Laboratory of Sensorimotor Research, National Eye Institute, NIH, Bethesda, MD 20892;
Division of Biomedical Science, Faculty of Medicine, University of Tsukuba, 305-8577 Tsukuba, Japan.
Proc Natl Acad Sci U S A. 2021 Jan 26;118(4). doi: 10.1073/pnas.2013623118.
Basal ganglia contribute to object-value learning, which is critical for survival. The underlying neuronal mechanism is the association of each object with its rewarding outcome. However, object values may change in different environments and we then need to choose different objects accordingly. The mechanism of this environment-based value learning is unknown. To address this question, we created an environment-based value task in which the value of each object was reversed depending on the two scene-environments (X and Y). After experiencing this task repeatedly, the monkeys became able to switch the choice of object when the scene-environment changed unexpectedly. When we blocked the inhibitory input from fast-spiking interneurons (FSIs) to medium spiny projection neurons (MSNs) in the striatum tail by locally injecting IEM-1460, the monkeys became unable to learn scene-selective object values. We then studied the mechanism of the FSI-MSN connection. Before and during this learning, FSIs responded to the scenes selectively, but were insensitive to object values. In contrast, MSNs became able to discriminate the objects (i.e., stronger response to good objects), but this occurred clearly in one of the two scenes (X or Y). This was caused by the scene-selective inhibition by FSI. As a whole, MSNs were divided into two groups that were sensitive to object values in scene X or in scene Y. These data indicate that the local network of striatum tail controls the learning of object values that are selective to the scene-environment. This mechanism may support our flexible switching behavior in various environments.
基底神经节有助于物体-价值学习，这对生存至关重要。其潜在的神经机制是将每个物体与其奖励结果相关联。然而，物体的价值在不同的环境中可能会发生变化，我们需要相应地选择不同的物体。这种基于环境的价值学习的机制尚不清楚。为了解决这个问题，我们创建了一个基于环境的价值任务，其中每个物体的价值取决于两个场景环境（X 和 Y）而发生反转。在反复经历这个任务后，猴子们能够在场景环境意外变化时切换对物体的选择。当我们通过局部注射 IEM-1460 阻断纹状体尾部的快放电中间神经元（FSIs）对中等棘突投射神经元（MSNs）的抑制性输入时，猴子们就无法学习场景选择性的物体价值。然后，我们研究了 FSIs-MSN 连接的机制。在学习之前和期间，FSIs 选择性地对场景做出反应，但对物体价值不敏感。相比之下，MSNs 能够区分物体（即，对好物体的反应更强），但这种情况仅在两个场景之一（X 或 Y）中出现。这是由 FSIs 的场景选择性抑制引起的。总体而言，MSNs 分为两组，它们分别对场景 X 或场景 Y 中的物体价值敏感。这些数据表明，纹状体尾部的局部网络控制着对场景环境具有选择性的物体价值的学习。这种机制可能支持我们在各种环境中的灵活切换行为。
