prefrontal cortex (VMPFC) is involved in mediating exploitative behaviors by tracking reward values unfolding after the behavioral switch. Amygdala, on the other hand, has been implied in coding the valence of stimulus-outcome associations and the ventral striatum (VS) has consistently been shown to code a reward prediction error (RPE). Here, we used fMRI data acquired in humans during a reversal task to parametrically model different sequences of positive feedback in order to unravel differential contributions of these brain regions to the tracking and exploitation of rewards. Parameters from an Optimal Bayesian Learner accurately predicted the divergent involvement of dACC and VMPFC during feedback processing: dACC signaled the first, but not later, presentations of positive feedback, while VMPFC coded trial-by-trial accumulations in reward value. Our results confirm that dACC carries a prominent confirmatory signal during processing of first positive feedback. Amygdala coded positive feedbacks more uniformly, while striatal regions were associated with RPE.
额皮质的几个区域与纹状体和杏仁核区域相互作用，以介导与奖励相关的信息的评估和随后的反应选择的调整。最近的理论讨论了背侧前扣带皮层（dACC）在行为转换中的特殊相关性；继而，腹内侧前额叶皮层（VMPFC）通过跟踪行为转换后展开的奖励值，参与了对剥削性行为的介导。另一方面，杏仁核已被暗示在对刺激-结果关联的效价进行编码，而腹侧纹状体（VS）则一致地对奖励预测误差（RPE）进行编码。在这里，我们使用人类在反转任务中获得的 fMRI 数据，对不同的正反馈序列进行参数建模，以揭示这些大脑区域对奖励的跟踪和利用的不同贡献。最优贝叶斯学习者的参数准确地预测了 dACC 和 VMPFC 在反馈处理过程中的不同参与：dACC 信号表示第一次，但不是第二次，正反馈的出现，而 VMPFC 则对奖励值的逐次积累进行了编码。我们的结果证实，dACC 在处理第一次正反馈时携带一个突出的确认信号。杏仁核对正反馈的编码更为一致，而纹状体区域则与 RPE 相关。
