Yin Chaoqun, Melin Maxwell D, Rojas-Bowe Gabriel, Sun Xiaonan Richard, Couto João, Gluf Steven, Kostiuk Alex, Musall Simon, Churchland Anne K
UCLA Neuroscience Interdepartmental Program.
Department of Neurobiology, University of California, Los Angeles.
bioRxiv. 2024 May 14:2023.06.26.546404. doi: 10.1101/2023.06.26.546404.
Existing work demonstrates that animals alternate between engaged and disengaged states during perceptual decision-making. To understand the neural signature of these states, we performed cortex-wide measurements of neural activity in mice making auditory decisions. The trial-averaged magnitude of neural activity was similar in the two states. However, the trial-to-trial variance in neural activity was higher during disengagement. To understand this increased variance, we trained separate linear encoding models on neural data from each state. The models demonstrated that although task variables and task-aligned movements impacted neural activity similarly during the two states, movements that are independent of task events explained more variance during disengagement. Behavioral analyses uncovered that during disengagement, movements become uncoupled to task events. Taken together, these results argue that the neural signature of disengagement, though obscured in trial-averaged neural activity, is evident in trial-to-trial variability driven by changing patterns of spontaneous movements.
现有研究表明，动物在感知决策过程中会在参与和脱离两种状态之间交替。为了理解这些状态的神经特征，我们对进行听觉决策的小鼠进行了全皮层神经活动测量。在这两种状态下，神经活动的试验平均幅度相似。然而，在脱离状态下，神经活动的逐次试验方差更高。为了理解这种增加的方差，我们根据来自每种状态的神经数据训练了单独的线性编码模型。这些模型表明，虽然任务变量和与任务对齐的运动在两种状态下对神经活动的影响相似，但与任务事件无关的运动在脱离状态下解释了更多的方差。行为分析发现，在脱离状态下，运动与任务事件解耦。综上所述，这些结果表明，脱离状态的神经特征虽然在试验平均神经活动中被掩盖，但在由自发运动模式变化驱动的逐次试验变异性中是明显的。
