This very strange object was peering out at me from the cover of last week's Nature…and "peer" is exactly the right word. Those are some of the eyes of a cubozoan, a box jelly, of the species Tripedalia cystophora. These eyes have some very peculiar features, and show that once again nature trumps the imaginations of science fiction artists.

a, The rhopalium shows the upper and lower lens eyes flanked by two pairs of simpler eyes. b, c, The live lower eye displays a mobile pupil. In b the eye was exposed for about 10 min to light intensities corresponding to direct sunlight, which is enough to close the pupil maximally. The fully open pupil in c is the result of total darkness for 10 min. Pupil adjustments take about 1 min.

The jellyfish has four sets of eyes, each set at one corner of the animal, and each set clustered in a club-shaped structure called the rhopalium. In each rhopalium, there are actually six eyes: the two dark smudges on each side in the above picture are simple pigmented pits with photoreceptors in the center. Then there are two complex camera eyes, one pointing up and a larger one pointing down, each with a retina and a lens and a pigmented epithelium—the larger downward pointing one even has an adapting iris that closes down in response to bright light, and opens in darkness. Those are remarkably sophisticated structures for a 'simple' box jelly, and even resemble superficially the eyes that stare back at me when I look at my zebrafish embryos!

This diagram will help you sort out the arrangement of the apparatus here. The jellyfish has an intricate and specialized camera complex hanging out on the end of a stalk in each rhopalium.

An accurate anatomical model. The sagittal section contains the statolith and the internal structure of the two lens eyes. The spherical lenses are surrounded by a cellular capsule, the inner part of which forms the equivalent of a vitreous body between lens and retina. Iris constriction in the large eye is caused by contraction of the outer part of the lens capsule. The lower eye is rotationally symmetrical, but the upper eye is only bilaterally symmetrical (front view shown to the right). Receptor outer segments fill the retina of both lens eyes. The alignment of receptor outer segments is unusual, especially in the upper eye, where receptor axes converge on a point at one side of the lens. Scale bars, 100 µm.

Even more impressive, the investigators have plucked out the lenses and investigated their optical properties, and discovered that they are actually very, very good lenses. The refractive index of the lens material varies from the core to the surface, naturally correcting for spherical aberration. The topmost eye is nearly perfectly corrected, and can focus light on a sharp point at a distance of 3.3 radii from the lens center. The lower lens is less well corrected, but still produces a point of focus at 2.6 to 3.7 lens radii.

Now here's the odd part, though: they have these very well tuned lenses, but look at the diagram up there: focusing light at 3.3 lens radii puts the focus way, way behind the retina. Jellyfish are horribly hyperopic, or far-sighted! It's as if they've gone to all the effort of evolving a nice set of lenses, and then thrown away all the advantages by grossly mis-focusing them. In the diagram below you can see the effect: light from a distant source is not sharply focused on a single photoreceptor, but at the plane of the retina instead lights up a whole diffuse region.

The optical models were used for tracing rays through the lens and retina and computing the absorption of light in selected single photoreceptors (red bars). Rays were traced in three dimensions, and by calculating receptor absorption at different incident angles of the ray bundle it was possible to generate receptive field maps for any receptor in the eyes.

Nilsson et al. analyzed from optical models what each photoreceptor should be able to see, and it turns out that they don't have much acuity at all—each photoreceptor gets light from an approximately 20° field of view.

What is the jellyfish doing with this eye?

The explanation probably lies in another simple fact: jellyfish don't have brains. They aren't going to be carrying out complex signal processing on visual information, and are only interested in very vague things: Is it day or night? Which side of me is in shadow? Where's up and down? They aren't going to be able to grasp the details of a copepod swimming next to them, and don't need to know about it—tiny motes dancing about are distractions, not information.

The authors point out that our brains do similar things. We retain more detailed spatial information for some kinds of processing, but other regions in our visual centers clump together the output of subsets of photoreceptors, generating the same effect at a higher level as the jellyfish do by the simple expedient of defocusing their lenses.

Because we do not yet know how the visual information from the lens eyes of box jellyfish is processed and used, we cannot tell what purpose the peculiar sensitivity functions might serve. But it is intriguing to note that many neurons in higher visual centres of the vertebrate brain also have large and geometrically complex receptive fields. A typical feature of animal visual systems is that higher processing occurs in parallel pathways where each pathway handles a specific aspect of information such as large-field motion detection or feature recognition. The large and complex receptive fields of neurons found in vertebrate higher visual centres represent highly filtered information needed for specific visual tasks. In box jellyfish we find these large complex receptive fields at the level of photoreceptors, indicating that the eyes might be specialized for a specific task only and that this allows complex filtering of information much earlier than in more general visual systems. The fact that box jellyfish have four different types of eye gives support to the idea that each eye type is highly specialized.

This is an article about anatomy, physiology, and optics, but again that wonderful integrating theory of evolution is an important part of the biology. The authors point out that complex structures evolve to provide immediate and local utility, and that we should not presume that because we use lenses and retinae to capture images with high spatial resolution, eyes would not necessarily have initially evolved to meet that specific task.

The early evolution of animal visual systems is likely to have started out with eyes that were involved only in single visual tasks. In this perspective it is interesting to note that high visual acuity is not necessarily desirable. The lens eyes of box jellyfish indicate that there might be visual tasks best served by a blurred image. Evolution of sophisticated eyes might therefore be a process with discrete stages representing the sequential addition of visual tasks. Our results also indicate that advanced lenses with graded-index optics might have evolved for tailoring complex receptive fields and not just for improving sensitivity or acuity.

Nilsson D-E, Gislén L, Coates MM, Skogh C, Garm A (2005) Advanced optics in a jellyfish eye. Nature 435:201-205.