Auditory aspects of multisensory working memory are retrieved better than visual aspects.

Working memory (WM) reflects the transient maintenance of information in the absence of external input, which can be attained via multiple senses separately or simultaneously. Pertaining to WM, the prevailing literature suggests the dominance of vision over other sensory systems. However, this imbalance may be attributed to challenges of finding stimuli that are represented in comparable ways across modalities. Here, we addressed this methodological problem by using a balanced multisensory "retro-cue" WM design. The to-be-memorized stimuli consisted of combinations of auditory (ripple sounds) and visuospatial (Gabor patches) patterns, which have been shown to undergo similar transformations during WM encoding and retrieval. Using a staircase procedure, the auditory ripple velocities and spatial frequencies of Gabor patches were adjusted relative to each participant's just noticeable differences (JND) separately in each modality, before the main task. The task was to audiovisually compare the probes to the memorized items. In randomly ordered trials, the probe either fully matched or differed from the memory item auditorily, visually, or audiovisually. The participants correctly rejected a significantly larger number of auditory non-match probes than visual non-match probes. Our findings suggest that, in the case of inter-sensory competition during feature maintenance, auditory attributes of multisensory WM items can be retrieved more precisely than their visual counterparts when complexity of the content and task demands are bimodally equated.