A successful chess player uses knowledge of past strategies and games to anticipate their opponent’s moves, adapting their play on the basis of the specific board configuration and planning several moves ahead. Context-specific past experiences provide important insights for making predictions about future events. In Nature Communications, Tarder-Stoll et al. describe how they explored, using immersive virtual reality and functional MRI, the way in which the brain represents context-specific temporal structure. Participants navigated through 16 naturalistic environments in a VR setting and were asked to create stories to remember the environment sequences, which varied in content and order. A day later, they were shown an image of one environment and were asked to anticipate upcoming environments in the sequence, several steps into the future, while undergoing fMRI scanning. The authors used multivoxel pattern similarity analysis to determine which environments were represented during anticipation. The study shows simultaneous representations of future and past environments, which were more temporally extensive in the hippocampus than in early visual cortex and were context-specific. Participants were also quicker at anticipating future events when hippocampal representations of distant environments were suppressed. These findings align with many theories of predictive processing in the medial temporal lobe.

Original reference: Nat. Commun. 15, 9094 (2024)

成功的国际象棋选手会利用过去的策略和游戏的知识来预测对手的棋步，根据特定的棋盘配置调整他们的游戏，并提前规划几步。特定于上下文的过去经验为预测未来事件提供了重要的见解。在《自然通讯》杂志上，Tarder-Stoll 等人描述了他们如何使用沉浸式虚拟现实和功能 MRI 来探索大脑表示特定环境的时间结构的方式。参与者在 VR 环境中浏览 16 个自然环境，并被要求创建故事以记住环境序列，这些序列的内容和顺序各不相同。一天后，他们看到了一个环境的图像，并被要求在进行 fMRI 扫描的同时，预测序列中即将到来的环境，即未来的几个步骤。作者使用多体素模式相似性分析来确定在预期期间表示哪些环境。该研究显示了未来和过去环境的同时表示，这些环境在海马体中比在早期视觉皮层中更广泛，并且是特定于上下文的。当海马体对遥远环境的表征受到抑制时，参与者也能更快地预测未来事件。这些发现与内侧颞叶预测处理的许多理论一致。

原始参考：Nat. Commun.15， 9094 （2024）