Disruption of short-term memory by distractor speech: Does content matter?

Abstract

Four experiments replicate the finding that auditory distractors that are lexically identical to the visual target items dramatically increase the irrelevant-speech effect on serial recall. This effect was previously attributed to interference of incompatible order cues. The present results suggest that a different interpretation of this effect is required. Experiment 2 replicates the order congruence effect observed by Hughes and Jones (2005), but shows that this effect is most likely due to an attenuation of interference that is caused by strategic attention shifts to the nominally irrelevant material. Experiments 3 and 4 show that the between-stream similarity effect generalizes to a condition in which the distractor items were drawn from the same category as the targets, but were not identical to them. By showing that nonacoustic distractor features can increase interference in serial recall of lists of supposedly “meaningless” items such as digits or consonants, the results are most consistent with models that postulate an integration of short-term and long-term memory such as the embedded-processes model and the feature model and are inconsistent with classical structural accounts of memory.

Keywords:

• Working memory models
• Auditory distraction
• Irrelevant sound effect
• Categorical similarity
• Attention and perception

Acknowledgments

We thank Gunnar Regenbrecht and Robert Hauke for their help during data acquisition.

Notes

¹ An anonymous reviewer suggested that the failure to find an order congruence effect in the lag 5 condition may be due to structural limits of the focus of attention (if the capacity limit is four items, having a lag of five items of “unattended” information would exceed the focus, while having a lag of two items should not), which is also consistent with the embedded-processes model (Cowan, 2000).