提取练习重塑自然情境连续再认数据集中的邻近记忆
Retrieval practice reshapes neighboring memories in a naturalistic continuous-recognition dataset
提取会促进还是损害相关记忆，以往证据多来自实验者设定关系的实验室范式。研究分析自然场景数据集中 72,402 次目标测验，把目标再认建模为邻近图像先前测验的函数。最近邻分析显示局部分离，同场次邻图提取正确预测目标再认更好，跨场次邻图正确和同场次邻图错误则预测更差；累积加权历史模型显示正确提取历史有更广的正向关联，同场次错误历史为负向关联。
Retrieval can facilitate or impair related memories, but most evidence comes from laboratory paradigms with experimenter-defined relationships. Here, we asked whether retrieval-related associations emerge in temporal neighborhoods of a large naturalistic continuous-recognition dataset. We analyzed 72,402 target tests from the Natural Scenes Dataset and modeled target recognition as a function of prior tests of neighboring images. Closest-neighbor analyses showed a local dissociation: accurate same-session neighbor retrieval predicted better target recognition, whereas accurate across-session neighbors and inaccurate same-session neighbors predicted poorer target recognition. Accumulated weighted-history models showed broader positive associations for accurate retrieval history and negative associations for inaccurate same-session history. These findings suggest that retrieval-related spillover in naturalistic memory depends jointly on retrieval success, session context, and neighborhood scale.
中文翻译还没有生成，先显示原文。
来源：Scientific Reports · nature.com