When does practice testing promote transfer on deductive reasoning tasks?

Abstract

Tran et al. (2015) evaluated whether engaging in practice testing versus restudy promotes transfer and concluded that testing does not enhance performance on a deductive reasoning task. The current research further evaluated Tran et al. (2015) and an alternative explanation for the observed effect-namely, that testing did not enhance memory for the information required to complete the deductive reasoning task. Learners studied premises from four scenarios and engaged in testing or restudy during practice. Following practice, learners completed a multiple-choice test that required them to deductively reason from premises. Results replicate Tran et al. (2015) by demonstrating that fill-in-the-blank testing does not promote transfer on a deductive reasoning task. Results also provide evidence suggesting that deductive reasoning depends on complete memory for requisite pieces of information and that testing effects on deductive reasoning are more likely to obtain under conditions that yield a memory advantage for testing versus restudy.

The Effect of Retrieval Practice on Transitive Inference

“…It is possible then, that if the retrieval practice used in the second phase of the current paradigm is of a sort that encourages inter-element processing (as free recall does), then retrieval practice might enhance the development of the abstracted knowledge structure and ultimately improve TI on the final test. Evidence relevant to this possibility comes from Wissman et al (2018), who investigated deductive inference using the materials and procedures of Tran et al (2015). Wissman et al found that for fill-in-the-blank retrieval practice, there was no enhancement to subsequent deductive inference (replicating Tran et al) but when cued recall was used as retrieval practice, a positive testing effect was found on deductive inference.…”

“…Recently, Eglington and Kang (2018) replicated the results of Tran et al (2015) in showing no testing effect on inference under the study and testing conditions used in Tran et al, but also showed that under somewhat different learning conditions, a testing effect on inference questions could be obtained. Likewise, Wissman, Zamary, and Rawson (2018) reported mixed results, under some conditions finding no testing effect on deductive inference but in others reporting a significant effect. We return to a more complete discussion of the latter two studies later in the paper.…”

“…1The relevant studies of deductive inference are listed under “Inference Questions Only” on p. 726, Pan and Rickard (2018), although it should be noted that this review does not include the results of Wissman et al (2018).…”

Testing enhances both memorization and conceptual learning of categorical materials.

“…However, their failure to observe the effect should be interpreted with caution because the researchers also failed to obtain a testing effect for verbatim memory, as demonstrated in their Experiment 4, in which the final test included both studied and nonstudied exemplars. As reported in Wissman, Zamary, and Rawson (2018), a necessary condition of transfer effects (i.e., conceptual learning) is to show that practice testing enhances memory of all the information needed to perform well on the transfer test.…”

Building a testing-based training paradigm from cognitive psychology principles.

“…Although extensive work has shown that retrieval practice can indeed aid retention (e.g., Carpenter et al, 2009;Carrier & Pashler, 1992;Dunlosky et al, 2013;Kang et al, 2013;Karpicke & Roediger, 2008;Pyc & Rawson, 2010;Roediger & Butler, 2011;Roediger & Karpicke, 2006a, 2006b, most of this literature is limited to direct memorization and does not typically involve true concept learning. Moreover, the limited work that has been conducted on this topic has yielded inconclusive results, as some of this work has shown a modest benefit of retrieval practice and testing on concept learning and transfer (Butler, 2010;Butler et al, 2017;Eglington & Kang, 2018;, but other studies have failed to replicate this finding (Peterson & Wissman, 2018;Tran, Rohrer, & Pashler, 2015;van Gog & Kester, 2012;Wissman, Zamary, & Rawson, 2018). It is therefore an empirical question as to whether these principles can be used to help people learn ecologically valid, complex concepts.…”