most birds. This is because in these species the eyes are placed on the side of the head and the visual field is often maximised, which requires use to be made of the entire image (Figure 3.3).
FIGURE 3.3 A diagrammatic section through the head of a bird showing a typical arrangement of eyes in the skull and how the visual fields of each eye combine. In all birds the eyes project laterally so that the axes of the eyes always diverge; no birds have forward-facing eyes. The fields of view of the two eyes are combined to give the total field of view, with a sector in front of the head where the two fields overlap to give a binocular field. A wide degree of variation in these basic arrangements is found in birds, resulting in different degrees of overlap, different width blind areas behind the head (though some birds have no such blind area). Just small variations in the width of the field of view of each eye, and of eye position in the skull, can result in large differences in visual fields between species. (Diagram by Nigel Hawtin, nigelhawtin.com.)
In some species, full use is made of an image from each eye that is more than 180 degrees wide. This gives the birds maximum visual coverage of the space around them. In many birds the width of the visual field behind the head is maximised in order to enhance the chances of detecting a predator, but this again can be achieved only by using peripheral optics (Figure 3.4).
FIGURE 3.4 Examples of the extremes of visual fields found in birds. In a Tawny Owl Strix aluco the axes of the eyes project laterally and forwards. The field of view of each eye is relatively narrow, and the eyes sit in the skull to give a relatively large degree of binocular overlap and an extensive blind region behind the head. In a Pink-eared Duck Malacorhynchus membranaceus the fields of each eye are extensive, a little over 180 degrees, and they project laterally to give the bird a small degree of binocular overlap both in front of and behind the head. This means that it sees all around its head in the horizontal plane. In fact the binocular region extends right above the head, and the duck has panoramic vision of the hemisphere around and above its head.
This lateral placement of the eyes in the skull is quite unlike the situation in ourselves. The optic axes of our eyes, and hence the best-quality optics, project directly forward. We do not try to look in the direction that we are travelling out of the sides of our eyes. This is, however, what all birds do to some extent. No bird species, not even owls, have eyes positioned to face directly forwards, and many birds look forward with the very periphery of their eyes’ optical systems. The consequence of this arrangement is that the best-quality optics in all birds projects laterally, away from the axis of the head, in some species markedly so. This has important consequences for understanding both the foraging behaviour and the role of vision in the control of locomotion in birds. It will be discussed in Chapter 5.
Another important property of the image is how much of the world is imaged at one instant. Does the imaging device have a wide or narrow field of view? This is important, since it determines from how much of the world around an animal’s head information can be gained at any instant.
The image analysis system
It is the retina that starts the process of image analysis. The retina extracts the essential information that the image contains, encodes it neurologically, and sends it via the optic nerve for further analysis by the brain.
When looking into any eye we see the pupil as a black void. We are looking through the optical system and through the thin transparent neurological layers of the retina to the uniformly black surface, the pigmented epithelium, that lies behind it. As outlined above, each retina contains many millions of individual neural cells arranged in distinct layers. The most prominent of these layers contains the photoreceptor cells, the well-known rods and cones, and these are discussed in detail below.
When light photons reach a photoreceptor, a neural signal is generated and relayed to a ganglion cell which in turn relays that information through the optic nerve to the brain. Of crucial importance at this first level of image analysis are the number, density, and distribution of photoreceptor and ganglion cells across a retina. The actual numbers of photoreceptors are very high. For example, in the eyes of eagles (and humans) the total number of cells in the whole retina probably exceeds 100 million. In all retinas, however, the number and density of photoreceptor cells are far from uniform. In some locations photoreceptor cells are packed close together, in others they are more sparse, and large differences in density can occur between locations less than 1 mm apart. In an eagle’s retina density can peak at about 450,000 photoreceptors per square millimetre, and in humans it reaches about 200,000, but less than 1 mm from the site of peak receptor density it drops to 16,000 photoreceptors per square millimetre. However, these changes in receptor density do not occur randomly across the retina; they occur in distinct patterns in the eyes of different species.
The patterns of photoreceptors can be revealed and characterised using isodensity contour maps (Figure 3.5). These link locations across a retina which have the same cell densities. In the same way that contour maps link locations of the same elevation and allow us to quickly appreciate the topography of a landscape, these density maps of retinal cells provide a quick way to compare retinas in different species and hence are a ready means of comparing some basic aspects of vision between species. A number of examples of receptor and ganglion cell maps will be discussed in later chapters.
FIGURE 3.5 Examples of isodensity maps of the ganglion cells in bird retinas. In each diagram the retina has been spread flat, but its orientation is as in the eyes in the intact birds shown above. Flattening the retina causes splits, hence the rather ragged shape. The densities of the ganglion cells (×1000 per square millimetre) have been analysed across the whole of the retina and points with similar density have been joined to give contour maps, much as a topographical map links places of the same height. Clear patterns emerge in these maps showing how the images projected onto the retina by the eye’s optics are analysed to extract different degrees of detail. On the left is the retina of a Manx Shearwater Puffinus puffinus, and the map shows that their retinas have a band running horizontally across the field of view in which details are particularly resolved (receptors are at high density, providing greater detail). On the right, the Rock Dove Columba livia shows a retina with two distinct areas from which detailed information is extracted: one looks out close to the axis of the eye (in this view almost directly out of the page), while the other also projects laterally but downwards and slightly forward within the bird’s field of view. (The diagram of the dove is redrawn from work published by Bingelli and Paul.)
The importance of these density patterns can be appreciated by considering the photoreceptors of a retina to be analogous with the photodiodes of the receptor surface of a digital camera. In a camera we understand that the photodiodes are responsible for pixelating the image, and we expect that the photodiodes are not jumbled but spread at an even density across the whole image-analysing surface. This guarantees that the same amount of detail is available across the whole of the image. However, in retinas the densities of receptors and ganglion cells vary markedly across the image surface. Furthermore, there are consistent and different photoreceptor and ganglion cell patterns in the retinas of every bird species. It is as though we were able to choose between cameras not just on their overall density of photodiodes, but also on how the photodiodes are placed across the image surface. It is as if we could choose between one camera that analysed the image in greater detail at its centre, another that could analyse with greater detail in a band horizontally across the middle of the image, another more to one side, and so on. An endless number of possible arrangements would be possible.
Such patterns are indeed found in bird
