Naming an object, a process, or an attribute involves little if any knowledge. Take the word poison. This term can be defined as any chemical substance that injures, impairs, or kills an organism. Note that this definition doesn't actually specify what chemical substances are in fact poisons or whether such substances exist. It merely states that if a chemical substance injures, impairs, or kills, then we will call it a "poison."
Does this definition of poison provide any non-trivial information about matters of fact? No, it does not. It is quite possible to know the definition of poison and yet know nothing of any specific poison. The definition of poison merely provides a naming convention. If you come across a substance that harms or kills an organism, it's "poison." But a naming convention is not knowledge. Knowing what to call things is different from knowing about things.
What are called "analytical truths" are merely deductions from naming conventions. This is why analytical truths are considered empty and tautological. They provide little if any information about matters of fact. They merely explain how various objects of thought are labeled. Hence the bad repute which definitional arguments have. Matters of fact cannot be determined on the basis of naming conventions.
In his essay on the Analytic Synthetic Dichotomy, Peikoff writes,
Denying that concepts have an objective basis in the facts of reality, nominalists declare that the source of concepts is a subjective human decision: men arbitrarily select certain characteristics to serve as the basis (the "essentials") for a classification; thereafter, they agree to apply the same term to any concretes that happen to exhibit these "essentials." [IOTE, 96]
Peikoff has provided here a malicious and distorted version of the theory I limned earlier in this post. He begins by declaring that nominalists deny "that concepts have an objective basis in the facts of reality." I'm not aware of a single nominalist who has ever said such a thing. Neither Locke, Hume, or Berkeley ever make such a statement. Peikoff really needs to provide specific examples and citations if he wishes to be taken seriously. Peikoff next attributes to nominalists the belief that the "source of concepts is a subjective human decision." Again, who among the major nominalists believes such a thing? Nominalists merely believe that words are "subjective" human decisions. Peikoff is here guilty of confusing concepts with words and then applying his confusion to the views of nominalists. Next we find Peikoff trotting out Rand's favorite scare-word, "arbitrary." The nominalists, according to Peikoff, believe that essential characteristics are "arbitrarily selected." This suggests that such characteristics are picked at random, like tickets at a lottery. This is yet another malicious distortion. Nominalists usually talk about concepts being formed on the basis of "convenience." This is little different from Rand's own theory, which claims that concepts are formed on the basis of cognitive efficiency.
Peikoff proceeds with a masterful bit of malicious distortion, as follows:
[Nominalists] commonly advance the [analytic-synthetic] dichotomy as a self-contained primary, independent of any theory of concepts. Indeed, they usually insist that the issue of concept-formation --- since it is "empirical," not "logical" --- is outside the province of philosophy. (!) (Thus, they use the dichotomy to discredit in advance any inquiry into the issues on which the dichotomy itself depends.) [IOTE, 97]
Peikoff begins here with an entirely unsubstantiated, and indeed, not entirely coherent allegation: nominalists take the analytic-synthetic dichotomy as a "primary"! I suppose what Peikoff is suggesting is that nominalists don't provide reasons for accepting the analytic-synthetic dichotomy. That allegation is simply not true. The analytic-synthetic dichotomy, as I pointed out in an earlier post, is an attack on rationalistic speculation. It attempts to demonstrate the futility of determining matters of fact by analyzing the meanings of words. It constitutes a frontal assault on Rand's own methods used in rationalizing Objectivism. That is why Peikoff resents the dichotomy and tries to refute it.
Having accused nominalists of taking the ASD as a primary, Peikoff next does an about face and accuses nominalists of being too empirical. Since such criticism could easily make Peikoff look bad, he places the word empirical in quotes, as if to suggest that the nominalist appeal to facts is phony. After all, aren't these the same folks who believe in the ASD for no reason at all? Peikoff, having distorted and caricaturized nominalism in this fashion can now deliver the final blow: nominalists have the nerve to insist that the issue of concept-formation is "outside the province of philosophy"!
Here we find the primary motive for Rand's and Peikoff's hatred of the ASD encaspulated in a single sentence. If by "philosophy" we mean "the attempt to determine matters of fact on the basis of logical, rhetorical, and moral constructions," then the nominalists are guilty as charged. The nominalist believes that matters of fact should be determined empirically, presumably through "scientific" and/or "peer reviewed" research. Rand and Peikoff believe that concept-formation should be determined in much the same way that Aristotle determined facts about the physical universe: that is, through arm-chair speculation vaguely based on notions derived from "common sense." It didn't work for Aristotle; why should we expect it to work for Rand and Peikoff? Concept-formation is an issue dealing with matters of fact. It is far too complicated a subject to be determined via arm chair speculation based on "common sense" or introspection. It requires sophisticated research techniques combined with exhaustive criticism and peer review. To render an appeal to "philosophy" (i.e., rationalistic speculation) as a kind of badge of honor demonstrates just the sort of empirical irresponsibility that characterizes the Objectivist modus operendi.
Peikoff next trots out Rand's theory of concept formation as the answer to the analytic-synthetic dichotomy. Once again, Peikoff demonstrates his cluelessnes about the ASD. The origin of concepts are irrelevant to this issue. The ASD, to the extent that it has relevance, is primarily about drawing a distinction between assertions based on empirical observations and assertions based on the analysis of meaning. If we wish to communicate meanings or consciously think about them, it helps to name them. But the act of naming fails to provide non-trivial information about matters of fact.
We can see that this is so by returning to the example of poison. Knowing the definition of poison does not provide any knowledge about poisons; it merely states that any chemical substance we come across that harms, impairs, or kills organisms we will label as poison. Knowledge only arises when we discover, through empirical research, substances that correspond to our definition. The definition itself tells us nothing of reality, only what some attribute or substance, whether real or not, may be labeled.
Now an Objectivist could argue, in response to this, that the concept poison could only be formed after some individual had discovered, through observation and experimentation, the actual substance. Even if this were true, it would be irrelevant. If the concept of poison could only be developed after an acquaintance of an actual chemical substance that harms, impairs, and/or kills, that is of little consequence: the definition of poison still remains a naming convention. Knowledge of matters of fact does not arise from meaning; it arises only when we assume that a given meaning describes an actual matter of fact. The fact that a given meaning was originally inspired by empirical observation does not in any way reflect on the potential validity or usefulness of the ASD. Meanings are neither true or false, valid or invalid, they just are. Rand's favorite tautology, A is A, applies, not to facts, but to meanings. The identification of matters of fact is a much more complicated matter which is best represented by the phrase A exists or A is. (And the identification of attributes involves predication, i.e, A is B.) Once a meaning is formed, it can be used to describe both fact and fiction, truth and error.
Further evidence for this point of view can be derived from the fact that some concepts really do begin their careers prior to experience. Robot, for example, is a term first introduced by the Czech playwright Karel Čapek, who used it to describe creatures produced by a factory in his play R.U.R. (Rossum's Universal Robots). But the meaning or concept of robots antedates the term and originates in ancient mythology. Atom is an example of another term that began prior to any experience or evidence of atoms. When scientific evidence for the existence of atoms began to emerge, the meaning of the term had to be revised. But that often happens with meanings. If a person comes across something new in the factual world, he will often hunt for some familiar meaning to describe it. This is the primary reason why words have multiple related meanings. Often when confronted with something new, people don't invent a new meaning to describe it; instead they adapt an older meaning to fit the new discovery.
By implication, Objectivism assumes that matters of fact can only be determined if concepts are originally based on facts. This theory involves a complete confusion of the role of concepts in cognition. Concepts are items of description. They are like the various colors of paint plastered on a canvas. By itself, a smudge of paint tells us nothing about matters of fact. Only when the paint is organized into a picture, can it represent an object in reality. Meanings act the same way. By themselves, they are cognitively inert. Only when organized into propositions can they represent matters of fact.