Abstract
Most research on visual-haptic integration has reported that visual perception dominates haptic perception. Accordingly, this phenomenon has been called “visual capture”. However, a few studies have shown a clear influence of haptics on vision, demonstrating that vision does not necessarily completely capture haptics. We hypothesized that the degree of dominance is determined by the statistical reliability of the available sensory information and can be represented by a linear weighting scheme. To test this hypothesis, we first measured visual and haptic size discrimination thresholds separately using a 2-IFC paradigm. We added noise to the visual display in order to vary the difficulty of the visual discrimination task. From those within-modality thresholds, we calculated predictions for the weights for cross-modal discrimination assuming that the weights depend only on the sensory noise. To determine the actual degree of visual dominance, the same stimuli were presented to both modalities simultaneously while specifying a size discrepancy. The perceived size of this conflicting visual-haptic stimulus was determined from PSE settings using a 2-IFC paradigm as similar as possible to the within-modal tasks. We found that the perceived size of the cross-modal stimulus was always in-between the sizes specified visually or haptically, indicating that both the visual and haptic information contribute to the combined percept. Furthermore, the cross-modal discrimination thresholds were always smaller than the individual visual and haptic thresholds, indicating sensory fusion. Finally, the weights derived from subjects' cross-modal settings were well predicted from the individual visual and haptic size discrimination thresholds (r = 0.985). We conclude that this form of cross-modal perception can be described by a linear weighting scheme in which the more reliable information is weighted more heavily.