Cognitive mechanisms of aversive prediction error-induced memory enhancements

While prediction errors (PEs) have long been recognized as critical in associative learning, emerging evidence indicates their significant role in episodic memory formation. This series of four experiments sought to elucidate the cognitive mechanisms underlying the enhancing effects of PEs related to aversive events on memory for surrounding neutral events. Specifically, we aimed to determine whether these PE effects are specific to predictive stimuli preceding the PE or if PEs create a transient window of enhanced, unselective memory formation. In a combined incidental encoding-fear learning task, participants (n=xxx) estimated aversive shock probabilities after unique stimuli. Physiological arousal and explicit PEs were measured during encoding to predict recognition memory tested either immediately after encoding (Experiment 3) or 24 hours later (Experiments 1-4). Our results show that PE-driven memory enhancement extends beyond predictive stimuli preceding the PE event to those encountered afterward. Furthermore, the retroactive memory enhancement induced by PEs may extend back longer than previously assumed, impacting stimuli presented 10 seconds before the PE. Importantly, our findings reveal that PE-related memory enhancement is specific to predictive stimuli, with uninformative stimuli not benefiting from PEs and even interfering with the PE-driven memory enhancement. This pattern demonstrates that PE effects are not unspecific but that PEs enhance memory for predictive stimuli encountered around a PE event. Notably, memory-enhancing effects of PEs persist even when controlling for changes in arousal. These findings provide insights into the cognitive mechanisms of PE-induced enhancements of memory, with potential implications for understanding aberrant emotional memory in fear-related disorders.

尽管预测误差（prediction errors, PEs）长期以来被认为在联想学习（associative learning）中至关重要，但新兴证据表明它们在情景记忆（episodic memory）形成中也发挥着显著作用。本系列四项实验旨在阐明与厌恶事件（aversive events）相关的PEs对周围中性事件记忆的增强效应背后的认知机制。具体而言，我们旨在探究这些PE效应是否特定于PE发生前的预测性刺激（predictive stimuli），抑或PE会创造一个短暂的、增强且非选择性的记忆形成窗口。在一项附带编码-恐惧学习（incidental encoding-fear learning）联合任务中，参与者（n=xxx）在接触独特刺激后估计厌恶电击概率（aversive shock probabilities）。编码过程中测量了生理唤醒（physiological arousal）和外显预测误差（explicit PEs），以预测编码后立即测试（实验3）或24小时后测试（实验1-4）的再认记忆（recognition memory）。我们的结果表明，PE驱动的记忆增强不仅限于PE事件发生前的预测性刺激（predictive stimuli），还延伸至PE事件发生后遇到的刺激。此外，PE诱导的回溯性记忆增强（retroactive memory enhancement）可能比之前假设的持续时间更长，影响PE发生前10秒呈现的刺激。重要的是，我们的研究结果揭示，PE相关的记忆增强特定于预测性刺激（predictive stimuli）；无信息价值的刺激（uninformative stimuli）不仅无法从PE中获益，甚至会干扰PE驱动的记忆增强。这种模式表明，PE效应并非非特异性的，而是PE会增强对PE事件前后遇到的预测性刺激（predictive stimuli）的记忆。值得注意的是，即使控制了唤醒水平的变化，PE的记忆增强效应依然存在。这些发现为理解PE诱导的记忆增强的认知机制提供了见解，并可能对理解恐惧相关障碍（fear-related disorders）中的异常情绪记忆具有启示意义。