Cross-modal correlations in ensemble averaging across vision, audition, and touch

The present study examined whether ensemble averaging of quantitative stimulus dimensions shows similar characteristics across sensory modalities. Although ensemble perception has been widely studied in vision, less is known about whether comparable processes operate in audition and touch. In particular, it remains unclear whether similarities across modalities reflect shared mechanisms or modality-specific processes. Two experiments were conducted using closely matched psychophysical procedures. In Experiment 1, participants estimated the average magnitude of stimulus sequences presented in the visual (size), auditory (loudness), and tactile (vibrotactile intensity) modalities. Responses were analyzed using psychometric functions, from which precision (slope) and bias (point of subjective equality; PSE) were derived. The precision of averaging performance remained relatively stable across set sizes in all modalities, suggesting that observers relied on processes that integrate information across the sequence rather than serial comparison of individual stimuli. Experiment 2 used a within-subject design with a newly recruited group of participants. Each participant completed visual, auditory, and tactile averaging tasks, and psychometric slopes and PSEs were estimated separately for each modality. Using an individual-differences approach, cross-modal correlations were then examined to test whether individuals who showed higher precision in one modality also showed higher precision in the other modalities. Precision measures were positively correlated across modalities, whereas bias measures were not. In addition, estimated averages tended to be larger than the physical means of the sequences in all modalities. Together, these findings suggest that ensemble averaging of quantitative stimuli involves partially shared magnitude-integration processes, while systematic biases remain influenced by modality-specific perceptual and decision factors.