Neural circuit mechanisms of social decision-making

为了了解决策的神经机制，研究集中在单个动物如何在控制良好的环境中做出选择。然而，现实生活中的决定很少由单一的代理人孤立地做出。相反，决策反映了多个决策者之间的动态互动，而成功取决于预测其他人的行动。计算机视觉的最新进展使得通过实时视频跟踪、闭环任务设计以及在社会互动过程中提取行为主题的特征来监控多个动物成为可能。利用这项技术，我们开发了一个定量框架来研究小鼠社会决策的神经回路机制，其中多只小鼠在与行为学相关的社会觅食任务中进行博弈论互动。为了最大限度地获得回报和社会效用，动物不仅应该跟踪对手的当前位置和选择，还应该预测他们未来的行动。我们使用多人行为的定量分析和建模来比较动物的实际策略和这些最优解决方案。越来越多的证据表明，前额叶皮质(PFC)和杏仁核参与了这些计算。但关于PFC-杏仁核回路的详细和因果贡献，仍有许多问题。我们模型中的潜在变量可以映射到动物的内部状态，以生成神经记录的可测试预测，并约束对PFC-杏仁核回路扰动的解释。我们将在我们的定量社会觅食任务中使用体内成像、电生理学和光遗传学的多区域和特定路径的方法，以测试PFC和杏仁核内部和之间的情感、奖励和社会信息如何促进大脑中的代理特定表征。这反过来将揭示社会动物如何在适应性和目标导向的行为中建立世界模型的一般原则。

To understand the neural mechanisms of decision-making, research has focused on how a single animal makes choices in a well controlled environment. However, real life decisions are rarely made by a single agent in isolation. Instead, decisions reflect dynamic interactions between multiple decision makers, and success depends on predicting the actions of others. Recent advances in computer vision make it feasible to monitor multiple animals with real-time video tracking, closed-loop task design; and extract features of behavioural motifs during social interactions. Taking advantage of this technology, we have developed a quantitative framework to study the neural circuit mechanisms of social decision-making in the mouse, where multiple mice engage in game theoretic interactions in an ethologically relevant social foraging task. To maximise reward and social utility, animals should not only track the current position and choice of their opponent, but also predict their future actions. We use quantitative analysis and modelling of multi-player behaviour to compare animals' actual strategies to these optimal solutions. Converging evidence have implicated the prefrontal cortex (PFC) and the amygdala in these computations. But many questions remain about the detailed and causal contributions of the PFC-amygdala circuits. The latent variables in our models can be mapped to internal states of the animal to generate testable predictions for neural recordings and constrain interpretation of perturbations of the PFC-amygdala circuit. We will use multi-regional and pathway-specific methods of in vivo imaging, electrophysiology and optogenetics in our quantitative social foraging task to test how affective, reward, and social information within and between PFC and amygdala facilitate agentspecific representation in the brain. This will in turn expose general principles of how social animals build a model of the world during adaptive and goal-directed behaviour.