readings> saccades and consciousness
Set the kids loose with the camcorder and the result is always
the same: a mad dancing picture as if the scenes had been shot from a
particularly hairy fair ground ride. Why can’t they let the
camera linger?
As any neurologist knows, the lurching lens is just imitating the
child’s saccading eye. Kids naturally expect to be able to
shift their attention in eyeball-swivelling leaps three or four times a
second and still see a perfectly stable world. So no mystery. Yet in
fact the business of saccades remains an especially troublesome one for
consciousness theorists.
In the 1990s, the big issue was the binding problem. If the cortex is a
hierarchy of processing modules, then how do the various computations
get glued together to make a seamless conscious whole? This question
has been reasonably well answered with the discovery that coherent
firing rhythms help knit widely scattered neurons into fleeting global
ensembles. Now mind scientists are asking the same kind of question but
concerning the unity of consciousness over seconds rather than
milliseconds.
The naive view of the brain is that it merely reflects
what the sense organs report. Yet if the eyes are skipping from
fixation to fixation, then the resulting cortex mappings must leap
about like the pictures from the kids’ camcorder. So how can
a steady stream of experience be constructed from such violently
wrenching foundations?
The need for saccades is clear enough. The cone-dense fovea covers only
1 degree of the visual field. So it’s reckoned that to have
foveal level acuity over the whole retina, the brain would need to be
hundreds of thousands times bigger and weigh ten tons!
Saccading itself
is a complex affair. Even when the eyes are fixated, there are constant
tremors and microsaccades that keep the retina refreshed. Saccades then
come at two rates depending on how habitual or routine the situation
happens to be. With highly predictable events, our eyes can flick to
the spot in as little as a tenth of a second in an express saccade. For
more exploratory looking, it takes a fifth of a second to find a new
fixation point.
Of course, during a saccade, retinal output is suppressed to cut down
the blur of motion - though recent experiments suggest it is may be
rather that the brain immediately forgets any visual input captured
while the eye is in transit. But either way, once we count in the
20,000 eyeblinks we make each day, we must be effectively blind about a
third of the time!
Actual fixations last around a third of a second. Another recent
surprise is that the time the eye lingers is mostly to do with the
processing needs of the brain. A visual image can be
“snatched” in a tenth of second. But it seems the
brain then holds the eye in place largely to give itself grace to
complete its processing. So fixation prevents fresh input before the
brain is ready.
Change blindness experiments, in which background visual features are
sneakily altered while a subject’s eyes are in mid-saccade,
raise yet further questions. It used to be thought that vision was
stabilised by a simple anticipatory routine. Each planned eye movement
generated an efference copy – an exact forward motor
projection – that recalibrated the visual cortex, cancelling
every apparent jump in the visual scene.
But change blindness studies
demonstrate that the brain can just gloss over some quite abrupt
changes in sensory input. It is only if the background features have
been highlighted by prior acts of attention that any changes become
detectable – apparently a required part of the forward model.
So how does the brain create a coherent stream of experience out of
such restless visual machinery? To cut a long story short, the
stability of our mental representations seem the result of a clever mix
of rapid forgetting and unconscious prediction. Raw sensory information
seems to flow through to the cortex more freely than some older
theories suggest. But it is just as freely ignored if it does not fit
into a general, expectancy-led, running model of the world.
The
lurching shifts in visual input are not gated or cancelled out but
instead downplayed, rendered vague, as the brain gets on with tracking
its own gist-based view of what is happening. So in a neat turnaround,
as with the child holding the camcorder, it is the inner narrative
which comes first.
The conscious stream – the so-called output – is
the foundation for the mental experience. And the wrenching changes in
visual input then only get cortically represented to the degree to
which they play into the inner tale being told.