2026年8月12日出版的《自然》杂志发表了美国科学家的一项最新研究成果。来自美国贝勒医学的Benjamin Y. Hayden小组开发出合成控制的神经基础。

为了理解连续选择的算法和神经基础，课题组人员研究了人类在执行连续捕猎任务时的行为和大脑活动。使用一种新开发的行为控制理论分解，该课题组人员发现追逐策略可以通过一个元控制器来很好地描述，该元控制器指示低级控制器的混合，每个低级控制器与特定的追逐目标相关联。前扣带皮层的神经元预测政策混合的主要变化，而海马神经元编码和更新支持早期计划的潜在政策状态。与此同时，眼窝额叶皮层的活动与任务当前价值结构的编码一致，而不是策略转换。总之，这些结果与海马体作为状态估计控制器，前扣带皮层作为元控制器，眶额皮质提供价值上下文信号的三边功能划分相一致。

据悉，自然主义的目标导向行为通常涉及针对动态变化目标的持续行动。正如微观经济学是离散选择的严格基础一样，控制理论也可以作为理解连续选择的基础。在连续环境中，行为是由目标的混合组成的，与选择最接近的类比是对目标特定控制策略的战略重新加权。

附：英文原文

Title: Neural basis of compositional control

Author: Chericoni, Assia, Fine, Justin M., Ismail, Taha S., Delgado Salazar, Gabriela, Franch, Melissa C., Mickiewicz, Elizabeth, Chavez, Ana G., Bartoli, Eleonora, Paulo, Danika L., Krishnan, Vaishnav, Hegazy, Mohamed, Goldman, Alica M., Lin, Lu, Banks, Garrett P., Giridharan, Nisha, Hasen, Mohammed, Provenza, Nicole R., Watrous, Andrew, Yoo, Seng Bum Michael, Sheth, Sameer A., Hayden, Benjamin Y.

Issue&Volume: 2026-08-12

Abstract: Naturalistic goal-directed behaviour often involves continuous actions directed at dynamically changing goals1,2,3. Just as microeconomics serves as a rigorous foundation for discrete choices, control theory can serve as a foundation for understanding choice in continuous ones3,4. In continuous contexts, behaviour is composed of blends of goals, and the closest analogue to choice is a strategic reweighting of goal-specific control policies5,6. Here, to understand the algorithmic and neural bases of continuous choice, we examined behaviour and brain activity in humans performing a continuous prey-pursuit task7. Using a newly developed control-theoretic decomposition of behaviour, we find that pursuit strategies are well described by a meta-controller dictating a mixture of lower-level controllers, each linked to specific pursuit goals. Neurons in the anterior cingulate cortex predict major changes in policy blends, whereas hippocampal neurons encode and update the latent policy state supporting early planning. Meanwhile, orbitofrontal cortex activity is consistent with an encoding of the current value structure of the task, rather than policy switching. Together these results are consistent with a tripartite functional division in which hippocampus serves as a state-estimating controller, anterior cingulate cortex serves as a meta-controller, and orbitofrontal cortex provides a value context signal.

DOI: 10.1038/s41586-026-10896-8

Source: https://www.nature.com/articles/s41586-026-10896-8