Feature-level costs and memory load effects in matching pitch and amplitude modulation features from memory

This study investigated the nature of auditory working memory (AWM) capacity, specifically examining whether it is best described by discrete slot or continuous general resource models. Through three experiments, participants engaged in auditory matching tasks where they reproduced cued features of a target sound after a retention interval. Experiment 1 required participants to adjust either the amplitude modulation rate (AM), carrier frequency, or both features, with results showing decreased accuracy and precision when both features were cued compared to a single feature. Experiment 2 allowed both features to vary during response collection to test for a potential contribution from Garner interference. This failed to replicate the dual-feature cost observed in Experiment 1. Experiment 3 increased memory load by introducing trials with three to-be-remembered sounds. Matching performance was worsened under higher load, consistent with a general resource model of working memory. Also, for the matching of AM, accuracy showed a significant cost of having to store pitch information in memory along with AM. The study partially supports the continuous resource model, indicating flexible allocation of memory capacity across items when AM is matched. However, matching for pitch did not appear to differ among conditions when Garner interference was controlled for. The flexibility to prioritize within-item features in working memory may vary across different types of features.