Too much of a good thing: frequent retrieval can impair immediate new learning

ABSTRACT

Interpolated testing can reduce mind-wandering and proactive interference, and improve note-taking. However, recent research using face-name-profession triads, has also shown that interpolated testing can impair new learning (Davis, Chan, & Wilford, 2017). In the current study, we further examined the impact of switching from testing to new learning, but with objectively-true materials. The study employed a 2 (Interpolated task: Test vs. Restudy) × 3 (Task-switch frequency: 0, 11, 35) between-participants design. In two experiments, participants restudied or retrieved originally-learned flag-country associations and learned new flag-capital (Experiment 1) or flag-export (Experiment 2) associations. Task-switch frequency varied such that participants switched to new learning trial(s) after every restudy/test trial (35-switches), after every three restudy/test trials (11-switches), or did not switch at all (0-switch). The results further demonstrate that retrieving previously-learned material can impair learning of new associations by replicating Davis et al. (2017) with objectively-true materials.

KEYWORDS:

• Task-switching
• interpolated testing
• testing effect
• retrieval practice
• test-impaired new learning

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 A pilot study was conducted wherein 23 participants were presented with forty flag-country-capital triads and were tested over all 40. Flags that produced a remarkably high rate of recall were excluded (e.g., Turkey, Sweden), as this indicated strong prior knowledge. We also excluded flags for which performance was at floor (e.g., Uzbekistan, Lesotho). The final eighteen flags had produced recall rates of 17% to 61% (versus 0% to 74% for all those pilot-tested).

2 The final sample size exceeded the minimum requirement described in our preregistration because we chose to run the experiment for a full semester (as we did in Experiment 1). Importantly, no analyses were conducted until data collection was complete.