学习判断(judgments of learning; JOLs)是指学习者对所学内容在未来测验中能够被成功提取可能性的预见性评估。以往研究主要聚焦于考察JOL的准确性及其内在产生机制等问题，然而近期研究表明JOL能够反应性改变记忆成绩本身，这种现象被称为反应性效应(the reactivity effect)。在以往研究中，个体学习完每个项目之后对该项目进行JOL，这种方式本质上是以学习者自我报告的形式来测量个体的元记忆监测能力。在个体进行这些自我报告时，其往往伴随对自身记忆过程的有意识监测，从而导致与不测量JOL的条件相比（即no-JOL条件），测量JOL的过程可能增加了人们对自身记忆过程的元意识成分，进而引起所学内容的记忆状态发生变化，引发反应性效应。

尽管先前研究发现了JOL对项目自身记忆成绩存在反应性效应，并提出多种理论用于解释该效应的内在认知机制。但是，迄今为止，尚未有研究考察JOL是否对项目间关系记忆产生反应性影响。项目间关系记忆对人们的日常生活（尤其是学习与教育）具有重要作用，其决定了学习者信息表征的组织形式。然而，反应性效应的现有理论均聚焦于项目自身记忆而提出的，其无法预测并解释JOL对项目间关系记忆的反应性影响，并且这些理论也难以解释测试类型对反应性效应的调节作用。因此，本研究基于项目自身与关系理论首次考察JOL对项目间关系记忆的反应性效应，并揭示该效应的内在认知机制。

项目自身与关系理论在先前大量研究中得到了反复验证，但是反应性效应的研究领域中从未考察该理论能否适用于解释JOL的反应性效应。根据项目自身与关系理论，我们预期：学习时做JOL会促进个体对项目自身特征的编码，从而促进项目自身记忆，产生积极反应性效应。然而，由于个体认知资源是有限的，当个体将更多的认知资源分配给项目自身特征加工时，分配给项目间关系加工的认知资源会相应减少，因此做JOL会破坏项目间关系记忆，产生消极反应性效应。

为了考察JOL是否对项目间关系记忆产生反应性影响以及项目自身与关系理论能否用于解释反应性效应的内在产生机制，本研究共开展了三项研究（包括11个实验和1项元分析），分别基于项目间时序关系（研究一）和项目间语义关系（研究二和研究三）深入探讨JOL对项目间关系记忆的反应性效应，以及该效应在不同关系记忆形式间是否具有一般性。本研究的主要结论如下：

（1）研究一考察了JOL对项目间时序关系记忆的反应性影响。研究1.1-研究1.3发现JOL对项目间时序关系记忆产生消极反应性效应。研究1.4和研究1.5发现，做JOL促进了项目自身记忆，但同时破坏了项目间时序关系记忆，且这两种反应性效应存在分离现象，进而支持了项目自身与关系理论。研究1.6的元分析结果表明，做JOL显著降低了顺序重建测试成绩，但提高了自由回忆和再认测试成绩，并且再认测试的积极反应性效应强于自由回忆测试的积极反应性效应，元分析的结果再次支持了项目自身与关系理论。

（2）研究二和研究三进一步考察JOL对项目间语义关系记忆的反应性效应，并检验不同关系记忆形式（即项目间时序关系记忆和项目间语义关系记忆）之间的反应性效应是否具有一般性机制。研究2.1和研究2.2采用押韵词对（且不同词对的目标词来自不同的语义类别）作为实验材料，发现做JOL破坏了项目间语义关系记忆，表明JOL对项目间语义关系记忆同样产生消极反应性效应。研究2.3通过降低项目间语义关联的凸显性，发现JOL对项目自身记忆产生积极反应性效应。研究二的结果再次支持了项目自身与关系理论，并进一步表明JOL对不同的项目间关系记忆的反应性影响具有一般性。

（3）研究三采用DRM范式，再次验证JOL对项目间语义关系记忆的反应性效应，并首次考察JOL对虚假记忆的反应性影响。研究3.1发现，逐项JOL破坏了DRM词语间的语义关联加工，进而降低虚假记忆的产生。与此同时，逐项JOL促进了已学词的项目自身加工，项目自身记忆出现积极反应性效应。研究3.2和研究3.3通过改变DRM词列呈现形式和JOL类型，发现整体JOL促进项目内语义关系加工，但其破坏项目间语义关系加工。研究三的结果再次支持项目自身与关系理论。

综上，本研究首次全面揭示出JOL对项目间关系记忆的反应性效应及其认知机制，并发现做JOL促进了项目自身加工，但同时破坏了项目间关系加工，并且这种破坏作用对不同的项目间关系记忆具有一般性。本研究结果不仅为项目自身与关系理论在反应性效应研究领域的应用提供了直接的证据支持，而且为元记忆监测策略作为提升学生学习效率的教育实践提供了重要参考。

Judgments of learning (JOLs) refer to learners' predictive evaluations of the likelihood that they will successfully retrieve learned content in future tests. Previous research has mainly focused on examining the accuracy of JOLs and the underlying mechanisms. However, recent studies have shown that JOLs can affect memory performance itself, which is called the reactivity effect. In previous studies, individuals made JOLs after learning each item, which essentially measured their metacognitive monitoring ability in the form of self-reports. When individuals make these self-reports, they often have conscious experience of their own memory processes, which may increase their awareness of their memory processes and lead to changes in memory performance compared to conditions without JOL measurement (i.e., no-JOL condition), causing the reactivity effect.

Although previous studies have found the reactivity effect of JOLs on memory performance and proposed various theories to explain the underlying cognitive mechanisms, to date, no study has examined whether JOLs have a reactivity effect on memory for item-item relations. Memory for item-item relations plays an important role in people's daily lives, especially in learning and education, and determines the organization of learners' knowledge representations. In addition, existing theories of the reactivity effect cannot predict or explain the reactivity effect of JOLs on memory for inter-item relations, and these theories also have difficulty explaining the moderating effect of test types on the reactivity effect. Therefore, the current research conducted the first investigation based on the item-specific relational account to examine the reactivity effect of JOLs on inter-item relational memory and uncover the cognitive underpinnings.

The item-specific relational account suggests that making JOLs during learning promotes encoding of item-specific features, thereby facilitating memory for items themselves and producing a positive reactivity effect. However, since cognitive resources are limited, when individuals allocate more cognitive resources to processing item-specific features, the cognitive resources allocated to processing inter-item relations will correspondingly decrease. Therefore, making JOLs will impair memory for inter-item relations and produce a negative reactivity effect. The item-specific relational account has been repeatedly validated in previous studies, but it has never been systematically examined in the field of the reactivity effect.

To examine whether JOLs have a reactivity effect on memory for item-item relations and whether the item-specific relational account is applicable to explain the reactivity effect of JOLs, this study conducted three studies (including 11 experiments and one meta-analysis) based on item-item temporal relations (Study 1) and item-item semantic relations (Studies 2 and 3) to investigate the reactivity effect of memory for item-item relations and whether this effect is generalizable across different forms of relational memory. The main conclusions of this study are as follows:

(1) Study 1 examined the reactivity effect of JOLs on memory for temporal inter-item relations. Studies 1.1-1.3 found that making JOLs produced a negative reactivity effect on memory for temporal inter-item relations. Studies 1.4 and 1.5 found that making JOLs facilitated memory for items themselves but impaired memory for temporal inter-item relations, and these two reactivity effects were dissociated, supporting the item-specific relational account. The meta-analysis results of Study 1.6 showed that making JOLs significantly reduced memory for temporal inter-item relations.

(2) Studies 2 and 3 further examined the reactivity effects of JOLs on semantic inter-item relational memory, and tested whether the reactivity effect on inter-item relational memory is generalizable from temporal to semantic inter-item relational memory. To achieve these aims, Studies 2.1 and 2.2 employed rhyming word pairs (with target words from different semantic categories) as experimental stimuli, and found that making JOLs disrupted memory for semantic relations between items, reflecting a negative reactivity effect on memory for semantic inter-item relations. But making JOLs produced minimal reactivity effect on free recall of target words. Study 2.3 found that, when the saliency of semantic relations between target words was reduced, making JOLs produced a positive reactivity effect on free recall of target words. The results of Study 2 once again support the item-specific relational account.

(3) Study 3 utilized the DRM paradigm to further examine the reactivity effect of JOLs on memory for semantic inter-item relations. Study 3.1 found that item-by-item JOLs disrupted processing of semantic relations among DRM words, reduced semantic activation of critical lures, and consequently reduced occurrence of false memory (i.e., reduced false alarm rates of critical lures). At the same time, item-by-item JOLs facilitated processing of the items themselves, resulting in a positive reactivity effect on item memory. Studies 3.2 and 3.3 changed presentation format of DRM word lists and asked participants to make global JOLs for a whole word list (instead made JOLs for each item). The results showed that making global JOLs facilitated processing of semantic intra-item relations, but instead disrupted processing of semantic inter-item relations. The results of Study 3 further support the item-specific relational account.

In summary, the current research provides the first comprehensive examination of the reactivity effect of making JOLs on inter-item relational memory (both temporal and semantic), and demonstrates that making JOLs facilitates item-specific processing, but disrupts inter-item relational processing. These findings not only provide direct evidence supporting the validity of the item-specific relational account for explaining the cognitive underpinnings of the reactivity effect, but also provide important implications for guiding educational practices aiming to improve individuals’ learning efficiency through metacognitive monitoring strategies.