top of page

"Was it Always True that Slavery was Wrong?": An Essay by Catherine Wilson

White house on hill

From The Philosopher, vol. 108, no. 3 ("Bodies").

If you enjoy reading this, please consider becoming a patron or making a small donation.

We are unfunded and your support is greatly appreciated.

Many of the habitual practices of antiquity, and even of the recent past, as well as the practices of distant cultures pose a philosophical problem for us.

By “us”, I mean – in the present context – philosophically interested readers of this journal. On the one hand, most of us quickly, intuitively and spontaneously judge slavery to be “wrong”, along with infanticide, torture, and gladiatorial combat, and to have been so in ancient times too. On the other hand, we seem to lack a sound justification for thinking that not only do we have feelings of sadness and moral offense in reading or learning about these ancient practices but that our judgements of wrongness are correct. Some people, finally, don’t have such feelings in the first place. They may even think something like: “well, for them, it made sense, and who are we to judge them anyway?”

To get clear about the institution under discussion, here are a few details about ancient slavery. Slaves in the ancient world performed the bulk of the labour in agriculture, mining, construction, and factory work (pottery, textiles). They were acquired in a variety of ways: through slave-hunting raids; as war-captives; through agreements by the destitute; through the collection of children abandoned by their parents; or through sale or inheritance as part of an estate. Greek and Roman law treated slaves as property that, as well as being bought and sold, could be beaten, killed, worked to death, or used for sexual purposes at will by their owners.

Like many other practices, slavery enjoyed both metaphysical and practical justifications. It was natural, Aristotle thought, because the universe itself was hierarchically arranged (Aristotle’s defence of slavery is found in his Politics Bk. I). Ancient philosophers, including Plato, were very interested in functions and ratings. It seemed obvious that, just as some teeth are good for grinding, others for tearing, some dogs for guarding, others for hunting, some horses for racing, others for ploughing, some humans were suited to leisure and intellectual activities, others to manual labour. Slavery was unavoidable because how else was work to get done? The fields don’t plant themselves as Aristotle joked, and who goes down the mines, he might have added, in the absence of the whip?

Probably not all ancient slaves had terrible lives: musicians, tutors and scribes were often slaves, and despite the legal differences in their rights, the actual lives of some slaves were probably indistinguishable from those of today’s agricultural or textile workers. Yet from our perspective, these practices were based in ideological illusions that preserved the position of the privileged (natural hierarchies); in reprehensible tendencies (sadism, love of domination, indifference to others’ preferences and feelings); and in groundless fears (the end of art, science and culture if a leisure class could not be supported by slave labour).

We know, however, that some of “our” individual and typical group behaviour is regarded with strong disapproval by people in other cultures and subcultures. Furthermore, future members of “our” group, if there still are any, will most likely be appalled by our prison systems and call centres, by the arms trade and atomic weaponry, identifying exactly the same features of ideology, indifference, and groundless fear as supporting them. They will wonder how “we” – meaning people alive at the time – tolerated these institutions for so long.


The apparently simplest epistemic position to take with respect to moral judgement is accordingly “Dogmatic Realism” (a position defended by Russ Shafer-Landau in Moral Realism: A Defence). According to the Dogmatic Realist, some practices and institutions are objectively permissible, others objectively forbidden or obligatory, and we are clear about the moral status of some but believe falsely about others. The ancients were breaking the moral law, as the ancient opponents of slavery knew, and as “we” know. So slavery was always wrong, although many people failed at the time to realise this. The sentence “Slavery is wrong” was always true, and is now known to be true.

Dogmatic Realism faces a host of objections: Here are just five:

1) How can we be sure that “we” know that slavery is wrong when our forebears who were philosophically reflective and had the same basic emotional dispositions as we do believed it to be acceptable? They had their set of justifications; we have a different set.

2) How can we be sure we know that slavery was wrong when we know, through induction, that moral opinions change in unforeseen ways?

3) The idea of “timeless truth” is puzzling. Was it true at the time of the big bang that slavery was always wrong? Or did it become true when the first person became a slave? Or when a word translatable as “slave” came into use? Or when slavery was legally defined? Sentences – marks on paper or another surface that are interpreted and understood by users of a language – are not true or false, only statements, made on a particular occasion by means of an uttered sentence, that express someone’s belief.

4) According to some philosophers, the concept of knowledge is prior to the concept of truth. If they are right, if no one knows that P, it can’t be true that P. Alternatively, if knowledge is justified true belief, how could the few opponents of slavery in the ancient world have been justified in believing that slavery was wrong before anti-slavery arguments had faced off against pro-slavery arguments and won?

5) The statement, expressing a belief, “Slavery was always wrong” has no possible use, so no meaning, so no truth value. By contrast, a statement like “The slave labour occurring today in diamond mines/T-shirt factories etc. is wrong” does have a possible use, so a meaning, and a truth value. It can be used to rouse opposition to current practices.

Although it would be better to have one irrefutable objection to a position rather than five contestable arguments against it, this is not how philosophy works. These five arguments against Dogmatic Realism, though individually contestable, can show what pitfalls are to be avoided in trying to frame a better account of the morality of past practices.

Faced with the weak points of Dogmatic Realism, it might be tempting to go over instead to what I will call “Anthropological Antirealism” (a position defended by Jesse Prinz in The Emotional Construction of Morals). On this view, moral beliefs and judgements are just anthropological data points, elements of a tribe’s – and “we” are just tribe members – mythology. No practices and institutions are objectively permissible, obligatory or forbidden. There are no “categorical imperatives”. All moral rules are “hypothetical imperatives” – the rules for operating in a society if you’d like things to run smoothly and avoid angering others or suffering retaliation, ostracism, gossip, and punishment. At the same time, because humans are psychologically prone to cast judgements and to experiencing strong feelings of guilt and self-righteousness, the delusion that moral rules are categorical is hard for most people to shake.

Because humans are psychologically prone to cast judgements and to experiencing strong feelings of guilt and self-righteousness, the delusion that moral rules are categorical is hard for most people to shake.

For the Anthropological Antirealist, all moral judgement is the expression of a preference, a liking or a disliking, an approval or disapproval. Claims like “Slavery was always wrong” as well as “The slave labour conditions obtaining today are morally unacceptable” have no truth value. Our tribe prefers that there not be slaves, so it was good luck that they were freed by legislation at some point in the past. Perhaps we also “wish” that those ancient people had had the same preferences and had freed their slaves, just as now we “wish” for an end to factory farming, but it is a pretty weak wish, and completely ineffectual. There is no way for the first wish, unlike the second, ever to come true.

The Anthropological Antirealist may concede that I can make disapproving judgements about past practices if there is a point to making them. By asserting “Slavery was always wrong” or even “It was always true that slavery is wrong”, I may emphasize my commitment to my tribe’s anti-slavery stance or my tribe’s commitment to the mythology of categorical imperatives such as: “Never be a slave holder!” or “Do not tolerate any slaveholding in this tribe!”

If I think those disapproving thoughts, says the Anthropological Antirealist, my stance is different from the position of those past people – that tribe – who thought that slavery was just, in the same way as my clothes are different from those of the generation of my own great-great grandmother – the tribe of ladies of the late 19th century. Even if I believe that it would have been better for the tribe of ancient people to have thought like me, or the tribe of ladies of the late 19th century to have dressed like me, I’m still expressing a culturally-located preference, either about what people should have thought or what they should have done, or both. And in most cases, this preference is idle: what point is there to thinking it was a shame that they had to wear whalebone corsets back then?

According to the Anthropological Antirealist, learning takes place only within a tribe. By a “tribe”, he or she is thinking of a group of people who acquire and share knowledge with one another by direct communication. Alternatively, however, we can regard any tribe as part of a larger human unit which has collectively, though not in all its parts, acquired knowledge or accomplished something, even when all the parts of the collective are not in direct communication. The extension of “we” is pragmatically rather than metaphysically demarcated.

For example, “we” have sent a man to the moon, but this does not imply that the inhabitants of the borough of Redbridge have sent a man to the moon, even though the inhabitants of the borough of Redbridge are a subset of human beings, as are the various tribal peoples of special interest to the anthropologist. And it is “known that” water is H2O, although this is not known by every human being, because not every human being has learned it. If moral learning, not just the learning of non-moral facts and practical techniques, is possible for an individual, it should be possible for any anthropologically selected tribe, and it should also be possible for a wider collective. “We” now know that slavery is wrong (although many individuals on earth cannot be said to believe it); alternatively, it is known that slavery is wrong. We as a species have learned it, through discussion, observation, responses to protest, and experimentation over the centuries.


The Anthropological Antirealist might decide to bite the bullet and deny that moral learning, as opposed to a succession of different states of belief and practice, actually takes place. The most radical version of this view would be the Humean position, revived by Galen Strawson in his 2004 article “Against Narrativity”, that “I” am simply a series of unconnected qualitatively different psychological states. I might think my beliefs and practices are better now than they used to be, but that is only to prefer carrying on as I do now to the way another person with some extrinsic features attached to my name did in the past.

There is an important insight in the Hume-Strawson position, in that acknowledging or denying continuity with a past self is again a pragmatic rather than a metaphysical issue. One might, for example, refuse to accept full responsibility for one’s seeming part in some unfortunate event on the grounds that one was acting under extraordinary pressures with limited perspective – or even that one has forgotten the whole thing, as happened to Locke’s drunkard in his discussion of personal identity in An Essay Concerning Human Understanding. This refusal is analogous to the reasoning by which we modern people might insist that the practices of our ancestors have nothing to do with “us”, or that those people were acting under extraordinary pressures and with limited perspective.

Alternatively, one can adopt the penitent outlook of someone who has learned something and who acknowledges one’s own past beliefs as false and actions as wrong, or the outlook of someone who simply acknowledges a progression from better to worse, like the beginner who learns eventually to play the piano to a certain standard.

If there is to be learning and progress in knowledge, technology, or morality, then the unit that has learned or made progress has to be seen as endowed with a history.

If there is to be learning and progress in knowledge, technology, or morality, then the unit that has learned or made progress has to be seen as endowed with a history. I can therefore acknowledge the ancient slaveholders as members of a collective that includes them along with us modern people, and rightly claim that “we” have learned something in the intervening centuries. The Anthropological Antirealist is right to point out that including them in the larger collective is a choice: Just as I do not look down on the inhabitants of the borough of Redbridge for not having made any particular contribution to putting a man on the moon or regard members of Amazonian tribes that don’t know that water is H2O as ignorant, I could make the choice to regard the ancients as blameless in their slave practices. But the epistemic payoff in including them in the “we” category of people who have learned that slavery is wrong is that it invites us to ask: how exactly did we learn that? What was the contribution of philosophical arguments? Of religion? Of protest? Of novels and autobiographies? There can also be a moral payoff to acknowledging the ancients as part of “us” in terms of increased sensitivity to current situations in which we haven’t made much or any progress.

To say that belief-and-practice set B is “better than” belief-and-practice set A is to say that the transition of an individual or a group from A to B would constitute learning or progress, while the transition from B to A would constitute a step back or loss of knowledge. At the same time, because the intended extension of “we” reflects choices that are not metaphysically but pragmatically determined, there can sometimes be good reasons to keep the extension narrow, and thereby to preclude comparisons of betterness. Anthropological Antirealism is thus partially vindicated.


Consider, for example, the treatment of aged relatives in a traditional Asian society, where their care is always a family responsibility, as opposed to the treatment in a traditional English society where there are “senior communities” and “care homes” outside the family where many old people end up. Both systems have advantages and disadvantages, and both are supported by different features of their contexts. It isn’t clear that “we” – all of humanity – could come to learn that one system was better than the other and that all accompanying practices should be revised so as to enable a switchover to the better option. The case of slavery is different: there we think that far-reaching changes to legal codes and a reorganisation of labour practices ought to take place even if they are disagreeable and even impoverishing to some.

If you prefer things to be the way they were before, and if you continue to act according to those outdated pre-MeToo standards, you will find yourself on the wrong side of history and subject to censure.

The same might be said for the contrast between a “puritanical” society that proscribes sex for both males and females until they are officially paired off, and a “liberated” society that exercises no controls, either internal or external, from a certain age. Both systems have their advantages and disadvantages, and it is hard to see how the transition from one to the other could be made on the basis of learning, or how one system could be considered objectively better than the other. Groups – and individuals – will just prefer one way of doing things to the other. By contrast, the “MeToo” movement reflected real learning on the part of an extensive – though not universal – “us”. If you prefer things to be the way they were before, and if you continue to act according to those outdated pre-MeToo standards, you will find yourself on the wrong side of history and subject to censure, like any modern day advocate of slavery amongst “us” who might try to get away with enslaving people or defending slavery as an institution.

The flexibility of the extension of “we” can explain and clarify the results of several of the striking thought experiments and the interpretations thereof advanced on behalf of relativism. To take two prominent examples:

1) In Ethics and the Limits of Philosophy, Bernard Williams argued that the statement “The Aztecs were wrong to kill and eat the hearts of their captives” is not false but meaningless.

2) In The Nature of Morality, Gilbert Harman argued that the claim that “Hitler was wrong to order the extermination of the Jews” was false on the grounds that we and Hitler do not share a common moral framework. Harman considered Hitler to be “outside our morality”. Within Hitler’s framework, Harman implied, his action was morally correct.

Williams’ assessment can be understood in this way: There is no historical connection between “us” – readers of this magazine – and the late medieval Aztecs. There is no candidate here for a group that could have learned something together – namely not to eat the hearts of one’s captives. It is hard to conjure up a context in which condemning the Aztec practice would be a contribution to an ongoing discussion of moral practices.

Harman’s argument that the claim that “Hitler was wrong to order the genocide” is false is not as sustainable. We can acknowledge a historical connection between “us” and past generations of persecutors and oppressors of Jews. Hitler and his enablers were members of “us” if we think broadly enough, and they did things that “we” have learned should never be done and therefore should not have been done. In declaring that Hitler was wrong, we condemn the practice and signal that there has been learning and progress within European society.

If you baulk at this embrace of responsibility, there is another way to endorse the judgement that Hitler was wrong. If, as individuals, Hitler and his enablers had changed their minds about perpetrating genocide on the basis of learning, that would have constituted moral progress on their parts. Their earlier beliefs would have been replaced by better ones.


So, was slavery “always” wrong? Was it always true that slavery was wrong? I’m at a loss how to interpret these sentences. Was water “always” H2O even before there was water? Was it “always true” that water was H2O even before there was water? Let’s just drop the references to “always”. Slavery in the ancient world must have been wrong. For the alternative judgements – that it was not wrong or that it was neither wrong nor not wrong – imply that we have learned nothing about the moral acceptability of slavery since ancient times. And that seems plainly false.

Catherine Wilson is a leading figure in the field of the history and philosophy of science. She is distinguished professor of philosophy at the CUNY Graduate Centre. Her latest book How to Be an Epicurean was published by Basic Books (and as The Pleasure Principle in the UK by HarperCollins).


From The Philosopher, vol. 108, no. 3 ("Bodies").

If you enjoyed reading this, please consider becoming a patron or making a small donation.

We are unfunded and your support is greatly appreciated.