G.E. Moore, answered from the texts and cited to the page.
Goodness is a simple, indefinable, non-natural property. The naturalistic fallacy is committed by anyone who tries to identify good with a natural property — with pleasure, with happiness, with what we desire, with evolutionary fitness, with what some community approves, with what God commands, or with any other property that could be specified in non-evaluative terms.1
The argument is the open-question argument. Whatever natural property N is proposed as the definition of good, the question "is N really good?" remains open and substantive in a way that would be impossible if "good" just meant N.2 If "good" meant "pleasure," the question "is pleasure good?" would be like asking "is pleasure pleasure?" — trivially affirmative, no substantive answer possible.
But "is pleasure good?" is not trivial; it is the central question hedonism asks. The same argument generalizes to any natural property. Goodness is therefore not identical with any natural property, and the attempt to define it in naturalistic terms commits the fallacy. Goodness is a simple property: it has no parts, it cannot be analyzed into other properties, it is grasped directly.3
It is non-natural: it is not the kind of property that could be observed, measured, or reduced to scientific terms. It is indefinable: any attempted definition either misses the property (by giving us a natural property) or fails to be a definition (by making the supposed definition trivially true).4
And there is, moreover, a further reason for treating Hedonism separately. That doctrine has, I think, as plainly as any other, owed its prevalence to the naturalistic fallacy; but it has had a singular fate in that the writer, who first clearly exposed the fallacy of the naturalistic arguments by which it had been attempted to prove that pleasure was the sole good, has maintained that nevertheless it is the…Principia Ethica (1922 (first ed. 1903))
The inference seems very natural; and yet what is meant by it is self-contradictory. For those who make it fail to perceive that their conclusion 'what possesses this property is good' is a significant proposition: that it does not mean either 'what possesses this property, possesses this property' or 'the word "good" denotes that a thing possesses this property.' And yet, if it does not mean one or other of these…Principia Ethica (1922 (first ed. 1903))
For I cannot hope to recommend the doctrine that things which are good do not owe their goodness to their common possession of any other property, without a criticism of the main doctrines, opposed to this, whose power to recommend themselves is proved by their wide prevalence. =25.= The theories I propose to discuss may be conveniently divided into two groups.Principia Ethica (1922 (first ed. 1903))
If I were to imagine that when I said 'I am pleased,' I meant that I was exactly the same thing as 'pleased,' I should not indeed call that a naturalistic fallacy, although it would be the same fallacy as I have called naturalistic with reference to Ethics. The reason of this is obvious enough.Principia Ethica (1922 (first ed. 1903))
Principia Ethica's argument against naturalism is one of the central twentieth-century positions in metaethics. The open- question argument has been defended, attacked, refined, and reinterpreted across the century. Frankena's 'The Naturalistic Fallacy' (1939) argued that what Moore identified is not strictly a fallacy. Pigden's contemporary work argues the open-question argument is question-begging. The Cornell realists (Boyd, Brink, Sturgeon) and Foot's neo-Aristotelian naturalism (Natural Goodness 2001) revive forms of naturalism Moore's argument was directed at. The expressivist tradition (Blackburn, Gibbard) takes Moore's argument as motivation for non-cognitivism rather than for non-naturalist realism. Contemporary non-naturalists (Parfit, Enoch, Shafer-Landau) defend the Moorean line in modified form. The position remains contested at the center of metaethics.