Chen Su-Hua, Pan Xiao-Chuan, Sakagami Masamichi
Institute of Cognitive Neurodynamics, East China University of Science and Technology, Shanghai 200237, China.
Brain Science Institute, Tamagawa University, Tokyo 194-8610, Japan.
Sheng Li Xue Bao. 2020 Dec 25;72(6):765-776.
It has been reported that single-unit activity in the prefrontal cortex (PFC) and striatum represented visual stimulus and reward information. But how to encode these pieces of information is quite complex from the view of single-neuron activity. Different neurons represented stimulus or reward information in different task epochs with increasing or decreasing their activities relative to their baseline firing rates. The present paper was aimed to study whether population neurons in the two brain areas could stably encode task-relevant parameters in a whole trial period. We recorded single-unit activities in the lateral PFC (LPFC) and striatum while the monkey was performing a stimulus- reward prediction task, and analyzed the neuronal activities by the method of a multi-variable regression model and the linear support vector machine. The results showed that, although proportions of task-related neurons in the two areas varied largely in the whole trial period, LPFC population neurons encoded reward and stimulus information stably and reliably. Population neurons in the striatum encoded only reward information, not stimulus information. A group of neurons in the two areas represented combined information of stimulus and reward. Further analysis showed that LPFC neurons encoded reward information for a group of relevant stimuli, while striatal neurons encoded reward information for a specific stimulus. These results suggest that both LPFC and striatal population neurons are able to stably represent task-relevant information, but from different aspects of the task. The different strategies to encode information in the LPFC and striatum suggest their different contributions in reward-based decision making.
据报道，前额叶皮层（PFC）和纹状体中的单神经元活动代表视觉刺激和奖励信息。但从单神经元活动的角度来看，如何对这些信息进行编码是相当复杂的。不同的神经元在不同的任务阶段代表刺激或奖励信息，其活动相对于基线放电率增加或减少。本文旨在研究这两个脑区的群体神经元是否能在整个试验期稳定地编码与任务相关的参数。我们在猴子执行刺激-奖励预测任务时记录了外侧前额叶皮层（LPFC）和纹状体中的单神经元活动，并通过多变量回归模型和线性支持向量机的方法分析了神经元活动。结果表明，虽然在整个试验期这两个区域中与任务相关的神经元比例差异很大，但LPFC群体神经元稳定且可靠地编码奖励和刺激信息。纹状体中的群体神经元仅编码奖励信息，而不编码刺激信息。这两个区域中的一组神经元代表刺激和奖励的组合信息。进一步分析表明，LPFC神经元为一组相关刺激编码奖励信息，而纹状体神经元为特定刺激编码奖励信息。这些结果表明，LPFC和纹状体群体神经元都能够稳定地代表与任务相关的信息，但从任务的不同方面。LPFC和纹状体中编码信息的不同策略表明它们在基于奖励的决策中具有不同的贡献。
