Viewing tilted images that contain spatial information about gravity and verticality, such as rotated landscapes, or leaning buildings, produces an eye movement response known as optostatic torsion (OST). OST consists of a small-amplitude rotation ∼1º of the eyes around the line of sight, in the same direction of image tilt. Here we aimed to determine whether illusory perceptions of visual tilt were sufficient to produce OST. The illusory stimulus was a variation of the Café Wall illusion, applied to four "walls" to approximate the appearance of a tilted room. In a first experiment we determined the perceived magnitude of tilt of the illusory stimulus using a two-alternative forced choice (2AFC) task, measuring the amount of tilt required to cancel the perceived illusory tilt for clockwise (CW) or counterclockwise (CCW) configurations. We found an illusory tilt of +3.67±0.57° and -3.80±0.93°, respectively. Then, in a second experiment, we recorded 3 dimensions of binocular eye movements in ten healthy subjects viewing one of four possible stimuli: 1) illusory, 2) non-illusory with small tilt, 3) landscape with small tilt, and 4) landscape with large tilt. Both the landscape (±4º tilt: 0.4±0.1, p<0.05; ±30° tilt: 0.5±0.1, p<0.01;) and control stimulus (0.2°±0.1°, p<0.05) produced a significant amount of OST when comparing left tilt and right tilt configurations while the illusory stimulus (0.11°±0.07°, p=0.15) did not. This indicates a potential dissociation between our perception of tilt and the processing of tilt that drives the motor response of OST.
Girum et al. (Mon,) studied this question.