When sentimental humans call their pet 'good dog', they usually mean only that the dog happens to have done what they wanted him to. We sometimes call babies 'good' for much the same reason, and not because they ought to be imitated by anyone who wants to live a good or an admirable life. Even when we say, 'So-and-so is a good man', we may mean only that the man is a useful worker, to be praised and cosseted and used. But we also know that good men and women ('morally' good men and women) are those who can be counted on to do what they ought, and for the right reasons. They are courageous and kindly, loyal, honest, temperate and just. They have good characters and do what they ought to do because they see they ought.

We usually assume, on the other hand, that animals let nothing stand in the way of their desires. Their wants are simple - like the dog with one thought for each paw (food, food, sex and food) - and anything that satisfies those wants will do. Whereas human beings do not willingly eat everything that is strictly edible (or there would be thriving cockroach farms in every American city), animals will eat anything that their stomachs can digest and that they can capture. Whereas human beings seem to love to make difficulties for themselves in sexual affairs, animals respond to lust as they would to an itch. This is not to say that animals do not have preferences, but they do not seem to have taboos. To live 'like an animal' (especially in the mouths of judges) is to live without any of the acknowledged restraints of decency, good manners or respect for persons. This sounds like a good idea to those romantically inclined to reject civilization, overturn tradition and begin again as noble savages. It usually sounds like a very bad idea to the rest of us. Civilization depends upon our not doing what we immediately and unthinkingly want to do (kill jay-walking pedestrians, steal books, seduce minors).

Those who disapprove of the behaviour of animals naturally feel a similar distaste for alien (presumably savage) human customs. Samuel Johnson, the great moralist and lexicographer who died two centuries ago, could not believe that 'savages', illiterate peoples, could have anything to teach him. According to James Boswell, in his Life of Samuel Johnson, he declared, 'Pity is not natural to man . . . but acquired and improved by the cultivation of reason. Savages are always cruel.' 'Natural affection is nothing: but affection from principle and established duty is sometimes wonderfully strong' — so savages have no more affection than do hens. Nor do they marry: 'a savage man and a savage woman meet by chance; and when the man sees another woman that pleases him better, he will leave the first.' Johnson's determined ignorance is now an embarrassment to his admirers (including myself). Why could he not have understood that other human tribes have their own arts and decencies, that they do not merely act out their momentary whims? He spoke from within his tradition, as the Greek philosopher Aristotle did when he declared that the more distant barbarians were beast-like, in that they lived 'only by perception', without — he supposed — being able to give principled reasons for their actions and without any long-term goals.

Every decent moralist is now conscious that all human tribes have inherited cosmologies and political systems. We hope, at least, that there are no 'natural slaves' of the sort Aristotle (unfortunately) taught exploring Europeans to expect, lacking any moral conscience and acting out of immediate desire or fear. But Johnson's attitude to animals — the view that they too are moved only by the prospect of immediate pleasure - is still widely held. 'Anthropomorphism' is the deadly sin of supposing that animals have customs, friends, serious emotions or needs beyond the merely physical. Most commentators recognize that contempt for 'savages' serves ideological and commercial interests, giving us an excuse for disrupting the savages' life, turning them out of their homelands and refusing to accept that they need to be able to control their own lives in accordance with their own traditions. That contempt for animals serves similarly ideological ends is not as widely recognized. To behave 'like an animal' is to have dropped out, to have abandoned cultivated manners and an awareness of one's place in the social universe. To be an animal is to be mere material for the purposes of human beings, whether those purposes are humane or not.

The older habit, and one reason why people now disapprove of anthropomorphism, was to describe animals in entirely human terms. People once believed it literally true that the lion was king of the beasts, that the world of wild things and the world of civilized humanity (which is to say, our kind of human beings) were built according to the same pattern. All animals were members of the kingdom of animals, as though difference of species was no more than a difference of class or profession (and the latter no less than a difference of species). Animals had their own mysterious language, their own law. They did what people did: if one animal killed another, he displayed the same sort of ferocity as would a human warrior. If one looked after another, it was out of maternal or comradely compassion. Moralists took examples from the behaviour of animals even though they also held that animals were really moved by nothing more than desire or fear. For these moralists, the animals were not thinking about what would happen to the creatures they affected and so could not 'really' be compassionate, or soldierly, or loyal.

The older way of thinking about animals was clearly confused, and some scholars were led to attribute far too much human intelligence and moral sensibility to animals. This led others to try to describe animal behaviour without committing themselves to any view about what purposes or perceptions they had. When modern students of animal behaviour say that an animal is 'aggressive', they mean only that the animal goes through certain motions that can usually be expected to result either in a fight or in the withdrawal of the opposing animal. They do not mean to imply that the animal actively wishes to hurt its opponent, or even that it knows it has an opponent. If a stickleback can be made to attack an unrealistic model of a fish with a red belly, we do not need to think that when it fights a real male rival it is 'genuinely' angry, in the way that we sometimes are. When we describe what people are doing we use our knowledge of their motives to distinguish between different acts: when Zachary kills Tamar it is not murder unless he really intended to kill her or hurt her very badly. Orthodox ethologists have abandoned the attempt to say what animals intend or want, and the words they use are not supposed to imply anything about the animals' own feelings. By keeping to what can be measured and recorded on camera or tape, they hope to avoid the perils of anthropomorphism.

This approach is a helpful one when we are dealing with some animals, those that have relatively few options and those with whom we do not readily empathize. If woodlice congregate in damp, dark patches, we need not suppose that they have some idea of what they are looking for, or an internal map of the territory, or any wish to greet their friends and neighbours. It is enough that they move faster when it is dry, and slow down when it is damp. If salmon can find their way back to the stream where they were born, we need not think that they know where they are going, nor do we need to imagine them fighting heroically with the current. They are only swimming towards a stronger concentration of some chemical in the water. If hunting wasps construct nests and supply their future progeny with paralysed caterpillars, it is not because they wish their offspring well but because they are acting out 'fixed action patterns', each one released by the successful completion of its predecessor. If the caterpillar is removed, the wasp will still seal up the nest and move on to the next one.

But though there are good reasons not to read too much of our own experience into the behaviour of animals, and though it is sometimes helpful to attempt as 'objective' a description as possible of what they do, the philosophical assumptions behind this programme are very odd. It is certainly often difficult to know what other people are feeling and thinking. It even makes sense, of a sort, to wonder whether the things we call people are perhaps really cleverly designed robots, whose behaviour is merely physical and who have no subjective life at all. But anyone who seriously concluded that this made it reasonable for him to treat people as if they were indeed nothing but insentient robots would be thought deranged. There are general difficulties about how we can form reasonable beliefs about minds other than our own. There are also general difficulties about how we can be sure that there are real material bodies: it makes sense to suppose that there are none, that all our experiences of closed doors and stubborn boulders are merely mental. Some modern physicists have indeed drawn the conclusion that electrons and photons and the rest of the particles that theory demands are fictitious, that the 'laws of physics' really refer only to the sorts of observation that physicists might make, not to any real world independent of their observations. It is one of the ironies of history that life scientists are much more materialistic than physical scientists. Irony apart, it is at least very peculiar that students of animal behaviour should think that merely 'physical' observations (e.g. how fast a thing is moving) are reliable indicators of the real world, while empathetic understanding of what others might be feeling and thinking can never be relied upon. If we cannot understand each other, all science collapses, since we need to be assured that our colleagues are honest and rational observers. If we refuse to let ourselves understand what animals are doing, if we never let ourselves see things, as it were, with a gull's eye, or a baboon's, how much are we likely to understand? Merely 'physical' description ('And now the chimpanzee's hands have contacted an empty oil can, and the can is rolling around the clearing, and the chimpanzee is emitting sounds') may be an aid to acute observation, but we have a more secure and useful understanding of the event when we know that the chimpanzee is taking advantage of human rubbish to impress his group.

In short: to say that a man has lost his temper is no less an observational statement than to say that he has lost his trousers. The evidence that either is true is, in a sense, compatible with its being false: maybe the man is acting, or maybe there is an optical illusion. Students may devise interesting and useful theories about what is going on at a neurophysiological or chemical level when a man has a tantrum. It may even be that a knowledge of the laws of chemical change would be enough to predict the motions through which his body will go (if only we had enough time to do the calculation). But even such a precise, physical theory would not prove that he was not genuinely in a temper, nor would it prove that his being in a temper was not a good explanation for his silence, his tense muscles, his inability to open an envelope tidily, his expressed belief that he has suffered a serious injustice and so on.

Scientists who profess to believe that animals have no accessible inner life are rarely consistent. If this were really their belief, they would consider it a waste of time to try to anaesthetize an animal and would certainly not draw conclusions about the psychology of human beings from the motions of non-human beings. It is perhaps more usual to think that animals do have feelings but that these feelings do not involve any lengthy foresight, nor any concept of the animal's place in the world and in society. This is in essence the traditional view: not that animals are machines, but that they are moved solely by immediate desire or pain. When a rat learns not to run over an electrified floor, this is held to be a mere conditioned reflex, not an intelligent assessment of the situation: although the rat is repelled by the sight or smell of the floor, it does not know why.

Although this view of things is not wholly unbelievable when applied, say, to amoebae, only those who still think that animals and human beings belong to separate kingdoms can easily suppose that chimpanzees are more like amoebae than they are like humans. If no animals except ourselves ever really think (i.e. grasp what is going on and what might be expected to follow, and respond not only to immediate sensations but also to the imagined causes of those sensations), how is it that human beings can think? Are we really alien or supernatural creatures? It seems very much more likely that our minds as well as our bodies resemble those of other animals -some more than others. If hunting wasps do not really care for their young, it is clear that primates do. Vervet monkeys, for example, not only recognize their own cub's cry but recognize too whose responsibility another cub may be (and look towards her). Adult affection for the young may not be genuinely altruistic — young hamadryas baboons must sometimes wish that their elders were not quite as passionate in their pursuit. But the tests that show that wasps do not have any interest in their young are ones that any reasonable higher primate can pass. Monkeys who are reared by human (but inhumane) psychologists in loveless environments, with only imitation 'mothers' to cuddle, themselves make lousy mothers and treat their offspring as they might a rat or an uncomfortable growth. Normal mammals respond appropriately to their offspring's call. Normal primates, in particular, recognize each other as individuals and have clearly personal relations with their fellows.

More generally, to say that animals are 'only' responding to sensory cues, and not to any more global grasp of the situation and their own role in it, is not really a simpler explanation. If an animal is to respond appropriately to a painful stimulus, it must be acting out an innate, fixed action pattern. Even to learn from experience we must already be acting, consciously or not, on the principle 'Do what brought us satisfaction last time.' So there seems no final reason why we should not admit the existence of other general principles of action — dispositions to behave in one way rather than another. Natural virtues arejust such dispositions. Moral virtues, indeed, are dispositions that the agent has deliberately acquired. A morally virtuous man has moulded himself to play some part in society that he and others reckon valuable. Maybe non-humans cannot train themselves. It does not follow that they have no natural dispositions or that they have no grasp at all of what goes on.

Even creatures whose behaviour does at first seem to be merely a response to sensory stimuli, in accordance with their natural disposition, may be more complex than we thought. While woodlice need have no internal map of their territory, worms perhaps do: at any rate, they reconstruct their tunnel systems. Many animals, indeed, so far from responding only to present stimuli, operate largely in terms of a learned map of the area, which is why bats sometimes bump into things, and laboratory mice can be induced to leap into empty space with the conviction that there is a safe landing. This is how self-consciousness arises, the capacity to locate oneself within physical and social space (like the vervets), to know where one is and whom one is dealing with and what is expected of one. There is good reason to think that animals may be self-conscious (in differing degrees) and that they can manipulate their companions because they can form an idea of how those companions will respond to their own actions. Witness the young chimpanzee who walks away from a luxury he is too low in the hierarchy to claim, knowing that the others will follow him: a little later he returns secretively to get the treat. Witness also such 'problem dogs' as manipulate their human owners into taking them for walks or never leaving them alone.

The ability to identify others as individuals and to recognize oneself as an object in public space is perhaps connected with the sort of upbringing animals receive. Creatures who produce a lot of young, of whom only a few will survive, are unlikely to recognize or care for them, or for anyone else, as individuals. Creatures who have few, slowly maturing offspring can be expected to care for those offspring. Since such care will require that they be able to provide for them, they will not, in general, wish to do what produces offspring unless they can count on ample provision - unless, that is, they have a territory that will support them (this is not to say that they necessarily think this policy out). This is why birds form couples only when there is territorial space available and may (in some species) be attached precisely to the space rather than to their individual partners. In those species the appearance of marital fidelity is an illusion: what generally keeps the same birds together is that each has an attachment to the territorial space. Those who are unable to make good their claim upon such a territory do not form couples or produce many offspring (except, of course, by 'cheating' -laying eggs in an established couple's nest or seducing the female). In other species the problem is dealt with by their ability to recognize each other as individuals and their being bound to marital fidelity. Barbary doves, for example, have been shown to be monogamous, to be faithful to their first partners even at unfamiliar nesting sites.

These patterns of preference can be shown to make sense in terms of the needs of the offspring and the nature of the animal's Umwelt (which is to say, the environment as it is for an animal of that kind, with those senses, capacities and preferences). Creatures that characteristically produce a few, slowly maturing offspring will not be indifferent to their offspring, or mate promiscuously (which would waste energy), or be unable to distinguish individuals of their own kind (unless they are strongly territorial creatures, which feel about a piece of land as a gander does about his goose).

This conclusion, that mammals, like ourselves, will care for their children as individuals, should not be exaggerated: though human beings have fewer and more slowly maturing offspring than, for example, the domestic cat, it does not follow that human beings cannot treat their young with a comparable sternness. What does seem clear is that some birds and mammals, at least, will be capable of forming personal attachments and will be aware of their own position in the world and in society. Without such attachments, without such awareness, creatures of their kind would not survive long enough to reproduce. Among the natural virtues of the higher mammals, at any rate, will be those of parental care and faithfulness. We can identify other virtuous dispositions that animals are likely to display. Members of the same species are natural rivals for food and  territory and  mates.  But it does not follow that arrogant individualists will be most successful in propagating their kind. Creatures that always fight to the finish, that will never accept defeat gracefully, that always kill or rob their rivals may win an occasional battle, but they must spend so much time and energy on forcing their will on others that any timid mutant which avoids fights will be able to leave behind more offspring. Rather than fighting directly for the goods they desire, animals are likely to try out their strengths in a way that does not seriously damage anyone. They do not usually use their most dangerous weapons against their rivals, and losers accept their lot. They may even (in disaster areas) let themselves starve to death while the dominant few eat relatively well. This last phenomenon need not be interpreted as a conscious suicide for the good of the tribe. It is more likely to be a byproduct of the usually 'successful' strategy of 'wait and see': better to wait for the dominant's leavings and hope for a return match later on than risk a real fight now. Conversely, it is often better for the dominants (those successful in the contests that define the eating and mating orders) to allow their subordinates lives of their own, sometimes even to assist them, and not to press home their attacks lest their victim turn upon them with the courage of despair.

The rules of 'war' were among the first to be noticed by ethologists: witness the 'merciful wolf, who spares his defeated rival when that rival rolls over and pretends to be a cub again. The assumption made by Konrad Lorenz, one of the few remaining ethologists not to be obsessed by the need to avoid anthropomorphism, that the wolf is 'inhibited' from killing his rival where a human victor would not be is questionable on at least two counts. First, there is good reason to think that human beings are very well able to kill each other but usually do not. There is no clear evidence that their record is much worse than the wolves'. Secondly, Lorenz gives no reason to describe the case as one of'inhibition'. The wolf does not (generally) kill — but that is evidence that he does not wish to, and perhaps never did wish to. He wishes to establish his superior rank, and that he does. We do not need to suppose that the wolf wants to kill his rival but disapproves of his wanting to do so and accordingly refrains. It is probably better to suppose that any desire he has to kill simply evaporates when his defeated rival resigns the contest. This sort of character is one that we could sensibly commend, but it is not quite as much like human moral virtue as Lorenz implies.

Other forms of animal behaviour might be compared with moral virtue more convincingly. It turns out that incest, or at any rate inbreeding, is much less frequent than we would expect if animals behaved like Johnson's savages. Female chimpanzees resist the advances of their brothers and of any other too familiar males. When Lucy Temerlin, a chimpanzee reared with humans and without experience of her conspecifics, reached puberty she rejected the attentions of her human foster-brother and foster-father while avidly pursuing any other human males. Here the reports do suggest much more strongly that Lucy experienced considerable conflict between her desire and her aversion, a conflict that might plausibly be compared with those of moralizing humanity. In this case an animal was inhibited from doing what she perhaps half-wanted to do. In another, reported by Jane van Lawick-Goodall from her observations of wild chimpanzees in the Gombe National Park, most of the chimpanzees ignored or bullied a companion who was partly paralysed. One chimpanzee, though disliking the smell (and one can reasonably assume) as much 'turned off by physical weakness and abnormality as his companions (and most humans), did none the less continue to treat the unfortunate and lonely ape as his old friend and companion. The disposition to friendship was stronger in him than the impulse to despise. This character too we would commend, even if we doubted that the altruistic chimpanzee thought he was 'doing his duty' - which some moralists have thought is the most important motive for moral action. He preferred one way of acting to another that he might have preferred.

Moral conscience, as we understand it, reflects on the actions and emotions of others as well as on one's own. To disapprove of oneself is also to disapprove of others similarly placed. Indeed, there is some reason to think that I come to know myself by knowing others who know me; I come to disapprove of myself by knowing, and approving of, others who might. So it is crucial to the existence of conscience (and self-consciousness in general) that there be social groups whose members attempt to regulate each other's behaviour. To do this they must be able to re-identify each other; they must have some grasp of past history; each must attend not merely to what other creatures are doing to him or her but also to what they are doing to each other. Such behaviour is not always found where we might expect it. The cannibalistic pair of chimpanzees discovered in the Gombe, for example, were (understandably) feared and resented by their fellows, but there has apparently been no attempt to ostracize or punish them. Other breaches of tribal discipline perhaps earn greater disapproval, notably (in some species) the attempt to dispossess an established nesting couple (if that is anything more than the mobbing of a supposed predator) or a failure of parental duty. If vervet females can respond to their own cub's cry, and also look (pointedly) at the mother of another crying cub, there is reason to think that they are operating in such a social system and that the roots of conscience are there.

By way of brief conclusion: to be a 'good dog' is to have those virtues of character that must be fairly widespread in a natural population if creatures of that kind are to survive and reproduce. A good dog is discriminating in her choice of mate, faithful to her cubs, prepared to spare her rivals and to accept her place in the social hierarchy of her group with good grace. Those animals that are of a kind that can be expected to identify others as individuals, and to reflect upon their own actions towards those individuals, may show some signs of having preferred the paths of virtue to those of easy gratification. Human animals alone, so far as we can see, have taken the next step, that of trying to assess their own sentiments in the light of reason. When they do, they are easily persuaded that they must not live 'like animals', out of immediate desire or fear. It would perhaps be better to remember that animals themselves do not live 'like animals'. Good animals of any kind (including the human) have some grasp of the physical and social worlds in which they live and prefer the paths of friendship and fidelity to those of war.