Testing can facilitate forgetting of tested items

Abstract

Previous research has shown that taking a test facilitates memory retention on later retesting, although facilitation is stronger when retesting is delayed. On the basis of the finding that testing prevents later forgetting without affecting memory recovery, we investigated immediate effects of taking a test on retrievability of a tested item. In two experiments, forty participants recalled studied items, then performed a retest of the studied items (test condition) and a non-episodic-memory task (distractor condition), and immediately afterward re-recalled the items. The test condition elicited more item losses (i.e. forgetting) than the distractor condition, whereas there were no condition differences in the number of item gains (i.e. remembering) and recall clustering. These results suggest that taking a test on a target item facilitates forgetting of the very same item for a short interval of time, a possible reason for the late appearance of the testing effect.

Keywords:

• Gain
• loss
• memory
• testing effect

Notes

1 In order to investigate loss and gain items, the present study introduced the baseline test before the experimental manipulation. Because restudy after test has been demonstrated to enhance memory performance (Arnold & McDermott, 2013; Rickard & Pan, 2018; Wissman & Rawson, 2018), restudy after the baseline test would not be adequate to estimate the number of loss and gain items in the test condition. We thus introduced a non-episodic memory task as the control condition (see Chan & McDermott, 2007; Huff, Coane, Hutchison, Grasser & Blais, 2012), in which participants could not obtain feedback on their responses to the baseline test.

2 That there was no significant difference in the number of gain items across conditions may be attributable to the small sample size of the present study (N = 20). However, this same experiment did show a significant difference in the number of loss items; the effect sizes of gain items were respectively d = 0.21 and d = 0.36 in Experiments 1 and 2, whereas those of loss items were respectively d = 0.51 and d = 0.74. The larger effect of forgetting may be the reason why the testing effect is less likely to appear on immediate tests.

Additional information

Funding

This work was supported by JSPS KAKENHI Grant Number JP15K12044.