#6 of Our Top Ten Evolution Stories of 2014: Phys.org Says the Argument for Suboptimal Design of the Eye "Is Folly"
Casey Luskin
In 2010, the pro-ID website Access Research Network (ARN) posted its Top Ten Darwin and Design Science News Stories of the year, and its No. 1 story was a paper in Physical Review Letters, "Retinal Glial Cells Enhance Human Vision Acuity." Why was this article ARN's top story for the year? Because it found that special "Müller glia cells" sit over the retina, acting like fiber-optic cables to channel light through the optic nerve wires directly onto the photoreceptor cells.
This refuted the old objection to intelligent design that the vertebrate eye is "poorly designed" because the optic nerve extends over the retina instead of going out the back of the eye. These cells ensure that there is no loss of visual acuity due to the presence of the optic nerve, as the paper found, revealing the retina "as an optimal structure designed for improving the sharpness of images." As New Scientist put it at the time, these funnel-shaped cells "act as optical fibres, and rather than being just a workaround to make up for the eye's peculiarities, they help filter and focus light, making images clearer and keeping colours sharp." We also reported on this here.
Now a new paper in Nature Communications, "Müller cells separate between wavelengths to improve day vision with minimal effect upon night vision," has expanded upon this research, further showing the eye's optimal design. According to the paper, Müller cells not only act as optical fibers to direct incoming light through the optic nerve, but are fine-tuned to specific wavelengths to ensure that light reaches the proper retinal cells. From the Abstract:
Vision starts with the absorption of light by the retinal photoreceptors -- cones and rods. However, due to the 'inverted' structure of the retina, the incident light must propagate through reflecting and scattering cellular layers before reaching the photoreceptors. It has been recently suggested that Müller cells function as optical fibres in the retina, transferring light illuminating the retinal surface onto the cone photoreceptors. Here we show that Müller cells are wavelength-dependent wave-guides, concentrating the green-red part of the visible spectrum onto cones and allowing the blue-purple part to leak onto nearby rods. This phenomenon is observed in the isolated retina and explained by a computational model, for the guinea pig and the human parafoveal retina. Therefore, light propagation by Müller cells through the retina can be considered as an integral part of the first step in the visual process, increasing photon absorption by cones while minimally affecting rod-mediated vision. (Amichai M. Labin, Shadi K. Safuri, Erez N. Ribak, and Ido Perlman, "Müller cells separate between wavelengths to improve day vision with minimal effect upon night vision," Nature Communications, DOI: 10.1038/ncomms5319 (July 8, 2014).)The paper presents Müller cells as a direct answer to the view that the vertebrate eye has a suboptimal wiring:
[T]he mammalian retina and the peripheral retina of humans and primates are organized in a seemingly reverse order with respect to the light path. This arrangement places the photoreceptors, responsible for light absorption, as the last cells in the path of light, rather than the first. Therefore, the incident light must propagate through five reflecting and scattering layers of cell bodies and neural processes before reaching the photoreceptors. This 'inverted' retinal structure is expected to cause blurring of the image and reduction in the photon flux reaching the photoreceptors, thus reducing their sensitivity. It has been recently reported that retinal Müller cells act as light guides serving to transfer light across the retina, from the vitreo-retinal border towards the photoreceptors.How do Müller cells accomplish this feat? The article continues: "A single Müller cell collects light at the vitreo-retinal surface from an extended retinal region, and guides it onto one coupled cone, located at its distal end." The shape of the Müller cells -- wide at the top where it collects light, and narrow at the bottom where it delivers light to the rods and cones -- presents a potential optimization tradeoff between day vision (which depends more on efficient light transmission to the cones) and night vision (which depends more on efficient light transmission to the rods).
They then ask an engineering question: "Can this cost-benefit optimization problem between day vision and night vision be solved, without significantly impeding one or the other?" They find that the retina is optimized to solve this problem:
[H]uman Müller cells separate white light according to its wavelengths; medium- and long-wavelength light is concentrated onto cones and short-wavelength light leaks to illuminate nearby rods. Next, we show similar theoretical calculations for the guinea pig Müller cells and describe imaging experiments in the isolated guinea pig retina, to find remarkable agreement between the experimental results and the computational model. These findings are consistent with the hypothesis that the wave guiding properties of Müller cells are wavelength-dependent in a manner that improves cone-mediated vision while minimally impeding rod-mediated vision.The paper explains that Müller cells give the retina a specialized architecture to foster light collection: "We could clearly identify distinct light guiding tubes across most of the retinal depth, spanning the retina from the retinal surface down to just above the photoreceptors. The only retinal structures that fit these light-guiding tubes are the Müller cells." They conclude:
The findings presented here indicate that the spectral separation of light by Müller cells provides a mechanism to improve cone-mediated day vision, with minimal interference with rod-mediated night vision. This is achieved by wavelength sorting of incident light by the Müller cells. Light of relevant wavelengths for cone visual pigments is directed towards the cones, while light of wavelengths more suitable for rod vision is allowed to leak outside the Müller cells towards the surrounding rods. This is a novel mechanism that needs to be considered when visual phenomena concerning cone- and rod-mediated vision are analysed.The implications of these findings have not been lost on expert optics commentators. A striking article at Phys.org about this new paper, "Fiber optic light pipes in the retina do much more than simple image transfer," reflects a keen awareness of the debate over whether the vertebrate eye is suboptimally designed. It concludes that the retinal architecture, as it now stands revealed, settles the debate. In the words of Phys.org, the notion that the vertebrate eye is suboptimally wired "is folly." Why? Because "Having the photoreceptors at the back of the retina is not a design constraint, it is a design feature." Here's the full passage from the article:
Having the photoreceptors at the back of the retina is not a design constraint, it is a design feature. The idea that the vertebrate eye, like a traditional front-illuminated camera, might have been improved somehow if it had only been able to orient its wiring behind the photoreceptor layer, like a cephalopod, is folly. Indeed in simply engineered systems, like CMOS or CCD image sensors, a back-illuminated design manufactured by flipping the silicon wafer and thinning it so that light hits the photocathode without having to navigate the wiring layer can improve photon capture across a wide wavelength band. But real eyes are much more crafty than that. A case in point are the Müller glia cells that span the thickness of the retina. These high refractive index cells spread an absorptive canopy across the retinal surface and then shepherd photons through a low-scattering cytoplasm to separate receivers, much like coins through a change sorting machine. A new paper in Nature Communications describes how these wavelength-dependent wave-guides can shuttle green-red light to cones while passing the blue-purples to adjacent rods. The idea that these Müller cells act as living fiber optic cables has been floated previously. It has even been convincingly demonstrated using a dual beam laser trap. In THIS case (THIS, like in Java programming meaning the paper just brought up) the authors couched this feat as mere image transfer, with the goal just being to bring light in with minimal distortion. (Emphasis added.)Take special note of the sentences I've put in bold at the end of the last paragraph. These recent discoveries about the retina were made by proposing that Müller cells behave like "living fiber optic cables" that have a "goal" to "bring light in with minimal distortion." This is an example of systems-biology-thinking, as I described it here recently, where you assume that biological systems function much like goal-directed technology, and then reverse engineer a system to determine how it works. Such teleological thinking is once again bearing fruit in biology.
But the Phys.org article explains that these Müller cells aren't just passive cables that transfer images -- they are dynamic structures that can adjust to the amount of incoming light to avoid distorting the image:
In considering not just the classical photoreceptors but the entire retina itself as a light-harvesting engine, it seems prudent to also regard its entire synaptic endowment as a molecular-scale computing volume. In other words, when you have many cells that have no axons or spikes to speak of, that can completely refigure their fine structure within a few minutes to handle changing light levels, every synapse appears as an essential machine that percolates information as if at the Brownian scale, or even below. [...]A Darwinian paradigm assumes that biological systems are cobbled together haphazardly by natural selection over eons of unguided descent with modification. Compare that paradigm with one that recognizes intelligent design and that accordingly predicts biological systems are built from the top down. In investigating the evidence of biology, ID expects to find goal-directed structures that are organized much like human technology -- except better, it often seems. Which model seems the more appropriate here?
Most incredibly, like the wings of a swallow, the retina more-or-less works right out of the box, even if it has not seen any exercise. In seeking to understand how it then further refines its delicate structure we should perhaps not overlook the pervasive organizing influence of the incoming photons themselves. Now that it is becoming abundantly clear that the whole works can "feel" them, the next question to answer is how.