Spatial orientation is the sense of one's orientation and self-motion relative to the environment. To estimate spatial orientation, the nervous system combines information from the visual and vestibular senses. The vestibular system consists of otoliths, which sense gravity and other linear accelerations of the head, and semicircular canals, which sense angular accelerations. The visual system senses scene orientation and optic flow, which are both relevant to spatial orientation.

In experiments reported here, visual and vestibular stimuli were manipulated independently. Observers made judgments about body orientation or self-motion. Results show that visual and vestibular senses are highly complimentary, in that they provide very similar types of information. Results also show that self-motion stimuli influence body orientation perception, and body orientation stimuli influence self-motion perception.

Chapter 1 introduces a Bayesian model for the disambiguation of gravitoinertial force by visual cues. Results from an accompanying experiment show that visual pitch influences perception of linear acceleration and that visual acceleration influences perception of body pitch. Furthermore, discrimination performance improves when consistent visual and vestibular stimuli are presented. Prior distributions are proposed to account for misperceptions of body orientation: the A-effect and somatogravic illusion.

Chapter 2 examines the relative influence of visual scene orientation and rotational optic flow for perception of body roll. Observers were rolled in a rotating chair and set a visible rod to earth-vertical in the dark and with four visual scenes: rotating dots, stationary cube, stationary room, and rotating room. Rotational optic flow influenced perception of body roll, but scene orientation had more influence. In addition, real-world scenes had more influence than abstract, oriented scenes, like the cube.

Chapter 3 investigates the role of vestibular signals for interpreting ambiguous optic flow. Single-cue thresholds were measured for discriminating linear and angular velocity based on optic flow only and vestibular stimulation only. Visual-vestibular thresholds were measured for discrimination of self-motion and object motion. Maximum-likelihood combination of visual and vestibular signals should yield a reduction in self-motion thresholds. Comparison of these signals should yield object motion thresholds equal to the sum of single-cue thresholds. These predictions are upheld for some observers, but not others.