Essentially, Rand provides a two part answer to this question. Concepts are classified (1) by their distinguishing characteristic(s), with their "measurements omitted"; and (2) concepts are classified in terms of "essential characteristics," which renders them cognitively efficient. In my last two posts, I refuted the first part of this answer. In this post, I will examine the second part.
In the chapter "The Cognitive Role of Concepts," Rand wrote:
It is the principle of unit-economy [i.e., cognitive efficiency] that necessitates the definition of concepts in terms of essential characteristics.
Typical for Rand, she here introduces a genuine insight, only to muddle it with one of the worst elements of her philosophy. She is correct to note the importance of cognitive efficiency. Where she goes astray is when she suggests that cognitive efficiency is a product of "proper" definition (and, ipso facto, of "proper" concept formation). In short, she has reversed cause and effect. Classification, she claims, arises from proper concept formation (which involves defining concepts in terms of essential characteristics). It's a conscious process. The individual must focus on the task at hand, and carefully guide the formation of what will eventually become "automatized":
Learning to speak is a process of automatizing the use ... of concepts. And more: all learning involves a process of automatizing, i.e., of first acquiring knowledge by fully conscious, focused attention and observation, then of establishing mental connections which make knowledge automatic. [IOTE, 65]
Why did Rand insist that concept making must be a fully conscious process? In the vast majority of instances, it clearly is no such thing. Many concepts are learned very early on, in the earliest years of childhood, well before schooling in logic or advanced thinking. Toddlers are not known for being especially careful or deliberate thinkers; they don't even have a particularly strong sense of reality. They often believe in Santa Claus and the Tooth Fairy. Rand's theory, on the face of it, seems grossly implausible. So why did she persist in it?
The most plausible explanation is that Rand wanted to make individuals solely (or at least largely) responsible for their development of their own "conceptual knowledge." The most important doctrine of Rand's epistemology derives from Rand's myth of the self-made individual, forming his knowledge and character on his own initiative, with little if any assistance from outward sources. If we stick closely to the logical implications of Rand's doctrine, we find ourselves confronted by the strange notion that every human being must recreate concepts, all on his own, from scratch. To be sure, Rand admitted that adults "could lend a helping hand." But given that "reason" can only be utilized by an individual mind, each individual has to do the lion's share of the cognitive leg work for himself.
It might be argued that Rand didn't mean any such thing. Admittedly, the doctrine she advanced is a bit vague. However, the implications are there in the doctrine for all to see. "There are many different ways in which children learn words," she explained. "Some proceed .. by treating words as concepts, by requiring a first-hand understanding of the exact meaning of every word they learn.... Some proceed by the road of approximations.... Some switch from cognition to imitation, substituting memorizing for understanding." [20-21]
Here Rand seems to be acknowledging that an individual could learn (or at least memorize) concepts without creating them from scratch. But she immediately condemns such an approach, calling it a "parrot's psycho-epistemology." People are merely learning the words, she insists, not the concepts underlying them. Is this really true? Remember, concepts are merely meanings; words are auditory symbols (as even Rand recognized) expressing those meanings. When children learns words, they ipso facto learn meanings (and therefore concepts). Are we really supposed to believe that some children use words without trying to express some sort of meaning? that when they speak they are merely uttering meaningless sounds?
These absurdities arise from regarding concepts (i.e. meanings) as knowledge. As I have repeatedly argued throughout this series on the Objectivist Epistemology, concepts are not knowledge. When concepts are regarded as knowledge, this implies that any concept without a real referent, such as unicorn or honest politician, is little more than meaningless gibberish and/or an inarticulate sound.
More important in the context of this post is Rand's implication that the each individual must form his own concepts. This view is almost certainly false. The words (and hence the meanings/concepts) that children learn are ready-made for them. Children learn all (or at least nearly all) of their meanings from the words spoken around them. We know this is true because the rare child who is isolated from all human intercourse fails to learn how to speak. Language, therefore, is a social product. But this does not mean, as apologists for Rand might maliciously assume, that knowledge is social as well. Not in the least. It is Rand who has (unwittingly of course) adopted the social view of knowledge. For if knowledge, as Rand contends, really were "conceptual," it would be "social," rather than individualistic. Ironically, Rand, in choosing concepts as the principle unit of knowledge, has (unwittingly) adopted a social view of cognition.
The fact that concepts are largely social products, learned from exposure to language, provides us with a far more plausible solution to the problem of classification than the one provided by Rand. Articulated concepts arise in the give and take of human interaction, in which there is a sort of Darwinian process of selection which favors cognitively efficient concepts over cognitively inefficient ones. Thus articulated concepts, although the product of human thinking and action, are not the product of design or specific conscious intention. Nor does each human being have to work out his own set of concepts, as Rand's theory implies. He just makes use of the ones that have been formed long before he was born and proceeds to use them to make conjectures about the real world.
Rand's insistence that cognitive efficiency must be achieved through "fully conscious, focused attention and observation" leads her to some rather glaring absurdities. In IOTE, we find her condemning cognitively unjustified concepts:
In the process of determining conceptual classification, neither the essential similarities nor the essential differences among existents may be ignored, evaded or omitted once they have been observed. Just as the requirements of cognition forbid the arbitrary subdivision of concepts, so they forbid the arbitrary integration of concpets into a wider concept by means of obliterating their essential differences --- which is an error (or falsification) proceeding from definitions by non-essentials....
Cognitively, such an attempt would produce nothing but a bad hash of equivocations, shoddy metaphors and unacknowledged "stolen" concepts. Epistemologically, it would produce the atrophy of the capacity to discriminate, and the panic of facing an immense, undifferentiated chaos of unintelligible data -- which means: the retrogression of an adult mind to the perceptual level of awareness. [71-72]
Is it really true that defining by non-essentials leads to a full state of panic? If so, would it be too much to ask for some documented examples? Curiously, Rand provides the following two examples of cognitively inefficiency: (1) forming a concept to designate beautiful blondes with blue eyes, 5'5" tall and 24 years old; and (2) regarding running as the essential characteristic of man. The problem with these examples is that they are completely fictitious. Who has ever formed a concept designating 24 year old blondes of a certain type? And who has ever claimed that running is the essential characteristic of man? We just don't find in practical experience examples of gross cognitive efficiency in the use of concepts. The meaning of terms used in common speech have been refined through centuries of trial and error. They are, in fact, far more efficient than concepts formulated through conscious attention (e.g.., concepts found in abstruse philosophical disquisitions). Classification is a very complex process. No human being, and certainly no child, would be capable of classifying the data of sense in an efficient manner solely on his own cognitive resources. Rand has here adopted an implausible approach. The concepts/meanings people use tend to be efficient from the get-go. They don't require any special formulation process, guided by "reason." They simply require being adopted and used.
Concepts are efficient because they are social. But even if man's concepts were not efficient, this would not have the the dire consequences that Rand moralizes about. Inefficient concepts would merely constitute a waste of cognitive effort. They would not lead to equivocations, shoddy metaphors, stolen concepts, or sheer panic. Once again, Rand is guilty of over-dramatizing an epistemological issue. Inefficient concepts (which essentially means: inefficient classification) are like a poorly written book. Bad writing is painful to read, but that doesn't that necessarily make a tedious book untrue or full of errors. Cognitive efficiency has little if anything to do with truth. A cognitively inefficient concept is merely a meaning that is difficult to use and comprehend. It's comparable to a convoluted sentence or a tedious, long-winded paragraph. It is much preferable to say more with fewer words. That is what cognitive efficiency amounts to in the end: saying more with less. But one could just as easily say more that is false with fewer words, or less that is true with more. Cognitive efficiency, in short, while eminently desirable, is no guarantee of truth.