This work presents a statistical thermodynamics-inspired framework that summarizes multichannel EEG and behavior using macroscopic state variables (entropy, internal energy, temperature, Helmholtz free energy) to quantify stability and reconfiguration in neuropsychological systems. Applied to mother-infant EEG dyads performing the A-not-B task, these variables dissociate neural reconfiguration from behavioral success across a large set of model and feature configurations. Informational heat increases during environmental switches and decision errors, consistent with increased information exchange with the task context. In contrast, correct choices are preceded by lower temperature and higher free energy in the window, and are followed by free-energy declines as the system re-stabilizes. In an independent optogenetic dam-pup paradigm, the same variables separate stimulation conditions and trace coherent trajectories in thermodynamic state space. Together, these findings show that the thermoinformational framework yields compact, physically grounded descriptors that hold in both human and mouse datasets studied here.

翻译：本研究提出一个受统计热力学启发的框架，该框架利用宏观状态变量（熵、内能、温度、亥姆霍兹自由能）来总结多通道脑电图与行为数据，以量化神经心理系统的稳定性与重构过程。将此框架应用于执行A-not-B任务的母婴脑电图配对数据时，这些变量能在大量模型与特征配置中区分神经重构与行为成功。信息热在环境切换与决策错误期间增加，这与任务情境中信息交换的增强相一致。相比之下，正确选择前窗口内呈现较低温度与较高自由能，随后系统重新稳定时自由能下降。在一个独立的光遗传学母鼠-幼鼠范式中，相同变量能区分刺激条件并在热力学状态空间中描绘出连贯轨迹。综合来看，这些发现表明热信息学框架可产生紧凑且具有物理基础的描述符，适用于本研究所涉及的人类与小鼠数据集。