Why White Noise?
What is a white noise visual stimulus?
First, don't be mislead by the word "noise". A white noise stimulus
is a signal, not noise. It is simply a specific kind
of time varying, random signal.
A temporal white noise stimulus would be, for example, changing the
intensity of a spot every 50 msec to a randomly selected new intensity
value. Such a random sequence has the property of containing all
temporal frequencies equally (for temporal frequencies slower than the
frame update frequency), which is all that is meant by calling it
"white noise".
Aside: A sometimes useful variant of this stimulus is to choose
the random intensities at very short time intervals, but then filter
out higher temporal frequencies, resulting in "filtered noise", which
appears as a more smooth or continuous fluctuation rather than an
abrupt flicker.
For spatio-temporal white noise, this process would be repeated at
every point in space (with some spatial resolution corresponding to
the size of the stimulus pixels).
Why use white noise?
One motivation behind using random dynamic stimulus is that
it mimics an aspect of real vision: the necessity to form the best
estimate of a unique and constantly changing or moving scene in real
time, based on only one "look" at it. This approach to visual
processing is in contrast to the more traditional one in which a
static stimulus (bar, grating, etc.) is presented for a long time
(hundreds of msec or even seconds), often many times over to come up
with an average neural response. The traditional approach emphasizes
visual processing more in the context of saccades and fixations around
a frozen visual scene.
White noise random stimuli in particular produce a wide spectrum of
stimuli, as if the visual system were designed to encode any
possible scene that could impinge on the retina. (This will be the
starting point of testing that hypothesis). If we randomize each
position in space as a function of time, we will in principle show
every possible image eventually. However, in practice, since we will only show
a tiny subset of these, any given image is almost infinitely unlikely.
copyright 1995 Pam Reinagel
Back to Comments (1995-1997)
Back to Pam Reinagel's home page
Back to Reinagel Lab