VisualJitter applet v0.50b
written by Ikuya Murakami Jul/16/98 using
CodeWarrior Pro 1 for Macintosh (Metrowerks) that supports
Java Developers Kit 1.0.2 (Sun Microsystems).
This software might show peculiar behavior depending on your local environment.
I and my friends have checked that it runs all right on
Netscape Navigator 4.04, 4.07 (for Macintosh),
Netscape Communicator 4.5 (for Macintosh and Microsoft Windows 95),
Microsoft Internet Explorer 4.01 (for Macintosh),
Metrowerks Java 1.0.7 (for Macintosh), and
Apple Applet Runner 2.0 (for Macintosh).
Save to your local disk (StuffIt + BinHex 52k)
A novel aftereffect: visual jitter in an unadapted static region after adaptation to dynamic random noise
((I. Murakami & P. Cavanagh)) ARVO '98.
Purpose. We report a new phenomenon: when dynamic random noise is replaced by
a static one after a period of adaptation, static random noise in an adjacent unadapted
region appears to "jitter" coherently in random directions for some time. Expanding on
this phenomenon, our study suggests a process for realizing a stable visual world despite
small eye movements. Methods. A fixation spot was provided throughout a trial.
During adaptation, dynamic random noise was typically presented in a circular patch, 6.7°
in diam., surrounded by a 13.3° outer-diam. annulus of static random noise.
Static noise was presented in both regions in the test period, during which the duration
of visual jitter was measured. For a stabilized image control, a high-intensity flash was
used to induce a long-lasting afterimage of the random-noise test field.
Results. (1) The effect usually lasted 5-10 s. (2) The two regions did not need to be
concentric; they could be any shapes if they were relatively close to each other. (3) During
adaptation, the surround could be filled with a blank field instead of static noise. (4) There
was no interocular transfer. (5) There was storage during a blank period (> 20 s) between
adaptation and test. (6) The jitter was synchronized in multiple test patches presented at
remote locations. (7) Retinal stabilization during test totally abolished the aftereffect.
Conclusions. Synchronized jitter at multiple locations and the loss of the effect in the
stabilized image indicate that the jitter aftereffect reflects retinal slip due to eye-position
fluctuation during test. As a model for compensating such retinal slip, we propose that motions
in the world are estimated as deviations from some baseline minimum in the retinal velocity
field. According to this model, adaptation to dynamic random noise transiently attenuates
responses to retinal velocities at adapted sites, creating a new minimum, and other retinal
velocities in the surround are computed as deviations from this new "baseline".
In unadapted regions, unattenuated retinal slip due to eye movements is now interpreted as a
jittering motion of those regions.
Notes
(1) If the cycling of the dot patterns is irregular or slow, try first quitting
all other applications. If the speed remains irregular, the jitter
aftereffect you will experience will be quite weak.
(2) The animation refresh rate is expected to be about equal to the scanning frequency of
the monitor you use (e.g., 67Hz),
but I found that the actual animation rate may be much lower for slower computers.
In that case, try to decrease "Inter-frame Interval" -- it might slow the GUI update of
the browser itself though.
(3) This demonstration simply stops animation during the test. In actual experiments,
however, a totally different static noise pattern is redrawn.
back to
Ikuya Murakami's page
back to
Vision Sciences Home Page