Structural differences along the hippocampal long axis are believed to underlie meaningful functional differences. Yet, recent data-driven parcellations of the hippocampus subdivide the hippocampus into a 10-cluster map with anterior-medial, anterior-lateral, and posteroanterior-lateral, middle, and posterior components. We tested whether task and experience could modulate this clustering using a spatial learning experiment where male and female participants were trained to virtually navigate a novel neighborhood in a Google Street View-like environment. Participants were scanned while navigating routes early in training and after a 2 week training period. Using the 10-cluster map as the ideal template, we found that participants who eventually learn the neighborhood well have hippocampal cluster maps consistent with the ideal—even on their second day of learning—and their cluster mappings do not deviate over the 2 week training period. However, participants who eventually learn the neighborhood poorly begin with hippocampal cluster maps inconsistent with the ideal template, though their cluster mappings may become more stereotypical after the 2 week training. Interestingly this improvement seems to be route specific: after some early improvement, when a new route is navigated, participants' hippocampal maps revert back to less stereotypical organization. We conclude that hippocampal clustering is not dependent solely on anatomical structure and instead is driven by a combination of anatomy, task, and, importantly, experience. Nonetheless, while hippocampal clustering can change with experience, efficient navigation depends on functional hippocampal activity clustering in a stereotypical manner, highlighting optimal divisions of processing along the hippocampal anterior-posterior and medial-lateral axes.

据信，沿海马长轴的结构差异是有意义的功能差异的基础。然而，最近数据驱动的海马分区将海马细分为 10 个簇图，其中包含前内侧、前外侧、后前外侧、中间和后部组件。我们使用空间学习实验测试了任务和经验是否可以调节这种聚类，其中男性和女性参与者接受训练，在类似谷歌街景的环境中虚拟地导航新的社区。参与者在训练初期和两周训练期后导航路线时接受了扫描。使用 10 簇图作为理想模板，我们发现最终很好地学习了邻域的参与者的海马簇图与理想状态一致（即使在学习的第二天），并且他们的簇映射在两周的训练中没有偏离时期。然而，最终学习邻域的参与者一开始的海马聚类图与理想模板不一致，尽管他们的聚类图在两周的训练后可能会变得更加刻板。有趣的是，这种改进似乎是特定于路线的：经过一些早期的改进，当导航新路线时，参与者的海马图会恢复到不那么刻板的组织。我们得出的结论是，海马聚类不仅仅取决于解剖结构，而是由解剖结构、任务以及重要的经验的组合驱动。 尽管如此，虽然海马聚类可能会随着经验而变化，但有效的导航取决于以刻板方式进行的功能性海马活动聚类，强调沿海马前后轴和内侧外侧轴的最佳处理划分。