At slow rates of change of 
egocentric direction—–as would 
often be true of slowly drifting 
clouds—–motion detection is 
poor. But change of relative 
location is more readily 
detected. The following 
experiment makes this point 
clearly. In a dark room, a single 
luminous spot can be set in 
motion at a speed below our 
threshold to detect its 
movement. If a second 
luminous stationary spot is 
introduced nearby, however, 
we immediately do see a spot in 
motion. Apparently, we are very 
sensitive to the changing 
distance between the two spots. 
Although we will tend to see 
one of the spots moving, we are 
equally often wrong as right as 
to which spot it is. In this 
experiment, the only usable 
motion information we are 
receiving is of a relative kind. 
Because such information is 
ambiguous, however, we 
cannot tell which object’s 
motion is producing the 
relative change.

    In the case of the moon and 
cloud, then, it is reasonable to 
suppose that the relative 
change of position between the 
two is paramount in our 
perception but that it is also 
ambiguous. Therefore, half of 
the time we should erroneously 
attribute the change to the 
moon’s motion. However, the 
moon will almost always appear 
to move when a cloud moves in 
front of it, not merely half the 
time. There is a further 
principle of induced motion 
that is applicable in this case. 
An object that surrounds 
another, or is much larger than 
it is, tends to be seen as 
stationary. The larger object 
therefore serves as a frame of 
reference with respect to 
which the relative 
displacement of other things is 
gauged. To prove this point, Karl 
Duncker, a Gestalt psychologist 
who pioneered investigation of 
induced movement in the late 
1920s, varied the experiment 
just described by replacing the 
moving point by a moving 
luminous rectangle that 
surrounded the stationary 
luminous spot. The stationary 
spot appeared to move on every 
trial.