原文标题：Why Testing Improves Memory: Mediator Effectiveness Hypothesis
原文作者：Mary A. Pyc* and Katherine A. Rawson
发表：science，2010.10.15
前文Science报道，通过实验证明retrieval（测试，考试）的效果要比restudy的效果要好。但是没有介绍，为什么这种现象产生。这边报道提供了一种假设叫做”媒介假设“。
文章开头：
An intuitive but incorrect assumption isthat learning only occurs during study and that testing is useful only forevaluating the state of memory (1). However, testing improves memory, asdemonstrated with a wide variety of materials and learners (2). Given thewealth of research establishing the empirical benefits of testing,it issurprising that the mechanisms underlying testing effects are not wellunderstood.
所以这篇文章是为了解释考试为什么提高记忆。
提出假设：
We propose the mediator effectivenesshypothesis, stating that testing improves memory by supporting the use ofmore-effective mediators during encoding (a mediator is a word, phrase, orconcept that links a cue to a target).What makes a mediator effective? Two keyfactors are mediator retrieval (i.e., mediator is recallable when prompted withthe cue) and mediator decoding (i.e., mediator elicits the target from memory)(3). The mediator effectiveness
hypothesis assumes that mediators generatedduring testing (versus restudy only) are more likely to be subsequentlyretrieved and decoded, increasing recall of target responses.
媒介假设---媒介可以是一个词、短语或者一个概念。这种媒介起作用必须有两个关键因素：1.根据提示能够回忆起这个mediator；2.通过mediator可以解码---可以回忆去目标。
设计实验：
To test this hypothesis, we presented 118participants with 48 Swahili-English translation pairs (e.g., wingu-cloud) foran initial study trial and then three blocks of practice trials. In the test-restudygroup, each practice trial for an item involved a cued recall test followedimmediately by restudy. In the restudy group, each trial involved only restudy.
分为两个组：test-restudy组，restudy组。
实验结果：
test-restudy practice produced almost a threefold increase in final testperformance in the C group (Fig. 1A) As converging evidence, the same patternis apparent when examining recall as a function of mediator retrieval in theCMR group (Fig. 1B).
讨论：
Results support the mediator effectivenesshypothesis and offer one theoretical explanation for why testing is beneficialfor memory: Mediators generated during encoding are more effective (i.e., morelikely to be retrieved and decoded) with test-restudy versus restudy practice.We are not claiming that mediator effectiveness is the only mechanismunderlying testing effects. However, mediator effectiveness may be an importantcontributor.
Why did testing yield more-effectivemediators?
Successfully retrieving mediators duringpractice may enhance their memory strength. Additionally, retrieval failuresduring encoding may promote shifting from less- to more-effective mediators(5), and retrieval failure occurs during testing but not restudy.
test-restudy相比restudy的方法更有效，可能是回顾媒介可以提高记忆强度，而且回忆媒介失败了也没关系，至少restudy的时候可以讲这种mediator从无效进步到有效。
