
The visual cortex appears to provide neural selectivity necessary
for perceiving objects moving in three-dimensional space. It also
appears to compensate the "contrast" of the internal
image over the wide ranges of contrast we encounter in natural
surroundings. For example, the contrast range is low for a ski-slope
on a cloudy day, but high after the sun appears and creates sharp
shadows. I have been making single-neuron measurements in the
cat's visual cortex that demonstrate smooth directional responses
to a moving textured object, just as we perceive such movement,
even though earlier neurons are strongly modulated by image-elements
in the texture. Our approach to understanding motion mechanisms
is to test a neuron with a stimulus that includes randomized brief
samples of motion to tease apart the effects of space and time.
The result is an unbiased measurement of a neuron's preferred
velocity (the slope of the obliquely oriented "motion kernel",
plotted below in space vs time for a directionally selective cortical
neuron). Differences in preferred velocity between the two eyes
would suggest selectivity for motion in depth. These measured
elemental motion responses are then reassembled to establish a
wide variety of neural responses that are compared with responses
of proposed mechanistic models. A successful model demonstrates
the same set of properties as the receptive field of the measured
neuron. Such models provide great insight into neural mechanisms,
including the nature and order of processing and connections among
earlier neurons. Similar measurements, after changing the contrast
of the randomized stimulus, demonstrate the changes in gain and
temporal properties that allow the visual system to operate efficiently
over the widely varying ranges of contrast that occur in our visual
environment.
