期刊
CEREBRAL CORTEX
卷 27, 期 2, 页码 1270-1284出版社
OXFORD UNIV PRESS INC
DOI: 10.1093/cercor/bhv333
关键词
fMRI; learning; parietal cortex; prediction error; prefrontal cortex
资金
- Stanford Center for Cognitive and Neurobiological Imaging
- Knut and Alice Wallenberg Foundation (Network Initiative on Culture, Brain, and Learning)
- NSF IGERT Traineeship Recipient [0801700]
Many decisions require a context-dependent mapping from sensory evidence to action. The capacity for flexible information processing of this sort is thought to depend on a cognitive control system in frontoparietal cortex, but the costs and limitations of control entail that its engagement should be minimized. Here, we show that humans reduce demands on control by exploiting statistical structure in their environment. Using a context-dependent perceptual discrimination task and model-based analyses of behavioral and neuroimaging data, we found that predictions about task context facilitated decision making and that a quantitative measure of context prediction error accounted for graded engagement of the frontoparietal control network. Within this network, multivariate analyses further showed that context prediction error enhanced the representation of task context. These results indicate that decision making is adaptively tuned by experience to minimize costs while maintaining flexibility.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据