Tuesday, November 10, 2015

The Pith and the Pendulum: How do slow visual signals keep up with fast ones from a moving object?


When we watch a moving object, we experience at least two visual signals: A fast one due to luminance contrasts (which also brings with it a high spatial resolution) and a slow one due to chromatic contrasts (which has much lower spatial resolution).  It is interesting---and functionally necessary---that we see an object in motion as recognizable and locatable, without any chromatic blur due the slow visual mechanism's failure to catch up with the fast one.

Chromatic blur is just one aspect of the visual signal that is suppressed when we look at a moving object. Also suppressed is our awareness of the persistent motion of the eye itself; despite this motion, we see the motion of, say a thrown baseball, as a simple arc.

Despite the general suppression of the chromatic blur, there is a simple demonstration that shows the chromatic signal sluggishly and fuzzily lagging the luminance signal. (I did this experiment at MIT about 40 years ago.) Set up a pendulum in a generally well illuminated room, in front of a white wall.
Now illuminate the scene with a diffuse blue light that nonetheless has a localized source. An example of such light is a slide projector without its lens, illuminating through a Kodak Wratten 98 filter.  The pendulum should be between the projector and the wall. The viewer should sit in such a way as to see simultaneously the pendulum and its shadow on the wall. If the room is brightly enough lit and the shadow is well defined, the blue light will be almost invisible, but the shadow will appear a rather intense yellow. Because the shadow (illuminated only by the white light) is almost the same luminance as the background (illuminated by white and blue lights together), the shadow will appear quite blurry. To see that it is the visual system that is creating the blur, replace the blue filter by a filter of another color and see the shadow appear as a relatively sharp image. The shadow edge is indeed exciting chromatic contrast with very low luminance contrast. (Note: If we instead had contrived the contrast edge to be truly isoluminous and restricted to excitation differences of the blue-sensitive cones, the edge would be completely invisible [1]. To quote Boynton [1], “The blue-sensitive cones […] seem to be free of any serious spatial or temporal responsibilities in vision.”)

Given our stationary shadow-casting pendulum, now set the pendulum in motion. You will notice immediately that the shadow lags the motion of the pendulum. At first it will seem to be an independent object, but then a weird thing happens. With each cycle, the shadow of the pendulum somehow reasserts its phase relative to the pendulum itself. If this were an independent object, the phase difference would continue to pile on, for the two pendula would have different frequencies. Eventually the pendula would be in counterphase, and then beyond counterphase. But this doesn't happen. Instead, each time the real pendulum reaches its turning point, the shadow reaches a turning point a fixed time thereafter. If you try to stare down the shadow pendulum to capture what should be a progressive phase lag, you will notice the shadow getting blurrier as your eye's involuntary motion takes over.

At this point, present and future vision scientists should take over. It should be possible to extract some scientific pith from this pendulum.

[1] R. M. Boynton, Ten years of research with the minimally distinct border. In J. C. Armington, J. Krauskopf, and B. R. Wooten (eds.), Visual Psychophysics and Physiology. Academic, New York (1978).
[image copyrighted by the Exploratorium, www.exploratorium.edu.] Among these colored shadows, the yellow hand is the one that waves to you most slowly.


Michael H. Brill
Datacolor