The testing effect and analogical problem-solving

ABSTRACT

Researchers generally agree that retrieval practice of previously learned material facilitates subsequent recall of same material, a phenomenon known as the testing effect. There is debate, however, about when such benefits transfer to related (though not identical) material. The current study examines the phenomenon of transfer in the domain of analogical problem-solving. In Experiments 1 and 2, learners were presented a source text describing a problem and solution to read which was subsequently either restudied or recalled. Following a short (Experiment 1) or long (Experiment 2) delay, learners were given a new target text and asked to solve a problem. The two texts shared a common structure such that the provided solution for the source text could be applied to solve the problem in the target text. In a combined analysis of both experiments, learners in the retrieval practice condition were more successful at solving the problem than those in the restudy condition. Experiment 3 explored the degree to which retrieval practice promotes cued versus spontaneous transfer by manipulating whether participants were provided with an explicit hint that the source and target texts were related. Results revealed no effect of retrieval practice.

KEYWORDS:

• Testing effect
• transfer
• analogical reasoning

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 From informal conversations with colleagues, we settled on paying $1 for every 10 min of work. Because the study took participants longer than 10 min but less than 20 min, we paid each participant $2. This same payment structure explains participant payment in Experiments 2 and 3.

2 Time limits were implemented during the different phases of the experiment to equate time on task across the two groups. Participants excluded for computer timing errors experienced a trial time of three seconds or greater (as compared to the intended time limit).

3 One may wonder whether the pattern of results differs if analysis is limited only to those who can identify the relationship. Interestingly, the effect is numerically stronger (t(57) = 1.50, p = .07, d = .39) than when including all learners. A similar pattern is observed in Experiment 2. However, collapsing across both experiments, outcomes from a 2 (re-exposure: retrieval practice, restudy) × 2 (able to identify the relationship: yes, no) ANOVA failed to yield any interaction (p = .54).

Additional information

Funding

This work was supported by James S. McDonnell Foundation(Understanding Human Cognition).