Witkin and his associates 
discovered that performance in 
this task was correlated with 
performance in other 
perceptual tasks. For example, 
those who were strongly 
influenced by the surrounding 
frame, the so-called field-
dependent observers, also 
found it difficult to isolate or 
find familiar figures embedded 
in a larger pattern. The so-
called field-independent 
observers, who resisted the 
effect of the frame on the 
perception of the rod, 
performed better in the 
embedded-figure task. The 
researchers also claimed that 
the two categories of observers 
tended to have different kinds 
of personality.

    If experimenters use a 
rectangle sufficiently large to 
serve as a surrogate frame of 
reference for the vertical-
horizontal coordinates of the 
environment, and they place it 
at, say, a tilt of around 30 
degrees, observers, on average, 
will set the rod at roughly 6 or 7 
degrees in the direction of the 
tilted frame. This result implies 
that a truly vertical rod seen in 
the frame would appear tilted 
by 6 or 7 degrees—–a sizable 
illusion. Quite a few observers 
do set the rod at this 
orientation; thus this effect is 
not simply based on averaging 
the two extreme types of 
subject. Such a result has been 
interpreted as a compromise 
between the two conflicting 
determinants: gravity and the 
visual frame of reference.

    How is the frame itself 
perceived? How it is seen 
presumably would influence 
the positioning of the rod. Were 
it to look upright, as does a 
tilted room in which the 
observer is enclosed, we should 
expect that the rod would be set 
parallel to its edges. Were it to 
look tilted, as tilted as it really 
is, should we expect any effect 
on the perception of the rod at 
all? If the frame is veridically 
perceived as tilted by 30 
degrees, presumably 
information about gravity is not 
being overpowered or captured 
by the surrogate frame of 
reference. In that event, the 
same information ought to be 
available to the observer to set 
the rod to the true vertical.