The claim that “morality is subjective” might be interpreted, not as an endorsement of some kind of moral relativism (as considered previously) but, instead, as an endorsement of some kind of moral anti-realism: the view on which there simply are not moral truths at all (stance-independent or otherwise). Broadly speaking, there are two varieties of moral anti-realism: (i) one on which moral judgments are not truth-apt in virtue of not expressing propositions, and (ii) one on which moral judgments do express propositions, but only ones that are false. Call the first variety non-cognitivism and the second metaethical nihilism. If one of these two positions is true, then we might think that morality is entirely subjective in the sense that there is no subject matter of morality about which we could possibly be right or wrong.
Non-cognitivists believe that when we make moral judgments, what we are really doing is just expressing attitudes of various sorts - so that when I say “murder is wrong”, for example, what I am expressing is merely my personal dislike of murder. Attitudes cannot be true or false: if I say that I enjoy the taste of chocolate or love watching Roger Federer play tennis, it would not make any sense for someone to say “false”, by way of response. The great 18th century Scottish philosopher David Hume is often identified as an early and important defender of non-cognitivism, and the view was, for various reasons, widely endorsed in analytic philosophy throughout much of the 20th century.
Two perennial problems for non-cognitivism are that (i) it seems to offer a poor account of what we take ourselves to be doing when we engage in moral discourse, and (ii) it cannot make sense of the logic of moral judgments. Regarding (i): most of us take ourselves to be doing more than merely expressing attitudes when we render moral judgments. When I say “murder is wrong”, I do not take myself to be merely expressing my personal disapproval of murder. Rather, I take myself to be expressing my belief that there is a universal reason not to commit murder. Regarding (ii): it seems possible to hold logically inconsistent moral beliefs. For example, if I believe that it is wrong tell lies, then it seems I am logically committed to believing that it is wrong to tell lies in a given case. But attitudes do not admit of logical appraisal in this sort of way: I can dislike tomatoes but like ketchup without being logically inconsistent.
Error theorists, on the other hand, believe that, although truth-apt, no moral judgments are true. A form of error theory seemingly common in today’s world involves (a) believing that there could be moral truths only if God existed to issue commandments, and (b) believing that God does not, in fact, exist. On such a view, “murder is wrong” is a truth-apt proposition just as it seems to many of us, but the condition needed to ground its truth simply is not in place. The important 20th century philosopher J.L. Mackie held that the only alternative to the kind of divine command ethic is some sort of moral Platonism (i.e., a view of morality along the lines of the one held by the ancient Greek philosopher, Plato), according to which moral values are objective and non-natural entities that possess strange properties (e.g., the property of being able to inspire motivation to act in various ways). The strangeness of these properties, Mackie believed, is itself reason to disbelieve in their existence.
One line to take against the error-theorists is, of course, simply to argue that we have reason to believe that the entities – God, or non-natural values – needed to ground moral truths really do in fact exist. Another is to argue that what Mackie perceives as the ontological strangeness of objective moral values does not count as any reason at all for thinking they do not exist: reality, we might insist, turns out to be quite strange (just read the things that quantum physicists say!); our question should not be whether some possibility is strange, but rather whether it is integral to our best account of the way things are. The anti-realist might at this point be tempted to say that presupposing a need for moral explanations is to beg the question in favor of moral realism. But, of course, the tu quoque writes itself: it can be alleged that in presupposing that there is no need for moral explanations, the anti-realist has begged the question. To break the tie, the moral realist can appeal to our deep psychological commitment to the idea that there really are moral truths. That, it might be thought, provides some evidence of an actual moral domain. Finally, one might challenge Mackie’s assumption that moral values would have to be ontologically strange. Many contemporary moral realists (e.g., W.D. Ross, Derek Parfit, Tim Scanlon, Kristine Korsgaard, Philipa Foot, Peter Railton, and many others) have attempted to defend varieties of moral realism that might be though less ontologically strange than those that Mackie had in mind.
In conclusion, my view is that, while anti-realists offer challenges that need to be taken seriously, it is far from obvious that any of these challenges are so strong that they should lead us to abandon the ordinary, commonsense commitment to the existence of a real moral domain.