Here are two facts you need to know about retinal ganglion cells to
make sense of Meister's experiment:
1) They have a receptive field which extends over a spatial region.
2) Most of them respond to stimulus with a large transient signal
followed by a smaller sustained signal.
The first thing Meister et. al. did in their experiment was record the
activity of retinal ganglion cells while flashing stationary bars of
light. This allowed them to map out the spatial extent of each
ganglion cell's receptive field. It makes sense to say that the CENTER
of each cell's receptive field corresponds to the position in space
which the brain maps onto that cell's activity.
Now what happens if you move a bar of light across the ganglion cell's
receptive field? IF the ganglion cell was not "transient" -- that is,
if it always gave a constant, sustained signal proportional to the
amount of light falling onto its receptive field -- then the signal
from the ganglion cell would start to rise when the bar of light
entered its receptive field, reach a maximum when the bar of light was
over the center of its receptive field, and drop as the bar moved out
of its receptive field.
But because ganglion cells have a strong initial transient at stimulus
onset in addition to their smaller sustained signal, something
different happens. When the moving bar of light first enters the
cell's receptive field, the cell starts to give its big, transient
signal. While the bar is still moving through the cell's receptive
field, the signal from the ganglion cell starts to drop as it sends its
smaller, "sustained" signal. And of course the signal from the
ganglion cell drops to zero as the bar of light moves out of the
receptive field. The net result is -- depending upon the speed with
which the bar of light is moving -- the signal produced by the ganglion
cell reaches its peak intensity BEFORE the bar of light has reached the
center of the cell's receptive field. In that sense, you could say
that the ganglion cell was "predicting" the motion of the bar. NOT
because it was firing a signal before the bar reached its receptive
field; rather, because the signal from a moving bar reaches maximum
before the bar reaches the CENTER of its receptive field.
We spent some time over lunch discussing whether or not Meister got a
big publication out of saying something which everyone has known for
decades -- namely, that most ganglion cells have a strong transient
signal. We decided that Meister really did show something a bit more
than that. (His results showed that this "predictive coding" effect --
which was a well-known psychophysical phenomenon for quite a few years
and was widely assumed to happen entirely in the visual cortex --
actually happens (to at least some extent) in the retina before signals
even reach the visual cortex. His results also show that at least some
of the ganglion cell's "fast adaptation" (switching from transient to
sustained signal levels) happens over the ganglion cell's entire
dendritic field -- rather than happening entirely at earlier stages of
processing in the photoreceptors or bipolar cells.)
Now for a few other points.
The BBC Sci-Tech article also said:
> The finding revolutionises many previous models of the eye, which
> assumed that it acted simply as a camera - capturing the image presented
> directly in front of it.
They need better editing and less hype over at BBC Sci-Tech.
Nobody who is actually doing research in retinas these days thinks of
the eye as a "simple camera."
I completely agree with one point Steve Jones was trying to make: the
retina is complex. There's a whole lot of signal-processing that gets
done in the retina before the signals ever reach the brain. We already
understand quite a bit of that complex wiring, and there's still quite
a bit that we don't have figured out. The retina is amazing!
I also agree that the "bad design" claim -- that the vertebrate retina
has a huge design flaw -- is scientifically premature.
The vertebrate retina is arranged so that light entering the eye has to
pass through several layers of ganglion cells, amacrine cells, bipolar
cells, horizontal cells, and Mueller cells (and in some cases, blood
vessels) before hitting the photoreceptors. This causes some optical
blurring. Also, when all the axons from the ganglion cells bundle
together to form the optic nerve, this forms a "blind spot" in the
retina where there can't be any photoreceptors. The invertebrate
retina is arranged in the opposite order. Light hits the
photoreceptors first. No blurring; no blind spot.
Is the invertebrate eye designed better? Actually, that's not clear.
The "backwards" arrangements of the vertebrate retina allows the
photoreceptors to be in contact with the pigment epithelium. This
tissue not only blocks further transmission of light into the head, it
also helps recycle the photoreceptors' used photopigments. Recycling
used photopigments is a metabolically intensive process. So there's an
advantage to having the photoreceptors right next pigment epithelium.
This arrangement allows tight spatial packing of photoreceptors, and
allows rapid recycling of photopigments.
How do invertebrates deal with the fact that their photoreceptors are
more distant from the pigment epithelium? I don't know. We could
speculate that the invertebrate eye has to sacrifice either some
photopigment recycling speed or some tight spatial packing, or a bit of
both. Maybe someone who studies invertebrate eyes knows the answer,
but I don't.
The point here is that vertebrate retina construction is not
necessarily an inferior design to the invertebrate construction. That
claim is scientifically premature.
But of course, the real problem with the claim that "bad design of the
vertebrate eye 'proves' that they weren't created" is a theological
problem, not scientific.
Things don't have to arise _de_novo_ to be "created;" things like
retinas can also be created through process. In addition, things like
retinas can even be created with 'sub-optimal design' (given our
definitions of "optimal") and still be declared "good" by a good
Having said that, I do think that comparing retina organization across
different species makes a pretty good case for common ancestry. You
see patterns of similarity in retinal organization (and almost
certainly in patterns of similarity in developmental programs and
genetic sequences) which are nested at the levels of genus, family,
order, and class. The ancestry tree inferred from these nested
similarities of retinal organization matches the ancestry tree inferred
from other biological features and from the fossil record.
Now, that doesn't prove that the eye evolved without miraculous
intervention. Common ancestry is consistent both with evolutionary
creationism and with some versions of episodic creationism which would
argue that the eye is just too complex to have evolved without
miraculous intervention. Well, IS the eye too complex for evolution?
Answer: it's way too soon to give a solid, empirical answer. Popular
literature on evolution gives hand-wavy arguments about how the complex
eye could have evolved from simpler systems. Some proponents of
episodic creation give hand-wavy arguments about how the eye is too
complex to have evolved from simpler systems. Whose hand-wavy
arguments do you want to believe? Neither side has the genetic data
across multiple species *necessary* to make a strong, empirical case.
I'd say the evolutionary biologists do have some corroborating evidence
on their side in that we do see many different levels of eye complexity
in nature if we look at different species. But that's not going to
settle the case. We need more data to settle it. I have other reasons
for favoring evolutionary creation, but none related directly to the
eye, so this seems like a good place to stop.