The production effect is consistent over material variations: support for the distinctiveness account

ABSTRACT

The production effect is the superior memory for items read aloud as opposed to silently at the time of study. The distinctiveness account holds that produced items benefit from the encoding of additional elements associated with the act of production. If so, then that benefit should be consistent regardless of item type. Three experiments, using three different sets of materials and three different methods, tested this hypothesis. Experiment 1, using recognition testing, showed consistent production benefits for high and low frequency words. Experiment 2, using free recall, showed consistent production increments for pictures and words. Experiment 3, using incidental learning, showed consistent production benefits for recognition of nonwords and words. Taken together, these results fit with the distinctiveness account: Production at encoding dependably adds information to the memory record, regardless of item type or method of testing, producing a consistently reliable memory benefit.

KEYWORDS:

• Memory
• production effect
• distinctiveness

Acknowledgements

We thank Jennifer Allman, Emily Bryntwick, Grace Hsiao, Molly Pottruff, and Jonathan Zimmermann for their assistance with the data collection.

Disclosure statement

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

Notes

1 After reporting all three experiments, and just prior to the General Discussion, we will consider the issue of the power in these experiments in more detail.

2 In many previous production experiments (e.g., MacLeod et al., 2010; Hourihan & MacLeod, 2008; Ozubko et al., 2012a), 2/3 of the items on the recognition test have been studied and 1/3 have been new. This makes the number of aloud, silent, and new items equal at test, but it also produces an unequal number of old and new items. We do know, though, from observing the benefit in recall experiments (Conway & Gathercole, 1987; Lin & MacLeod, 2012; Experiment 2 here) and in other recognition experiments in which the numbers of old and new items were equal (e.g., Forrin et al., Experiment 2, 2012) that this choice does not matter. Nor does counterbalancing the colours that signal aloud versus silent reading matter (see Lin & MacLeod, 2012). The extension of the production benefit to a fill-in-the-blanks test in Ozubko et al. (2012b, Experiment 3) further shows that the production effect does not rely on certain specific procedural features either during study or during test.

Additional information

Funding

This research was supported by Discovery Grant A7459 from the Natural Sciences and Engineering Research Council of Canada.