Search

Practice testing is a highly robust learning strategy that promotes long-term retention, especially in comparison to more passive strategies such as restudying—a finding referred to as the testing effect. However, learners do not always appreciate the memorial benefits of practice testing over restudying, which could limit their use of practice testing during self-regulated learning. The current investigation explored the extent to which learners’ metacognitive judgments about the testing effect can be improved via test experience, direct instruction, or a com-bination of both techniques. Prolific participants underwent two learning cycles. In the first cycle, participants were randomly assigned to either (a) experience a testing effect in their own memory performance (i.e., study unrelated word pairs, practice half the pairs through restudying and half through testing with correct-answer feedback, complete a critical test on the pairs, and receive feedback regarding their performance after using each strategy); (b) imagine they had to learn word pairs and read a passage on the purported benefits of practice testing; or (c) undergo both procedures. In the second cycle, all participants learned a novel set of word pairs. Across both learning cycles, participants estimated memory performance for material learned through testing versus restudying. Both test experience and direct instruction—independently and in combination—led to more accurate memory estimates across learning cycles, but no technique was more effective than the other. In summary, people can learn about the memorial benefits of practice testing when they experience a testing effect on their own memory performance and/or when they receive instruction about its benefits.

When learning material that differs with respect to importance, leaners tend to remember more-important over less-important material (Castel et al., 2009; 2011; 2013). For instance, when studying words that differ by how many points (e.g., 1-10) each will be worth on an upcoming test, learners differentially encode and better remember higher-value items (e.g., FENCE: 10, SWIM: 9) compared to lower-value items (e.g., APPLE: 1, WEIGHT: 2); this effect is known as value-directed remembering (VDR; see Knowlton et al., 2022 for a review). The focus of most of the work in this field has been on memory outcomes, and in only a few instances have researchers explored the impact of value on learners’ self-regulated learning. Prior work has shown that learners tend to spend more time studying high-value word pairs over lower-value word pairs (Soderstrom & McCabe, 2011, Experiment 2) and choose to restudy high-value single words more often than lower-value single words (Middlebrooks & Castel, 2018, Experiment 2). Research with basic materials is informative; however, much less is known about how value influences learners’ study choices with educationally-relevant materials. In day-to-day life, a common scenario in which one may encounter value-associated information is when using a scoring rubric. To illustrate, instructors might use a rubric to score a writing assignment for a class, and this rubric may indicate, for example, that the introduction section is worth 30 out of a possible 100 points on the assignment, whereas the method section is worth 20 points. Some prior work has found that giving learners access to the rubric by which their assignment will be scored increases learners’ self-reported, self-regulated learning. Specifically, compared to learners who do not have access to a rubric during preparation, learners who do report engaging in more monitoring, planning, and evaluating of their learning during study (Panadero & Romero, 2013). Thus, making rubrics available to learners as they prepare for an assignment may benefit them by making the expectations for that assignment clear, allowing learners to plan their approach accordingly (Panadero, Tapia, & Huertas, 2013; Reddy & Andrade, 2010). As stated by Panadero and Jonsson (2020), the majority of research on how learners use rubrics has relied on self-report measures and personal anecdotes. Moreover, there exists a wide variety of type and quality of rubrics used in prior work, with some rubrics containing subjective criteria (e.g., “all or most of the instruction involves active engagement on the part of the students”), some containing value-based information (e.g., “active engagement is worth 4 points”; Andrade & Du, 2005), and some containing script-like, scaffolded instructions (e.g., “underline key phrases in the rubric with colored ink, then, underline evidence in your draft as having met the standard with the same color ink”; Andrade, 2010; 2017). The current study, therefore, aims to systematically measure how value as provided by a rubric influences self-regulated learning. By doing so, we will apply the VDR framework to a new, realistic context: using a rubric to prepare for a writing assignment. Specifically, we will examine how rubrics that differ with respect to the information they contain regarding points, influence learners’ studying decisions as they prepare for a writing assignment about mineral formation. Participants will be randomly assigned to one of four rubric groups: 1. Varied points rubric: two high-point (12 point) concepts, one medium-point (8 points) concept, and two low-point (4 points) concepts 2. Reverse-varied points rubric: same as varied points rubric but with opposite high-point (12 points) and low-point (4 points) concepts 3. Constant points rubric: all concepts associated with medium-points (8 points) 4. No points rubric: no points provided (only concept names appear) During the task, participants will learn about the process of mineral formation by reading about 5 concepts (from Ariel et al., 2021) in anticipation of completing a writing assignment. Participants can spend as much time as they want preparing for and completing the writing assignment. While preparing for the writing assignment, participants will see a rubric, 5 concept names, and the question, “which concept would you like to study?”. Participants can choose to study any of the 5 concepts by clicking on its name. Once a concept has been selected, a brief description of the concept will appear on the screen, and participants can spend as much time reading it as they want. For example, if participants select “Compounds”, they will be directed to a page containing a paragraph about compounds (in the context of mineral formation). When participants are finished studying a concept, they will be redirected to the choice-screen. After studying at least one concept, the choice-screen will also have an option that allows participants to move onto the writing assignment. Participants can study any concept they want, in any order they want, and revisit a concept as many times as they want. Beyond the requirement that participants must study one concept, participants can also choose to skip any concept (i.e., they can skip 4 of 5). On the writing assignment, which will be self-paced, participants will be asked to write as much as they can about the process of mineral formation underneath each of the 5 concept names. Following the writing assignment, participants will answer a series of self-report questions such as, “how did you determine you were ready for the writing assignment”, as well as a series of demographic questions.

Making judgments of learning (JOLs) while learning related word pairs can enhance performance on tests that rely on cue-target associations (e.g., cued recall) compared to studying alone. One possible explanation for this positive JOL reactivity effect is that the prompt to make JOLs – which typically occurs halfway through presentation of each pair – may encourage learners to devote more attention to the pair during the second half of the encoding episode. To investigate this idea, an online sample of participants (Experiment 1) and undergraduate students (Experiment 2) studied a set of moderately related word pairs (e.g., dairy – cow) in preparation for a cued recall test. Some participants made JOLs for each pair halfway through presentation, whereas other participants did not. As well, some participants were presented with a fixation point half-way through presentation, whereas other participants were not. The goal of this fixation point was to simulate the possible “reorienting” effect of a JOL prompt halfway through each encoding episode. In both an unsupervised online context and supervised laboratory context, cued recall performance was higher for participants who made JOLs compared to those who did not make JOLs. However, presenting a fixation point halfway through presentation of each pair did not lead to reactive effects on memory, which suggests that positive JOL reactivity for related pairs is not solely driven by attentional reorienting during encoding.

Investigations of learners' beliefs about pretesting as a function of experience and external support.

What constrains people’s ability to learn from experience about the effectiveness of practice testing versus restudying for memory (i.e., the testing effect)? Across two cycles, participants studied word pairs, practiced each pair through either restudying or testing, predicted how many pairs they would recall for each strategy, then completed a critical test on the pairs. During this test, participants either received feedback about the number of pairs they had correctly recalled or made postdictions about their performance for each strategy (i.e., generated their own feedback). During both cycles, participants predicted they would recall an equivalent number of tested and restudied pairs, although they actually recalled more tested pairs. However, when participants experienced a larger testing effect, they estimated recall performance more accurately for each strategy and updated their knowledge about the testing effect. Thus, peoples’ ability to learn from experience about the testing effect is primarily constrained not by a failure to initiate the metacognitive processes required to monitor and track recall performance by strategy, but by the metacognitive burden of discriminating between small differences in recall between tested versus restudied material. In summary, people can learn from experience about the testing effect when the metacognitive burden is lifted.

Do JOLs still influence recall when performance is similar for both category and letter-cued words?

Does knowledge of cue or strategy effectiveness influence the effects of JOLs on recall performance? If so, how?

Why does making JOLs influence subsequent memory, and when learners make judgments of learning (JOLs) for some items but not others, how is recall performance affected? To answer these questions, participants studied related and unrelated word pairs and made JOLs for half. Pair type was either randomly intermixed within a list (Experiment 1) or blocked (Experiment 2). We evaluated two hypotheses. The changed-goal hypothesis, proposed by Mitchum, Kelley, and Fox (J Exp Psychol Gen, 2016), states that making JOLs leads learners to notice differences in item difficulty and allocate more resources to learning easier pairs, ultimately leading to higher recall for easier (i.e., related) pairs and impaired recall for more difficult (i.e., unrelated) pairs. In contrast, the positive-reactivity hypothesis predicts increased recall performance for both related and unrelated pairs. As predicted by the positive-reactivity hypothesis, recall performance was higher for pairs that were judged versus not judged on both a mixed and blocked list of related and unrelated pairs. In Experiment 3, we evaluated one proximal mechanism for increased performance for judged pairs: The use of more effective encoding strategies during acquisition. Making JOLs did not influence strategy use, which suggests that the benefit of making JOLs on memory performance results from increased attention. These and other findings converge to support the claim that the requirement to monitor learning benefits memory.

Investigations of learners' beliefs about pretesting as a function of experience.