美国哈佛大学Naoshige Uchida、HyungGoo R. Kim等研究人员合作建立不同时间尺度下多巴胺信号的统一框架。这一研究成果于2020年11月27日在线发表在国际学术期刊《细胞》上。

研究人员使用虚拟现实开发了将奖励预测误差（RPE）与价值区分开来的实验范例。研究人员检查了各个阶段的多巴胺回路活动，包括体细胞突增、体细胞和轴突处的钙信号以及纹状体多巴胺浓度。结果表明，多巴胺信号的上升与RPE一致，而不是与价值一致，并且在所有检查的阶段均观察到了这种上升。

多巴胺信号的上升可以由动态刺激来驱动，从而表明一种逐渐采取奖励措施。研究人员提供了对快速相位和缓慢倾斜的多巴胺信号的统一计算理解：多巴胺神经元在时间基础上对价值执行类似导数的计算。

据了解，中脑多巴胺神经元的快速相位活动被认为是RPE的信号，类似于机器学习中使用的时间差异误差。但是，最近报道缓慢增加的多巴胺信号研究提出，它们代表状态值并且独立于体细胞突波活动而出现。

附：英文原文

Title: A Unified Framework for Dopamine Signals across Timescales

Author: HyungGoo R. Kim, Athar N. Malik, John G. Mikhael, Pol Bech, Iku Tsutsui-Kimura, Fangmiao Sun, Yajun Zhang, Yulong Li, Mitsuko Watabe-Uchida, Samuel J. Gershman, Naoshige Uchida

Issue&Volume: 2020-11-27

Abstract: Rapid phasic activity of midbrain dopamine neurons is thought to signal reward predictionerrors (RPEs), resembling temporal difference errors used in machine learning. However,recent studies describing slowly increasing dopamine signals have instead proposedthat they represent state values and arise independent from somatic spiking activity.Here we developed experimental paradigms using virtual reality that disambiguate RPEsfrom values. We examined dopamine circuit activity at various stages, including somaticspiking, calcium signals at somata and axons, and striatal dopamine concentrations.Our results demonstrate that ramping dopamine signals are consistent with RPEs ratherthan value, and this ramping is observed at all stages examined. Ramping dopaminesignals can be driven by a dynamic stimulus that indicates a gradual approach to areward. We provide a unified computational understanding of rapid phasic and slowlyramping dopamine signals: dopamine neurons perform a derivative-like computation over values on a moment-by-moment basis.

DOI: 10.1016/j.cell.2020.11.013

Source: https://www.cell.com/cell/fulltext/S0092-8674(20)31530-0