Abstract
The visual horizontal as judged by 4 normal subjects (authors and 2 experienced test pilots) was recorded every 2 seconds during periods lasting up to 23 minutes. Each subject was tested in an upright, then in a recumbent (left side) position. In both positions the procedure was identical: Empirical visual cues serving as a background to the luminous line target were alternately illuminated for 2 minutes, then darkened completely for 5 minutes during each period. It was found that a lack of visual cues did not appreciably influence the accurate and relatively stable localization of the horizontal in the upright position; however, in the recumbent position removal of these cues caused, after a brief lag period, a gradual spontaneous rotation of the phenomenal horizontal up to a maximum displacement typical for each subject. Superimposed upon these perceptual changes was the considerable fluctuant movement in horizontality which was described as a form of autokinesis (rotary). These illusions disappeared almost instantaneously when a visual framework was provided. Qualitatively, the time course of these perceptions was similar and highly reliable for all subjects. The pilot subjects, however, differed from the nonpilots by indicating a slightly greater time lag at onset, a slower rate in reaching the maximum deviation position, significantly less total deviation (about one-third), and less rotary autokinesis. In a subsequent (recumbent) trial, 2 subjects, the authors, observed the target for 30 minutes in the dark. The error perceived remained essentially at the same level for one subject, but decreased significantly for the other.