The Subjective Visual (SVV) and Haptic (SHV) Vertical indicate the perceived direction of gravity. Previous methods have relied on participants adjusting a rod manually, potentially introducing movement-related confounds. Here we used robust psychophysical techniques to directly compare the SVV, SHV, and a novel multimodal paradigm.

Participants judged if a probe rod (30.5 cm long, 0.9 cm diameter) was tilted to the right (clockwise) or left (counterclockwise) with respect to gravity, during whole-body tilt 30° leftwards in the roll plane. A motor set the orientation of the rod; tested orientations were chosen using a PSI adaptive staircase optimized to estimate the reliability of each judgment. For SVV, the central 5 cm of the rod was illuminated and viewed through a diffusing screen that reduced the reliability of judgments to a level comparable to SHV judgments. For SHV, participants explored the rod by touch. In a third, multimodal condition, participants used both visual and haptic cues simultaneously.

The standard deviations of the SVV, SHV and multimodal estimates were 3.9° ± 0.2°, 5.5° ± 0.9°, and 2.9° ± 0.2° respectively. Across subjects, the variability and orientation of combined-cue estimates were consistent with the maximum likelihood estimate model.

We conclude that, as when comparing haptic and visual information for shape (Ernst and Banks, 2002), perceived orientation also results from optimal combination of available cues and that this estimate is available for making cross-modal judgements with the perceived direction of gravity.

multisensory perception; subjective vertical; perceived vertical; Maximum Likelihood Estimate; Subjective visual vertical; subjective haptic vertical; visual; haptic; sensory integration