Higher Order Visual Location Learning Does Not Explain Multisensory Enhancement of Sound Localization (Reply to Vroomen and Stekelenburg 2021)

Abstract

In a recent study, we reported that multisensory enhancement (ME) of auditory localization after exposure to spatially congruent audiovisual stimuli and crossmodal recalibration in the ventriloquism aftereffect (VAE) are differently affected by the temporal stimulation frequency with which the audiovisual exposure stimuli are presented. Because audiovisual stimulation at 10 Hz rather than at 2 Hz selectively abolished the VAE but did not affect the ME, we concluded that distinct underlying neural mechanisms are involved in the two effects. A commentary on our paper challenged this interpretation and argued that the ME might have been spared simply because participants had acquired higher order knowledge about the loudspeaker locations from the visual stimulus locations in the ME condition, or because the ME was generally more reliable than the VAE. To test this alternative explanation of our results, we conducted an additional control experiment in which participants localized sounds before and after exposure to unimodal visual stimulation at the loudspeaker locations. No significant reduction of auditory localization errors was found after unimodal visual exposure, suggesting that higher order visual location learning cannot sufficiently explain the significant ME that was observed after audiovisual exposure in our previous study. These new results confirm previous findings pointing toward dissociable neural mechanisms underlying ME and VAE.

Keywords: crossmodal learning; recalibration; spatial hearing; ventriloquism.

FIGURE 1

Changes in sound localization error after exposure to unimodal visual stimuli relative to baseline. Positive values indicate error reductions (i.e., better performance) and negative values indicate error increases (i.e., worse performance) at posttest. Bars indicate group means, and superimposed dots indicate individual values. Three types of error measures were considered: absolute errors (reflecting accuracy and precision), which were calculated by averaging the absolute values of the single‐trial localization errors; constant errors (reflecting accuracy only), which were calculated by averaging the signed single‐trial localization errors; and variable errors (reflecting precision only), which were calculated as the standard deviation of the single‐trial localization errors.