On the costs of refocusing items in working memory: A matter of inhibition or decay?

Abstract

The present study investigates the mental processes that are applied to previously attended items of working memory. In an object-switching task, participants counted the number of sequentially presented objects. In Experiment 1 the processing time increased when the object category switched from the prior trial compared to a repetition. More importantly, the further in the past the last instance of a current category was presented, the more processing time was necessary—an observation suggesting passive decay rather than inhibition of previously attended items. However, results differed when only two object categories were employed. Experiment 2 suggests that the lack of a clear indication of decay with small numbers of categories was due to participants’ expectancy of category switches rather than repetitions. Taken together, the results suggest that working memory items become less accessible the longer they have not been attended to, when strategic processes are controlled.

Acknowledgements

We thank two anonymous reviewers for helpful comments on an earlier draft of this manuscript, and Sebastian Krüger for technical support. Parts of this work were supported by a grant of the German Research Foundation (DFG) to Wilfried Kunde.

Notes

¹As a consequence, the interval between items (ITI) varied with RT. One could argue that with long RTs (long ITI) the decay of previously focused items would be more progressed than with short RTs, which consequently requires less inhibition of the previous item (see Gade & Koch, 2005, for this idea in task switching). Yet keeping the ITI constant would mean to vary the response–stimulus interval (RSI = ITI – RT). This would open the door for other uncontrollable effects. Participants might simply press the spacebar once they perceptually recognised the current category, and update the counter during the (rather long) RSI. Also this would allow more preparation for the next item the shorter the RT in the current trial, obviously compromising the interpretation of the RT measure as well. We therefore opted to keep procedural details of the experiment as close as possible to previously published work on item switching (Bao et al., 2006; Oberauer, 2003).

²We thank a reviewer of the manuscript for this suggestion.