The presence of an additional external upright frame was studied in th
ree experiments to separate the role of visuovestibular, global, and l
ocal mechanisms in the rod-and-frame illusion (RFI). In the first expe
riment, carried out in a dark room, the external frame surrounded a la
rge tilted frame. Rod-setting errors to the vertical were abolished wi
th the additional-frame condition (at 22 degrees inner-frame tilt) con
firming earlier findings. However, small, residual direct (at 11 degre
es inner-frame tilt) and indirect effects (at 33 degrees inner-frame t
ilt) were still present, indicating the persistence of global visual p
rocessing. In the second experiment, the RFI in the dark was compared
with the RFI with the lights on. Turning the light on abolished the ef
fect at 22 degrees and 33 degrees frame tilt; however, a small direct
effect was maintained at 11 degrees frame tilt. These two studies indi
cate that the addition of veridical vertical information abolishes the
effect owing to visuovestibular mechanisms. In the third experiment,
a small rod and frame was used with the lights on (a condition abolish
ing visual-vestibular interaction). In the case of a small gap between
the rod and the inducing frame (a condition which maximises local pro
cessing), the effect of the outer upright frame was negligible; this i
ndicates that the additional frame had no effect on local processing.
In the case of a large gap (a condition which minimises local processi
ng), the external square reduced the illusion, indicating its modulati
ng effect on visual global processing. Overall, an upright external fr
ame exerts a differential influence depending on which mechanisms cont
ribute most to the RFI in a given experimental condition.