Since a word is a symbol for a concept, it has no meaning apart from the content of the concept it symbolizes. And since a concept is an integration of units, it has no content or meaning apart from its units. The meaning of a concept consists of the units — the existents — which it integrates, including all the characteristics of these units.
Observe that concepts mean existents, not arbitrarily selected portions of existents....
Metaphysically, an entity is: all of the things which it is. Each of its characteristics has the same metaphysical status: each constitutes a part of the entity's identity.
Epistemologically, all the characteristics of the entities subsumed under a concept are discovered by the same basic method: by observation of these entities. The initial similarities, on the basis of which certain concretes were isolated and conceptually integrated, were grasped by a process of observation; all subsequently discovered characteristics of these concretes are discovered by the same method (no matter how complex the inductive processes involved may become).
The fact that certain characteristics are, at a given time, unknown to man, does not indicate that these characteristics are excluded from the entity — or from the concept. A is A; existents are what they are, independent of the state of human knowledge; and a concept means the existents which it integrates. Thus, a concept subsumes and includes all the characteristics of its referents, known and not-yet-known. [IOTE, 98-99]
Rand's strange doctrine that "a concept means the existents which it integrates" leads her to adopt a quasi-Platonist view of concepts. Concepts are not seen as psychological processes occurring within an actual human brain, but as ideal forms which real concepts can approach, if never quite match. Peikoff complains about the Platonic "essence-accident" dichotomy, which he suggests is at the root of the ASD. But there is a dichotomy in Rand that is far more perplexing and gratuitous: the dichotomy between concepts as immaculate representations subsuming all the characteristics of a referent, and the actual concepts human beings hold within their minds. Apologists for Rand will undoubtedly cry foul over my description of the Objectivist theory. But I have merely drawn out what is clearly implied by the theory itself. "A concept subsumes and includes all the characteristics of its referents, known and not-yet-known," declares Peikoff (undoubtedly speaking for Rand). How can a concept existing in a human mind include not-yet-known information? The only way this theory makes any sense is to assume that a concept is a kind of ideal construct, which our minds are trying to approach. Actual concepts (i.e., concepts existing in human minds) must fall short of that ideal construct, because human beings, as even Rand admits, are fallible. On this view of concepts, the aim of knowledge would be to approach that ideal concept as much as is humanly possible.
I suspect no Objectivist would ever put Rand's theory of meaning in quite these terms. They might instead insist on the phrase "known and not-yet-known." A concept, they might argue, is merely what is potentially known; it does not assume infallibility. This take on the theory, however, merely introduces even more troublesome implications. Not-yet-known is every bit as impossible a standard as all the characteristics of a referent. For how does one know whether one knows all the characteristics, or that one has attained the limits of the potentially knowable? That is a receding horizon, impossible ever to reach. On either interpretation, a dichotomy is created between Rand's ideal concept and the concepts that exist in human minds.
While Rand's view of concepts verges toward a non-naturalistic Platonism in one direction, in another it embraces a strange kind of quasi-posivitism. Concepts, Rand declares, mean the existents which they integrate. This suggests that concepts must mean only things that exist, never something that might not exist, like centaurs or laissez-faire capitalism. This doctrine seems little more than exaggeration and gross caricature of the very worst sort of positivism. It is odd to find so curious and hideous a growth emerging from what is supposed to be a stringently anti-positivist philosophy.
Rand could have avoided both the quasi-Platonism and the quasi-positivism in her theory of meaning if she had merely accepted the obvious fact that concepts are idealized descriptions which can be used to describe both fact or fiction, both truth and lies. Rand recognized that words are symbols; she failed to realize, however, that concepts, percepts, and, for that matter, all mental datum are symbols as well. Knowledge is merely symbols mediated by conjecture. Knowledge doesn't just name or identify things; on the contrary, it makes assertions about what it names and identifies. Only when concepts are combined into propositions can they become conjectures about reality which can be tested, either scientifically or in the practice of everyday life, and either falsified or corroborated.
Rand's theory of meaning, if it were ever consistently applied, would make the attainment of knowledge very difficult, if not impossible. For it would disparage many of the essential tools of trial and error, such as hypothesis and counter-factuals, because they depend on "invalid" concepts (i.e., concepts that aren't integrated from observed existents). Luckily for Rand and her Objectivist followers, her theory of meaning was never meant to be applied in real life. It's mere window dressing, a rationalization devised to beat down epistemological theories Rand and her disciples disagreed with. If taken serious and applied consistently, Rand's theory collapses into palpable absurdities.