HOW THE BRAIN MAKES POTENTIAL 175With a little refl ection this is apparent in other ways. For example, when
we move a book around on a table, something invariant about its “book-
ness” persists in spite of radical changes in appearance. Likewise, the end-
on appearance of a car is quite diff er ent from the view from the side. Yet
we somehow retain a coherent image of a single object and not diverse
snapshots—so long as it can be integrated with our own movements.
So, what a sensory system particularly needs is the structure that
comes from motion, that is, the deeper statistical information it provides.
Indeed, the importance of visual structure has been suspected— and
neglected— for de cades. In 1953, Hans Wallach showed that if a station-
ary three- dimensional fi gure (e.g., a wire- form cube) is illuminated from
behind so that its shadow falls on a translucent screen, an observer in
front of the screen will see a two- dimensional pattern of lines. But if the
same object is rotated, the observer will immediately identify it as a
three- dimensional cube, even though only two- dimensional informa-
tion is presented. Wallach called it the kinetic depth eff ect. (Many dem-
onstrations are available on the internet.)
Another popu lar demonstration indicates how whole fi gures can be
recognized from a few light points moving with appropriate coordination
but not in static pre sen ta tion. One of the best illustrations is the point-
light walker (fi gure 6.1). Th ere are many other examples. So crucial is the
structure in motion that most animals fail to respond to a perfectly mo-
tionless object, as if they cannot see it. If a dead fl y on a string is dangled
perfectly still in front of a starving frog, the frog literally cannot sense this
meal to save its life. In humans it has been shown that perception of a ret-
i nal image held perfectly still on the eye (by pasting the image to a kind
of contact lens) quickly fades and dis appears. Again, it seems to be the
structure, not the particulars, that is crucial to the evolved visual system.
So what kind of structure can be so informative? A widely accepted idea
is that the structure exists in the form of statistical associations between
the variables in motion. For example, the way that light spots, or other as-
pects of a viewed object, move together in space and time, or appear and
dis appear together. As Horace Barlow put it, “Th e visual messages aroused
by the world around us are full of very complicated associative structure.”
He demonstrated that single cells in ret ina and brain are sensitive to them,
and he suggests that the brain may use “higher- order” associations.^7
This content downloaded from 139.184.14.159 on Tue, 17 Oct 2017 13:54:00 UTC