Jonathan Sarfati, a particularly silly creationist, is quite thrilled — he's crowing about how he has caught Richard Dawkins in a fundamental error. The eye did not evolve, says Sarfati, because it is perfectly designed for its function, and Dawkins' suggestion that there might be something imperfect about it is wrong, wrong, wrong. He quotes Dawkins on the eye.
But I haven't mentioned the most glaring example of imperfection in the optics. The retina is back to front.
Imagine a latter-day Helmholtz presented by an engineer with a digital camera, with its screen of tiny photocells, set up to capture images projected directly on to the surface of the screen. That makes good sense, and obviously each photocell has a wire connecting it to a computing device of some kind where images are collated. Makes sense again. Helmholtz wouldn't send it back.
But now, suppose I tell you that the eye's 'photocells' are pointing backwards, away from the scene being looked at. The 'wires' connecting the photocells to the brain run over all the surface of the retina, so the light rays have to pass through a carpet of massed wires before they hit the photocells. That doesn't make sense…
What Dawkins wrote is quite correct, and nowhere in his refutation does Sarfati show that he is wrong. Instead, Sarfati bumbles about to argue against an argument that no biologist makes, that the eye is a poor instrument for detecting patterns of light. The argument is never that eyes do their job poorly; it's that they do their job well, by a peculiar pattern of kludgy patches to increase functionality that bear all the hallmarks of a long accumulation of refinements. Sarfati is actually supporting the evolutionary story by summarizing a long collection of compromises and odd fixes to improve the functionality of the eye.
There's a fundamental question here: why does the vertebrate eye have its receptors facing backwards in the first place? It is not the best arrangement optically; Sarfati is stretching the facts to claim that God designed it that way because it was superior. It ain't. The reason lies in the way our eye is formed, as an outpocketing of the cortex of the brain. It retains the layered structure of the cortex, even; it's the way it is because of how it was assembled, not because its origins are rooted in optical optimality. You might argue that it's based on a developmental optimum, that this was the easiest, simplest way to turn a light-sensitive patch into a cup-shaped retina.
Evolution has subsequently shaped this patch of tissue for better acuity and sensitivity in certain lineages. That, as I said, is a product of compromises, not pre-planned design. Sarfati brings up a series of odd tweaks that make my case for me.
The vertebrate photoreceptors are nourished and protected by an opaque layer called the retinal pigmented epithelium (RPE). Obviously, you couldn't put the RPE in front of the visual receptors, so the retina had to be reversed to allow it to work. This is a beautiful example of compromise: physiology is improved at the expense of optical clarity. This is exactly what the biologists have been saying! Vertebrates have made a trade-off of better nutrient supplies to the retina for a slight loss of optical clarity.
Sarfati makes the completely nonsensical claim that the presence of blood vessels, cells, etc. in the light path do not compromise vision at all because resolution is limited by diffraction at the pupil, so "improvements of the retina would make no difference to the eye's performance". This is clearly not true. The fovea of the vertebrate eye represents an optimization of a small spot on the retina for better optical properties vs. poorer circulation: blood vessels are excluded from the fovea, which also has a greater density of photoreceptors. Obviously, improvements to the retina do make a difference.
It's also not a condition that is universal in all vertebrates. Most birds, for instance, do not have a vascularized retina — there is no snaky pattern of blood vessels wending their way across the photoreceptors. Birds do have greater acuity than we do, as well. What birds have instead is a strange structure inside their eye called the pecten oculi, which looks kind of like an old steam radiator dangling into the vitreous humor, which seems to be a metabolic specialization to secrete oxygen and nutrients into the vitreous to supply by diffusion the retina.
Sarfati also plays rhetorical games. This is a subtly dishonest argument:
In fact, cephalopods don't see as well as humans, e.g. no colour vision, and the octopus eye structure is totally different and much simpler. It's more like 'a compound eye with a single lens'.
First, there's a stereotype he's playing to: he's trying to set up a hierarchy of superior vision, and he wants our god-designed eyes at the top, so he tells us that most cephalopods have poorer vision than we do. He doesn't bother to mention that humans don't have particularly good vision ourselves; birds have better eyes. So, is God avian?
That business about the cephalopod being like a compound eye is BS; if it's got a single lens, it isn't a compound eye, now is it? It's also again pandering to a bias that our eyes must be better than mere compound eyes, since bugs and other lowly vermin have those. Cephalopods have rhabdomeric eyes, meaning that their photoreceptors have a particular structure and use a particular set of biomolecules in signal transduction, but that does not in any way imply that they are inferior. In fact, they have some superior properties: the cephalopod retina is tightly organized and patterned, with individual rhabdomeres ganged together into units called rhabdomes that work together to process light. Their ordered structure means that cephalopods can detect the polarity of light, something we can't do at all. This is a different kind of complexity, not a lesser one. They can't see color, which is true, but we can't sense the plane of polarity of light in our environment.
I must also note that the functions of acting as a light guide (more below) and using pigment to shield photoreceptors are also present in the cephalopod eye…only by shifting pigments in supporting cells that surround the rhabdome, rather than in a solid RPE. Same functions, different solutions, the cephalopod has merely stumbled across a solution that does not simultaneously impede the passage of light.
Color vision, by the way, is a red herring here. Color is another compromise that has nothing to do with the optical properties of the arrangement of the retina, but is instead a consequence of biochemical properties of the photoreceptors and deeper processing in the brain. If anything, color vision reduces resolution (because individual photoreceptors are tuned to different wavelengths) and always reduces sensitivity (you don't use color receptors at night, you may have noticed, relying instead on rods that are far less specific about wavelength). But if he insists, many teleosts have a greater diversity of photopigments and can see colors we can't even imagine…so humans are once again also-rans in the color vision department.
Sarfati is much taken with the discovery that some of the glial cells of the eye, the Müller cells, act as light guides to help pipe light through the tangle of retinal processing cells direct to the photoreceptors. This is a wonderful innovation, and it is entirely true that in principle this could improve the sensitivity of the photoreceptors. But again, this would not perturb any biologist at all — this is what we expect from evolution, the addition of new features to overcome shortcomings of original organization. If we had a camera that clumsily had the non-optical parts interposed between the lens and the light sensor, we might be impressed with the blind, clumsy intricacy of a solution that involved using an array of fiber optics to shunt light around the opaque junk, but it wouldn't suggest that the original design was particularly good. It would indicate short-term, problem-by-problem debugging rather than clean long-term planning.
Sarfati cannot comprehend why the blind spot would be a sign of poor design, either. He repeats himself: why, it's because the eye needs a blood supply. Yes, it does, and the solution implemented in our eyes is one that compromises resolution. I will again point out that the cephalopod retina also needs a blood supply, and they have a much more elegant solution; the avian eye also needs a blood supply, but is not invested with blood vessels. He gets very circular here. The argument is not that the vertebrate eye lacks a solution to this problem, but that there are many different ways to solve the problem of organizing the retina with its multiple demands, and that the vertebrate eye was clearly not made by assembling the very best solutions.
Sarfati really needs to crawl out of his little sealed box of creationist dogma and discover what scientists actually say about these matters, and not impose his bizarre creationist interpretations on the words of people like Dawkins and Miller. What any comparative biologist can see by looking at eyes across multiple taxa is that they all work well enough for their particular functions, but they all also have clear signs of assembly by a historical process, like evolution and quite unlike creation, and that there is also evidence of shortcomings that have acquired workarounds, some of which are wonderfully and surprisingly useful. What we don't see are signs that the best solutions from each clade have been extracted and placed together in one creature at the pinnacle of creation. And in particular — and this has to be particularly grating to the Genesis-worshipping creationists of Sarfati's ilk, since he studiously avoids the issue — Homo sapiens is not standing alone at that pinnacle of visual excellence. We're kinda straggling partway down the peak, trying to compensate for some relics of our ancestry, like the fact that we're descended from nocturnal mammals that let the refinement of their vision slide for a hundred million years or thereabouts, while the birds kept on optimizing for daylight acuity and sensitivity.