The visual equipment of box jellyfish includes eight optically advanced eyes that operate with only a rudimentary nervous system. As they produce blurred images, their function remains an open question.
According to conventional wisdom, information-processing in visual systems is a hierarchical process1. It starts at the level of the receptor layer, the retina, where raw sensory data are taken up from the outside world, and continues by transferring this information to increasingly higher centres in the brain. During this upstream process, exactly those features are extracted from the retinal response patterns that enable the animal to cope successfully with its ecological world.
The findings that Nilsson et al. describe on page 201 of this issue2 run counter to this view. A lowly animal inhabiting the tropical seas — a box jellyfish, or cubomedusa (Fig. 1) — is equipped with eight surprisingly sophisticated lens eyes of the camera-type, but there is no common brain behind them. In nearly every respect, these lens eyes resemble those of animals such as fish and cephalopods, but the ‘central nervous system’ behind the eyes consists only of a diffuse nerve net accompanied by a marginal nerve ring3.
Nilsson and colleagues' anatomical, optical (microinterferometrical) and modelling studies reveal two phenomena. The first is amazing, and the second peculiar. Amazingly, the tiny spherical lenses, which are only a tenth of a millimetre wide, are able to form sharp images, which are free of spherical aberration. This is due to a refractive index gradient within the lens — exactly as one would expect from optical theory, and exactly as is found in the much larger spherical lenses of fish and cephalopods4. Peculiarly, however, the focal plane of each jellyfish lens, and hence the sharp image formed by the lens, lies far behind the retina. This underfocusing leads to a blurred image, and thus to a loss of the fine visual detail that the lens is able to provide.
So why should evolution have produced highly sophisticated optics that have only poor resolution? Isn't higher resolution always better, irrespective of the visual function to be fulfilled? Obviously not. Cubomedusae are strong, agile swimmers and active predators3 living in near-shore habitats such as mangroves5. But how they are visually guided within these cluttered environments has remained elusive. Here we face the problem of what could be dubbed ‘reversed neurobiology’, analogous to the case of ‘reversed genetics’6, in which the genes are known but their functions are not. We know what kind of visual cues the eyes of jellyfish are best at extracting, but not the visual tasks that the animals have to accomplish.
In insects, there are two examples in which the degradation of spatial resolution is a design feature of a particular visual subsystem known to serve a specific function. One is course stabilization, which involves horizon detectors that are built into small, single-lens eyes, the ocelli7. The other is skylight navigation, which is based on patterns of polarized light and mediated by a small, specialized part of the insect compound eyes8. But what are the jellyfish's eyes designed for?
This question is even more compelling, as Nilsson et al.2 find that the photoreceptors of the jellyfish eyes possess wide and often complex (for example, asymmetrically shaped) receptive fields. In mammals, for instance, such complex receptive fields result from multi-level processing and hence are confined to higher cortical centres in the brain9. But in jellyfish they are generated by the optics of the lens and the position of the photoreceptors within the retina. This extreme case of peripheral filtering again hints at a very particular task that these eyes must accomplish. It is further corroborated by the observation that, in each of the animal's four sensory clubs (or rhopalia), there are not only the two types of lens eye studied by Nilsson et al. — one looking upwards, the other horizontally — but also two other (simpler) types of eye (Fig. 2). Within these batteries of eyes, which exhibit quite some diversity of optical design, each type of eye could be specifically adapted to a particular aspect of the animal's lifestyle.
Nilsson and co-workers2 provide us with one of the most dramatic examples that vision is not a general-purpose sense. Box jellyfish evidently have different eyes for different tasks, and have delegated such tasks, which are usually accomplished by neural processing, to the optics of the eyes. Furthermore, as the outputs of the eyes are channelled directly into the pacemakers for the swimming movements, and as these pacemakers are also located in the sensory clubs10, visuomotor processing occurs at an extremely peripheral level. Specialization of eyes for particular tasks and peripheral coding seem to go hand in hand — during the course of evolution, box jellyfish have clearly not had the need to feed the information provided by their total of 24 eyes into a central processing unit, or brain.
Nicholls, J. G., Martin, A. R., Wallace, B. G. & Fuchs, P. A. From Neuron to Brain 4th edn (Sinauer, Sunderland, MA, 2001).
Nilsson, D. -E., Gislén, L., Coates, M. M., Skogh, C. & Garm, A. Nature 435, 201–205 (2005).
Satterlie, R. A. Can. J. Zool. 80, 1654–1669 (2002).
Land, M. F. & Nilsson, D. -E. Animal Eyes (Oxford Univ. Press, 2003).
Coates, M. M. Integr. Comp. Biol. 43, 542–548 (2003).
Weissmann, C. Trends Biochem. Sci. 3, 109–111 (1978).
Stange, G. J. Comp. Physiol. A 141, 335–347 (1981).
Wehner, R. J. Comp. Physiol. A 189, 579–588 (2003).
Schweigart, G. & Eysel, U. T. Eur. J. Neurosci. 15, 1585–1596 (2002).
Satterlie, R. A. J. Comp. Physiol. 133, 357–367 (1979).
Laska, G. & Hundgen, M. Zool. Jb. Anat. 108, 107–123 (1982).
About this article
Natural Computing (2018)
Journal of Evolutionary Biochemistry and Physiology (2009)
Evolution: Education and Outreach (2008)
The American Biology Teacher (2005)