NEXT time you walk past a field or hillside full of sheep do not suppose that they are busy automatons with eyes only for the next mouthful of grass. They have sensed your presence, identified you as a human and established whether you are moving towards them or not. If you do not pose a threat, they will take stock of which individual animals are in the near vicinity and, providing all seems well, tackle the essential problem of which clump of grass or clover presents the best bet for the next tasty mouthful, or settle down for a spot of rumination. Above all, the sheep is aware of every change in its environment and its senses are geared to assess such changes with optimum speed and accuracy.
During the past decade at the Agricultural and Food Research Council Institute at Babraham, we have studied the way that sheep and other farm animals perceive their environment, to better understand their needs in an agricultural context. But the research also throws light on how we, too, recognise important objects and individuals in our environment.
Advertisement
Sheep can see almost completely around themselves, but their sight is most acute in the frontal visual field where the two eyes overlap. They rely heavily on their keen eyesight, as Valerius Geist established 20 years ago in his observations of feral breeds in their mountain habitats. In his book Mountain Sheep, Geist wrote: ‘Their eyesight is marvelled at by hunters, and a popular myth circulates in North America that sheep vision is equal to that of man aided by 8-power binoculars.’
He found that the animals can spot both humans and coyotes at distances greater than 900 metres, even when they are partially hidden by shrubbery or when conditions make it impossible to rely on odours. Sheep can also use visual cues from specific body features to recognise another animal and its social position. In most feral breeds, as well as some domestic ones, the presence and size of horns indicate an individual’s position in the dominance hierarchy – the larger the horns the more socially dominant the animal is likely to be. Simply removing an individual’s horns can have drastic effects on its position in the social hierarchy. The size of horns also provides a guide to sex, as females either lack horns or have smaller horns than males.
Earlier work by Elizabeth Walser at the institute has shown that ewes probably visually recognise their lambs by looking at their faces. In her experiments she coloured various parts of the lamb’s body and observed how the mother ewe reacted. Only when the head region was coloured, or the whole body including the head, did the ewe have difficulty in recognising its lamb. My own observations on an orphaned lamb that I raised by hand suggested that even though the lamb was just a few weeks old, it could distinguish me from other humans by sight and that it clearly had trouble recognising me from a distance if I covered my head. But as the lamb grew, it became better at recognising me from a distance, so experience must be important too. Sheep probably use visual cues to help them to identify food as well. Even where they cannot smell particular foods, they still find their favourite ones easily. Similarly, in the field they seek out patches of dark green, lush grass, or clover in preference to less tasty, but more plentiful, normal light-green grass. Field tests at the Welsh Plant Breeding Institute in Aberystwyth have shown that when lines of clover are sown in a field full of grass, the sheep will graze the clover lines and leave the grass untouched.
In more controlled laboratory settings, Bob Baldwin at the institute has shown that sheep can be trained to distinguish between a variety of geometrical shapes or those placed at different orientation. Similar experiments have established that sheep have some form of colour vision, although no one has yet proved that they actually use colour, as opposed to differences in luminance, to distinguish between objects. Essentially, the eyes and brain of a sheep process some visual images in the same way as we do. But their visual system has some distinctive features. With its eyes positioned slightly to the side of the head, a sheep’s field of view spans 290Degree , of which only 50Degree is the frontal field where the two eyes overlap to give the sharpest vision. Compared with the eyes of a primate, those of the sheep are less well adapted to the fine discrimination of objects. Their eyes lack an accommodation reflex, which in our eyes alters the shape of the lens to bring objects into focus. Their eyes are also probably astigmatic, with the curvature of the lens differing in different planes, which suggests that sheep see less clearly than a person with ‘perfect’ vision. Both sheep and people process a visual image first in the retina at the back of the eye. The sheep’s retina has a central region where vision is more distinct, similar to a primate’s fovea. The density of ganglion cells in the retina probably indicates visual acuity, and in a sheep’s eye this density is midway between that of a cat and a human.
In the next stage of processing, information from the retina passes down the optic nerves and, after further analysis by thalamic nuclei, reaches the primary visual cortex at the back of the brain. In the late 1970s, P. Clarke, D. Whitteridge and V. Ramachandran at the University of Oxford recorded the electrical activity of single nerve cells in this region, and found that the structure and organisation of a sheep’s primary visual cortex is similar in structure and organisation to that in primates. Unlike newborn primates, however, lambs have a fully mature visual cortex at birth, and so can use their eyes to learn about the world right from their first day of life. At what stage in the processing of a visual image do we – or sheep – recognise objects or individuals? Initially, a visual image is divided up into millions of different components as it is successively processed by the retina, thalamic nuclei and the primary visual cortex. In the various layers and subdivisions of the visual cortex, particular neurons become active, as they code for simple information relating to aspects such as form, colour, brightness and movement. But there is little evidence that the neurons are specifically coding for meaningful objects or individuals.
Recognition does not seem to happen until the information is relayed to other parts of the brain – to areas of association cortex or subcortical structures. In these regions of the brain we start to find populations of neurons that change their activity in response to the sight of specific objects or individuals, as opposed to the myriad components from which they are composed. These regions also connect with areas of the brain involved in memory and the control of emotions and behaviour. The behavioural studies of sheep suggest that, like primates, this species should have centres in the brain that specialise in the visual recognition of important individuals and food. We have tested this possibility by recording the electrical output of single nerve cells in a variety of cortical and subcortical structures of a sheep’s brain while it is shown pictures of a variety of animal or human faces and foods. Nerve cells in the brain communicate with each other through the production of electrical impulses. Information is coded by the frequency of these impulses, and the idea behind this technique is to find cells that change the frequency of their electrical output only when the sheep sees pictures of faces or food.
Our studies have shown that a small population of cells in the temporal cortex of the brain codes specifically for the sight of faces (see Figure 1). This suggests that in sheep, like nonhuman primates, this region is specialised for processing information relevant to identifying individuals from their faces. In humans, too, damage to this general area of the brain, due to strokes or other traumas, can lead to problems in recognising faces. These cells usually respond within 100 milliseconds of the facial image becoming visible.FIG-mg17164101.GIF
In a sheep’s brain, specific groups of faces or facial features are coded for by separate subpopulations of cells. Most cells respond only to faces with horns, and the larger the horns the greater the change in their output frequency. Even crude line drawings of sheep faces show this effect of horn size (see Figure 2). A second group of cells responds only to faces of animals of the same breed (such as a Dalesbred sheep) and particularly familiar individuals. A third group of cells responds exclusively to the faces of humans and dogs. So sheep deal separately with information relating to dominance features, faces of other familiar sheep and faces of potentially threatening species, such as humans and dogs.FIG-mg17164102.GIF
The orientation of the facial stimulus is also important (see Figure 3). Frontal images are more potent stimuli than profiles or views of the back of the head. This probably reflects the fact that direct eye contact, as sheep view each other face to face, is more likely to lead to some form of social or aggressive encounter. But unlike the temporal cortex cells in primates, these in sheep do not respond to faces upside down. Perhaps this is because sheep, in contrast to arboreal primates, have no need to recognise each other upside down! Sheep can recognise important individuals from their faces over only a fairly short range. Yet they can spot a dog or human approaching from a long way off. How do they accomplish this? The answer is that they probably use cues from a body’s shape and posture. A sheep’s temporal cortex also has cells that respond to the sight of a human shape. Their activity tells us a great deal about how a sheep recognises a human and interprets the significance of one appearing in its environment.FIG-mg17164103.GIF
Sheep cannot distinguish between humans, their sex, what they are wearing, whether the back or front view is presented or if the head and shoulders are covered. Equally, these cells do not respond to these various aspects of people. They are, however, influenced by the direction the human is moving in. Most cells respond only when a human moves towards the sheep; only a few respond to stationary or withdrawing images. This is not surprising, as an approaching human is a much greater potential threat to the sheep than a stationary or retreating one. When humans are viewed from the side, or in a quadrupedal, as opposed to a bipedal, posture the responses of these cells are diminished or disappear (see Figure 4).FIG-mg17164104.GIF
So sheep may not be able to recognise humans if we change from our normal bipedal posture, or present a difficult viewing angle. ‘This could explain why they, and other domestic animals such as cattle, will often let a crawling human approach much closer to them than a walking one. As standing on two legs is obviously one of our more distinguishing features, it would seem logical for animals to use this to identify us. (This is not to say, of course, that the sheep would not eventually learn to redefine their criteria for human if we always appeared to them on all fours.) In another subcortical region of the brain, the hypothalamus, we have found populations of cells that appear to specialise in the visual recognition of foods. These cells are similar to those that Edmund Rolls and his colleagues in Oxford found in primates in the late 1970s. The cells respond to the sight, but not smell, of known palatable foods and not to non-food objects. The more the animal likes a particular food, the better these cells respond to the sight of it. When the animal no longer wishes to eat a particular food, the cells cease to respond to the sight of it.
These cells can even indicate whether an animal has learnt to recognise a new food. Initially, the cells do not respond to the sight of an unknown food. But if a sheep eats the food just once and likes it, the cells will respond the next time it is seen, even if the sheep has not seen it for a month or more.
Our combined studies of the behaviour of sheep and the way that their brains process important information provide unique insights into the way sheep categorise and learn to recognise important objects in the world they see around them. It turns out that the principles of neural processing embodied in a sheep’s brain are remarkably similar to those found in monkeys and probably in humans as well. This is not to say that sheep are as intelligent as monkeys or humans. What it does show is that the way the brain carries out the complex task of visual recognition is remarkably similar in different species.
Another important principle to emerge from this work is that the neural processes underlying visual recognition are intimately linked with the processes governing a sheep’s emotional or behavioural responses. For the sheep, visual recognition of important objects at the level of the temporal cortex or hypothalamus is not simply a matter of detecting complex features. Objects are also seen in the context of their emotional or behavioural importance to the animal. For example, some cells respond similarly to canine and human faces. The emotional significance of these two faces is the same, whereas their appearance is clearly not. If such cells were simply feature detectors, one would expect a dog’s face to be categorised along with the faces of sheep – a dog’s pointed muzzle is clearly shaped more like a sheep’s than a human’s face.
So the visual analysis of images is influenced by their emotional or behavioural consequences. For this to happen, there must be complex interactions between brain regions controlling the two processes. Our perceptions are not simply coloured by our emotions, they are inextricably linked with them.
Why should the brain merge the two mechanisms: visual recognition and the emotional and behavioural responses to what is seen? The most obvious answer is that this approach to analysing visual objects speeds up the whole process, so that an animal, or human, can respond quickly to what it sees. Decisions about the appropriate responses are being made at the level of sensory analysis, some time before information has passed to the parts of the brain responsible for executing such responses. The evolutionary advantage of this system for animals with many predators is obvious.
But there is a trade-off between speed and accuracy, and the accuracy of visual identification takes a back seat to speed of response. For example, a sheep may approach a crawling human, or a dog that looks like a sheep, before it has had time to check the accuracy of its identification. We have all experienced the results of this same trade-off when we mistakenly approach and greet as a friend a stranger we have only glimpsed briefly.
Learning clearly plays an important part in determining the way the brain categorises objects. We have shown how quickly sheep can learn to recognise food, and we are confident that learning must also play an important role in the recognition of individuals. We are currently investigating, for example, the effects of bringing up horned animals with non-horned mothers. Does a lamb reared with no experience of horned animals have cells that selectively respond to horned faces? If not, how long does it take them to develop when they have had such experience. This ability to examine the influence of learning at the level of a single cell is an extremely powerful model for investigating the way the brain carries out complex tasks. From the point of view of understanding the behaviour of sheep, such information will also give us a direct measure of the way their environment dictates the way they perceive and value different objects.
Medicine may also benefit, if we can go on to understand how learning modifies the interaction between sensory processing of images and the making of emotional and behavioural responses. In people suffering from schizophrenia or autism there is often a breakdown in the link between recognising objects and individuals and responding appropriately to them. If we can show how learning influences the integration of these two processes, we may come to a better understanding of these disorders.
Dr Keith Kendrick is at the Agricultural and Food Research Council’s Institute of Animal Physiology and Genetics Research (Cambridge Research Station), Babraham, Cambridge.