Find Enlightenment

The Universality and Employment of Concepts

Bryan Register

Forum: The Journal of Ayn Rand Studies, Vol. 1, No. 2
Thought is surrounded by a halo. . . . We are under the illusion that what is peculiar, profound, essential, in our investigation, resides in its trying to grasp the incomparable essence of language. Ludwig Wittgenstein ([1945-49] 1958, para. 97)

1. Introduction

The purpose of this paper is to explore and revise Rand's theory of concepts. As I see it, Rand's theory is not exactly a theory of concepts at all. Rather, the theory provides a plausible account of two things:

1) The objectivity of categorizations of things. Traditional realists have held that categorizations of things can be objective because the members of a category all share some special ontological feature in common, typically known as a universal. All of the members of a category are related in some special way to the same universal, and no non-members are likewise related to that universal. The goal of our conceptual apparatus is to divide the world into categories such that we place into a single category all those things that share relations to a common universal.

Rand, however, would call this approach 'intrinsicist.' Intrinsicism is the classical account of objectivity wherein objectivity consists in the subject treating objects as they are in themselves, without making any contribution to their having certain features like category-membership, value-ladenness, and so forth. Thus, the intrinsicist account of categorizations is that category members are intrinsically in categories. Traditional realism accounts for this intrinsic category membership by postulating real universals and demanding that our categorizations line up one-to-one with these ineffable ontological entities. Rand would argue that traditional realism makes objectivity in conceptual apparatus an impossible goal. Contrary to the traditional realist, there are no universals, and, hence, there is never any way to make our categorizations line up with them. Further, realism, by setting these impossible standards, plays into the hands of skepticism about objectivity. Nominalist skeptics (those Rand might call 'subjectivists'), arguing that there is no way to meet realist standards of objectivity, will conclude that there are no appropriate standards of objectivity.

Rand is a nominalist about universals, but she retains realism's stress on the importance of epistemic norms and objectivity—a synthesis that constitutes her 'Objectivist' solution to the classical problem. Rand's alternative 'Objectivist' account views objectivity as a complex causal interaction between a subject and its nature and needs, and an object's nature and capacities. She holds that category members are not intrinsically members of the same category but rather that they are members of the same category in virtue of their relation to a knowing subject, the cognitive needs of that subject, and their own natures. Her theory accounts for category-membership in a way that, like traditional nominalism, does not multiply ontological entities beyond necessity but, like traditional realism, allows for the objectivity of our categories (and insists on such objectivity as a normative matter). [1]

2) The universality and abstractness of certain mental entities (states, processes, abilities—somethings) known as concepts. Rand's theory accounts for the categories of things in the world by invoking the cognitive needs of a subject and the subject's mental acts that divide the world into categories. But likewise, having created the divisions, the subject is then in a position to treat members of different categories differentially and appropriately.

Let me explain this important terminology. When I say that a subject can deal with things differentially, I mean that she can deal with these things in a way that respects their difference(s) from other things. When I say that a subject can deal with things appropriately, I mean that she can deal with these things in a way that respects their own common nature. Not trying to read a pair of scissors, for instance, is treating scissors differentially from written text, while using the scissors to cut paper is using scissors appropriately given their nature, which they share with other pairs of scissors.

Objectivist epistemologist David Kelley (1984) treats universality and abstractness as essential features of concepts. No Objectivist writer has, to the best of my knowledge, treated any other mental entity (or whatever) as possessing either of these traits. Objectivist usage, then, seems to indicate that universality and abstractness are partially definitive or constitutive of concepts and nothing else. These features are important for our present purposes, because it seems to be the universality and abstractness of some capacity that accounts for our differential and appropriate handling-abilities. Any ability to deal with an open-ended category (or at least a category with multiple members) of things (properties, events, relationships), which are non-essentially different from one another, will appeal to universality and abstractness. Universality is the multiple applicability of the concept to all of the members of a (potentially open-ended) category. Abstractness explains how these category members can be non-essentially different, that is, different from one another within some range. Any capacity to deal with category-members in a way that is appropriate to the members of that category will rest on the members of that category being, in some respect, essentially the same as (or, not essentially different from) one another. Moreover, for such a capacity to deal with any member of the category, it must be a capacity to deal with them just insofar as they are members of the category, so the capacity must be universal (or at least multiple) in scope.

Universality and abstractness are features of and only of concepts; these features allow the subject to deal with things differentially and appropriately, according to their membership in specific categories. Any ability, then, that involves the differential and appropriate handling of things will rest on a concept that has some special relation to those things. Colloquially, we can say that this is the concept 'for' those things.

I will argue that, if Rand's theory is true, then it may explain the abstractness and universality of our concepts, and the objectivity of the categories of things which concepts are 'for.' However, a great deal goes unexplained. In this paper, I will examine the nature of concepts, their relationships with words, and the inadequacies of Rand's theory. I shall argue that there are many abilities to handle category members differentially and appropriately, and that these abilities thus require something in the mental background that is universal and abstract in nature. Rand, however, seems to focus exclusively on linguistic abilities (and only one aspect of those!), leaving these other capacities unexplained and even unexplainable. Subtly dealing only with the theory side of the theory/practice distinction, and, less subtly, confusing concepts with words, Rand deals with concepts in a way that fails to account for their involvement with most human activity.

I will propose a two-step way to fill in the gap. The first step will be to incorporate H. H. Price's account of concepts as recognitional capacities into Objectivist theory. The second will be to revise Rand's notion of concepts of method. With these additions and alterations, Rand's theory should be able to explain all of our differential and appropriate abilities. [2]

2. Price on Concepts

I want to start by discussing Price's theory of concepts, and move, in the next section, to compare his theory of abstraction with Rand's. The comparison between Price's and Rand's theories will bring out the most important features of the process of abstraction for our purposes.

For Price (1969), the possession of a concept is marked by a number of capacities, including mastery of words, capacity to rephrase, a feeling of familiarity when given an instance of the concept, the ability to produce appropriate mental imagery, and so forth (334-55). But he suggests that one capacity, the recognitional capacity, is central:

the recognition of instances is the primary and fundamental way in which a concept is manifested, and all the other manifestations it has, or may come to have as it develops towards completeness, are dependent upon this one. Fundamentally, a concept is a recognitional capacity, whatever else it may be besides. (355)

Price suggests three lines of argument. First, he proposes that recognition must necessarily precede abstraction (35). Especially in discussions of Rand, it is typical to use 'abstraction' to mean the process of concept-formation. But by 'abstraction,' Price seems to mean the entertainment of appropriate mental imagery, that is, imagining a concept-instance in the absence of a real concept-instance (in 'abstraction' from a real concept-instance). [3] So Price is not suggesting that recognition precedes the formation of concepts. He is suggesting that one conceptual ability, recognition, precedes another conceptual ability, the entertainment of mental imagery.

This is surely right, since if we could entertain mental images of concept-instances without being able to recognize concept-instances, we would not recognize the mental images as being (imaginary) concept-instances; thus, we would have no guide in entertaining some particular mental image rather than another. Exceptions, such as the concepts dragon [4] or ghost, are derivative concepts. Our capacity to mentally envision dragons and ghosts derives from our prior capacity to mentally envision instances of whatever concepts these concepts are derived from. For instance, our capacity to envision a dragon is derived from our capacity to envision (say) lizards, snakes, wings, fire, flying, and so forth. But the capacity to mentally envision these things is based on our capacity to recognize them.

Second, Price points out that none of our practical skills could exist if we did not have the capacity to recognize instances of relevant concepts (36). One could not, for instance, play a violin without being able to recognize one. This argument is intended to establish the priority of recognitional over performative capacities, but I do not think it succeeds. Playing the violin is one kind of performance, which, as I shall argue, is concept-based. But some bodily tasks, which do not employ tools, are also performative capacities, just as is playing the violin, and these do not require the concept body for their performance. Learning to whistle, say, is a conceptual skill, but one need not have the concepts mouth, exhale, and so forth to whistle. Price's point establishes that very many, but not all, of our performative capacities must be based on a prior recognitional capacity. [5]

Finally and most importantly, Price argues that the recognitional capacity is prior to the possession of words in the sense of knowing which words refer to which things, and, moreover, that one can have a particular recognitional capacity without possessing a word which refers to the recognized things (35-37). He points out that most of our recognitional experiences do not involve using any word that refers to the recognized object. For instance, when we plan to enter a room through a door, we do not mentally vocalize 'door,' 'doorknob,' 'entrance' and the like when we recognize the door. Moreover, before we can perform such mental vocalization, we must already have recognized the thing, the word for which we then proceed to vocalize.

Consider especially the case of recognizing words. When we recognize a word as being the word it is, we have an immediate recognitional experience, which does not involve saying to ourselves: “This is the word X.” Such a requirement would make word-recognition impossible in an obvious way. If I say to you, “The cat is on the mat,” and you have to say, “This is the word 'the,'” to recognize the first word, you would then have to say, “This is the word 'this,'” to recognize the first word of the sentence by which you recognized the first word of the original sentence, and so forth.

Two more suggestions should clinch the case. The first is that pre-verbal children seem to recognize things as what they are without possessing the words for them; we know this because they often respond differentially and appropriately to things. Moreover, for a pre-verbal child ever to learn language, she would first have to notice that the sounds produced by other persons are somehow meaningfully correlated with things, and that the same sound typically goes with the same thing or a thing of the same kind. Now, this recognitional experience of 'same again,' both for the words and the things to which the words refer, is a conceptually based experience, so some concepts must be possessed before any language.

It would be possible to take the point too far. Not every concept is primarily a recognitional capacity. How does one recognize inflation, mental focus, or something's being fictitious? Perhaps not at all. Nevertheless, Price's arguments do establish that recognition is a basic conceptual ability, and that it must precede linguistic abilities for at least some of those concepts that are recognitional capacities.

So Price has given us two very good reasons to believe that recognition is a basic conceptual ability, which is certainly prior to language and the entertainment of visual imagery. So far, we can agree. But what about the process that gives rise to recognitional capacities?

3. Price and Rand on Abstraction

Here, Price's theory is inadequate. Price refers to realism about universals as 'The Philosophy of Universals' and to nominalism as 'The Philosophy of Resemblances.' After discussing the differences between these two, he concludes that they are simply different and equally problematic vocabularies for saying the same things (30-32). The problem with the vocabulary of realism is that it leads us to reify similarities into universals, which are theoretical, non-physical, non-experienced entities that cannot be shown to exist. But this reification is not an accidental side-feature of realism about universals; it is the essence of realist theories. So Price is understating the significance of his difference with realists. If we are not to believe in universals, then we have only the ontological commitments of the nominalist.

For the nominalist, it is similarities between things, rather than common relations to a universal, that justify our treating them as the same. If Price rejects universals, then he must provide a theory of how we can apprehend similarities in things that would justify our treatment of those things as the same. The Philosophy of Resemblances, or nominalism, must be that theory. Unfortunately, Price's version of nominalism leaves him liable to certain standard criticisms.

Emmet (1954), who directs it specifically at Price's account, provides one of these criticisms. The objection amounts to a puzzle about kinds of similarities. If something is an instance of the concept red, then it must be similar to other instances of the concept. Specifically, if we tell some subject that something is red if and only if it is similar to certain paradigm red objects in the same way that they are similar to one another, we have failed to take account of the many respects of similarity between the red objects. For instance, an apple, a tomato, and a globe of Mars are similar in their shape as well as their color. How can we focus attention on the specific respect of similarity we want our subject to attend to? Or, to put it developmentally, how is the attention of the concept-acquiring child focused on the proper respect of similarity (where 'proper' indicates no more than 'socially accepted' but may, normatively, indicate more)?

The other question is an epistemological version of Russell's proof of universals (see Russell [1912] 1988, 95-97 [ch. 9]). Russell sought to justify the introduction of universals as a theoretical ontological entity by raising a different puzzle about similarity. If there are no universals, then things are in categories only because they are similar to one another. But the similarities in question—for instance, all of the relationships of similarity in respect of color between all white patches—are themselves all the same thing, similarities. So nominalists don't get away from universals after all.

As Price rightly points out, this argument fails (1969, 23-26). The similarity between two white patches is not identical to the similarity between two glasses of water, because the one relation has different relata from the other. So the two similarities are themselves only similar to one another. There is no reason at all that there cannot be higher-level relations between relations without invoking universals.

One might argue that the '=' signs in a=b and c=d seem to indicate the same relationship between the flanking names, even though the relata of the relationships are different. Likewise, the similarity between two red things seems to be the same relationship, similarity, as the similarity between two blue things. Nevertheless, though the two identity relationships are both identity relationships, they are not the same identity relationship. Nor are they both identity relationships because of their respective associations with a universal for identity. They are both identity relationships because it best meets the cognitive needs of a human subject to treat these two relationships as indistinct at a certain level of abstraction. Likewise for the various similarity relationships. Further, it is this kind of relationship (and our cognitive need to attend to it) that justifies our treating different red things as the same in respect of color regardless of the fact that they have no common universal.

But Price fails to take account of an epistemic version of Russell's argument that is especially dangerous to his own account of concepts as recognitional capacities. This version would not prove that there is a universal for similarity, but rather that similarity is an innate concept—one not formed through any process of abstraction. Such an argument would go as follows:

In order to form the concept red, you must first detect the similarity relationship between various red things. But since concepts are recognitional capacities and are necessary for detecting such repeatable relationships as similarity, then to detect the similarity relationship between red things you must first possess the concept similarity. So, at least one concept, similarity, must be possessed before any concept can be formed.

The merit of Rand's theory of abstraction is that it dissolves both of these objections to nominalism. [6]

Rand's solution to the problems for nominalist theories of abstraction is to analyze similarity into levels of difference. The recognition of two things as similar relies on experiencing them as less different from each other than from some contrast object which they are both experienced as being more different from. This enables both problems to be solved.

The Russell problem can be dealt with as follows. An innate recognitional capacity for similarities is not necessary to detect similarities if similarities have been analyzed into levels of difference. The capacity to experience things as different is, indeed, innate, but it is not conceptual. The perceptual mode of experience is sufficient, [7] because this mode of experience distinguishes entities from their backgrounds; thus, entities are experienced as different from whatever constitutes their backgrounds, which will at least sometimes be other entities (and not, say, undifferentiated fields of color like walls or the sky). When our attention is called to two entities that are not as different from their mutual background (other entities) as they are from one another (considered as each other's background), then they are being experienced as [8] similar, but no recognition of similarity is required.

The presence of the background of other entities also yields up the respect of similarity necessary for the solution of the Emmet problem. For two entities to be experienced as similar, they must be experienced as less different from one another than either of them are from some third entity. But the specific way that they are both similar to one another and different from the third entity will then be the respect in which they are similar. Thus, the contrast object brings the respect of similarity into salience.

We could recast this into Emmet's way of speaking about getting a subject to acquire a concept by telling her that the instances of a certain concept are all of the things which are similar to certain paradigms. In Emmet's argument, there are many respects in which the paradigms will be similar, and so we will not be able to teach the concept simply by referring to 'the' similarity between the paradigms. However, if we try to teach a concept by telling the subject that the instances of the concept are all of the things that are similar to certain paradigms, and different from certain contrast objects, then the respect in which the paradigms are similar to one another and different from the contrast objects will be the respect of similarity that is important for concept-instancehood. For instance, two red objects will be similar in many respects, but if we exhibit a blue object as contrast, then we draw attention to the relevant respect in which the two red objects are similar. The respect of difference would be brought out with special salience were we to exhibit to the subject a contrast object that differed from the paradigms only in the appropriate respect.

An example may help. One might ask which of the following are more similar to, or more different from, one another: a red tomato, a green tomato, and a red book? The two tomatoes are more similar in respect of shape, while the red tomato and the red book are more similar in respect of color. The problem, then, is how to keep the two red things together on grounds of similarity, while keeping the green things apart from them on grounds of difference, despite the fact that the red tomato is just as similar to the one as to the other. However, all of the similarity and difference relations are clear because similarity is always similarity in some particular respect. The red book is different in respect of shape from the two tomatoes and can serve as a contrast object along this dimension, while the green tomato is different in respect of color from the red tomato and the red book, and can serve as a contrast object along this dimension. There is no single, absolute level of similarity and difference between whole entities; rather, the distinctions between entities' properties give rise to multiple respects of similarity and difference.

Thus, Rand's theory of abstraction seems to be adequate in that it can meet certain classic objections to nominalism without reifying universals. But Price's theory of concepts seems to be adequate in that it persuasively argues for recognition as the (or at least a) fundamental and pre-linguistic conceptual capacity. Price thus needs Rand's theory of abstraction, and so we can see the advantage of a synthesis from his point of view. But has Rand any need for Price's theory? I think so.

4. Rand on Concepts

I want to begin by asking Rand, “What is a concept?” Unfortunately, her answer is not a very good one. Rand (1990) defines concept [9] as “a mental integration of two or more units possessing the same distinguishing characteristic(s), with their particular measurements omitted” (13). First, the definition is not in genus-differentia form (unless we are overly charitable about what can count as a genus or differentia), as Rand suggests definitions must be (41). But an important problem with this definition is that it defines a concept only as the product of a certain process, without specifying anything about the product other than that it is the product of that process. A concept, by this definition, is just what(ever) you get when you mentally integrate two or more units possessing the same distinguishing characteristic(s), with their particular measurements omitted. But what it is that you get out of this process is not said. (This would be like defining, say, 'scholarship' as 'what scholars produce.') [10] , [11]

But my broader critique of Rand's theory is that she confuses concepts and words and that this is mistaken. Regarding words and concepts, Rand says:

In order to be used as a single unit, the enormous sum integrated by a concept has to be given the form of a single, specific, perceptual concrete, which will differentiate it from all other concretes and from all other concepts. This is the function performed by language. . . . Every word we use . . . is a symbol that denotes a concept, i.e., that stands for an unlimited number of concretes of a certain kind. (10)

First, note the motivations implicit in the first sentence. Rand says that one cannot use a concept without having a word for it. But if using a concept rests on having a word for the concept, then using a concept must consist in linguistic activity. Presumably, no non-linguistic conceptual ability would rest on using words, so Rand is saying that there are no non-linguistic conceptual abilities. This point will emerge from a few more passages, which I will cite for other reasons.

But I want to attend more closely to the set of relations between concept, word, and things. A word 'denotes a concept,' but Rand apparently means this to be just the same (because of the 'i.e.') as 'standing for' the instances of the concept. However, the denoting relationship cannot be the same as the standing-for relationship. A word bears a denoting relationship to the concept with which it is associated, but the concept is mental. A word bears a standing-for relationship to the instances of the concept which it denotes, but the instances are usually not mental. Presumably, a word cannot have the same relation to something mental that it has to something non-mental. So denoting is not standing-for. Rand does not define either 'denote' or 'stand for,' (though we can make a guess about the latter) so all that we know about the two relationships is that they are not the same thing: a word relates somehow to things, and a concept also relates somehow to things, and, furthermore, a word's having its standing-for relation to things is conditioned by the word having a mysterious denoting relation to a concept, and the concept's having its mysterious relation to things. But we can guess that the 'standing for' seems to be 'standing in for'; that is, when we wish to think about things, we can avoid the burden of thinking of too many things by having a word stand in for them in our mind. Instead of having to mentally envision all of the instances of a concept, we can internally verbalize the word for the concept and we will have succeeded (somehow) in thinking of the instances. [12]

The appeal, of course, is to Rand's notion of unit-economy. This notion is worth reflecting on for a moment, because of the context in which Rand introduces it. She introduces it by way of explaining 'The Cognitive Role of Concepts.' She doesn't discuss any other role which concepts may have. By way of concluding her discussion (before turning to polemics), Rand says: “The primary purpose of concepts and of language is to provide man with a system of cognitive classification and organization . . .” (69). Again, she offers no other purposes. It seems that for Rand, concepts are exclusively cognitive in nature.

Rand says that “[w]ords transform concepts into (mental) entities . . .” (11) and also that “[a] word is merely a visual-auditory symbol used to represent a concept . . .” (40). A number of questions arise. What exactly is a mental entity? If a concept is not a mental entity before it is transformed into such with a word, what is it? How does a word represent a concept, and what is a concept without being represented? Is denoting something the same as representing it? How can an auditory or visual symbol transform anything into a mental entity?

First, I want to get a grip on what it is for a concept to be a mental entity. Rand does admit that this is metaphorical:

. . . we can call them “mental entities” only metaphorically or for convenience. It is a “something.” For instance, before you have a certain concept, that particular something doesn't exist in your mind. When you have formed the concept of “concept,” that is a mental something; it isn't a nothing. (157)

So mental entities are not to be thought of as entities on the model of physical substances; they are not independently existent. But Rand does agree with an interlocutor when he suggests that “a mental entity is a concrete” (156). Rand does not specify what the meaning of 'concrete' is, but when we look at the use of the word by Rand and her interlocutors, it seems that a concrete is something that can be perceptually or introspectively experienced and can be a concept-instance. We are encouraged in this interpretation by another of Rand's comments: “Take another, similar case: the concept of 'emotion.' What are its concretes? The various emotions which you observe introspectively, which you are able to conceptualize” (156, emphasis added). Here, Rand seems to regard an emotion as being an inner something that is open to inspection (“you observe introspectively”) and to being a concept-instance. [13] It seems that it is specifically in virtue of these two attributes that an emotion is a concrete. Likewise, then, a concept, to be a mental entity, must be open to being inspected and being a concept-instance (of the concept concept).

Since Rand is vague about how a concept relates to its instances, we can't follow up very well on a concrete's being a potential concept-instance. But we can follow up on the idea that a concept must be open to inspection. Rand is aware of this when she says that a concept must “be given the form of a . . . perceptual concrete.” Now, if a concept must be given the form of a perceptual concrete, then it doesn't natively possess such concrete form. So concepts, as such, are not concrete, and thus not inspectible and thus not entities. Rand seems to support this conclusion when she asserts that “[w]ords transform concepts into (mental) entities . . .” I can see no reason for Rand to draw this conclusion other than that she regards words as inspectible and concrete, but concepts as not so, and, thus, if concepts are to be entities, they must be associated with words, which are the form in which we experience the concepts. (Strictly speaking, and as we will see, there is no concept not associated with a word.)

But if this is right, what was a concept before it was transformed into a mental entity? It was not a concrete, because without the word, the concept is not inspectible. Rand's answer is that is an 'implicit concept': “The 'implicit' is that which is available to your consciousness but which you have not conceptualized” (159). One of Rand's interlocutors raises the obvious objection: If something is implicit if it is not yet conceptualized, and a concept is (obviously) always conceptual, then there can be no implicit concepts. Rand does not offer a response. However, another of the interlocutors suggests that the critique proves too much: “. . . you would also say that it is a self-contradiction to describe a fertilized egg in the womb as a 'potential man,' because a man is defined as a rational animal and the egg is not yet a rational animal.” Rand agrees with the response (161).

Let's explore this analogy between the potential man and the implicit concept. A potential man is not yet a man; 'man' is being used homonymously. To qualify something as potentially p is one way of negating its being p. Likewise, then, if an implicit concept is not conceptualized, it is not really a concept. To refer to something as an implicit concept is one way of saying that it is not a concept. But, in both cases, the negation is temporally indexical. Calling something potentially p says that it is not p, but also says that it will be p. Likewise, calling something an implicit concept says that it is not a concept but that it will be.

Under what conditions can an implicit concept become a concept? Rand becomes more specific when she says that “[a]n implicit concept is the stage of an integration when one is in the process of forming that integration and until it is completed.” When is the concept implicit and the process of forming it not yet completed? “At any stage before [the concept-acquirer] is ready to grasp the word [for the concept]” (162). So Rand's theory seems to run as follows: A subject begins to notice relations of similarity and difference between things. The awareness of these similarities is an implicit concept. When the subject begins to employ a word to refer to these similar things, then she has formed a concept for them.

In one sense, then, the choice of the word 'implicit,' while confusing, is very appropriate. Something is implicit if it is not stated, explicit if it is stated. If there is not yet a word for a concept, it has not and cannot be stated, so it is implicit. For Rand, who views concepts as a means to linguistic use, a concept that has no word associated with it (an implicit concept) is not (yet) a concept at all. [14]

Rand makes this spectacularly unclear but seemingly important statement on the relation between words and concepts: “the word is not the concept, but the word is the auditory or visual symbol which stands for a concept. And a concept is a mental entity; it cannot be perceived perceptually. That's the role played by words” (163, emphasis added). The phrase 'perceived perceptually,' of course, demands interpretation. Sometimes Rand uses the word 'perceive' correctly, to refer to a means of awareness of the physical world that is pre-conceptual but more integrated than the sensory; something like the direct sensory awareness of physical objects. But sometimes, she uses it incorrectly, to refer to any kind of consciousness. Presumably, Rand is not pointlessly repeating herself in this sentence. But she would be repeating herself if she were using 'perceived' correctly, because any perceiving, properly so speaking, is done perceptually. So let us assume that she is using 'perceived' incorrectly on the pattern just noted, but that she is using 'perceptually' correctly. In this case, Rand is saying that concepts are not something of which we can be perceptually conscious.

She then notes: “That's the role played by words.” The context seems to suggest that 'that' role is the role of being 'perceived perceptually'—words play the role of being something of which we can be perceptually conscious.

What's odd here is that Rand has said that “a concept is a mental entity; it cannot be perceived perceptually.” The phrasing suggests that the second clause is somehow a consequence of the first, so that a concept cannot be 'perceived perceptually' because it is a mental entity. But why not? Surely not because it is an entity; perception is all about the direct awareness of entities. Presumably, then, because it is mental. And indeed, it would be odd to perceive something mental, because perception is of the physical world, not the contents of one's own consciousness.

This passage, then, is highly puzzling. Two interpretations present themselves. On the one hand, Rand may be suggesting that words are always in the physical world; they are always marks on paper or sounds in the air. In this case, we can be perceptually conscious of them. But this eliminates the possibility of unspoken thoughts (i.e., subvocalizations). On the other hand, Rand may mean to say that words are how we experience concepts both internally and externally. The subvocalization of a word 'in' one's mind is how we employ in thought the concept with which that word is associated. But in this case, then, the 'perceptually' in 'perceived perceptually' is also being used wrongly. If words can be experienced 'perceptually' at the same time that they are objects to introspection, introspection must be something done 'perceptually'—an awareness's being done 'perceptually' is just its not being done conceptually. Thus, something is 'perceived perceptually' if we are introspectively or perceptually aware of it.

I would prefer to opt for the latter interpretation, because the former seems to make Rand into a behaviorist, while the latter simply makes her unclear. Unfortunately, when we re-examine the first sentence, we note that Rand says that a word is the “auditory or visual symbol which stands for a concept.” But an auditory or visual symbol is one in the physical world; it is not mental. So words are, indeed, always external. There is no introspective experience of words.

But why, then, has Rand said that “[w]ords transform concepts into (mental) entities . . .”? Focus on the word 'mental,' and recall that words are auditory or visual. How an auditory or visual symbol can transform anything into a mental entity is quite unclear, as auditory and visual symbols are non-mental. Indeed, this seems quite contradictory.

This is another version of a lesson learned earlier. Recall that Rand does not specify how it is that words relate to concepts or to things, or how concepts relate to things. Here, Rand seems to be trying to say that words relate to concepts by making them into mental entities—something introspectively observable. But under analysis, her claim makes no sense. So again, an attempt by Rand to explain how words relate to concepts collapses into terminal unclarity.

We are left wondering where concepts fit in. We have a process, called concept-formation, during which we have an implicit concept. But since 'implicit' performs a kind of negation in this context, we have no concept at all. At the end of the process, we have a word which refers to things. Why is a concept necessary at all? Why not deal simply with words? Indeed, Rand seems to drop concepts out of the picture when she says that “as far as the process of concept-formation is concerned, the word is the result of the process” (164). One might have expected that the process of concept-formation would result in a concept, but Rand says that a word is the result of the process. Rand also agrees when one of her interlocutors suggests that “until the word was interposed, there would not in the strict sense be a concept” (165). Unfortunately, she does not say where—between what things—the word is interposed.

However, these comments suggest a possibility. While Rand is quite unclear about the relation between words and concepts, I want to propose that, however vaguely she puts it, the clearest and most consistent relationship into which Rand puts words and concepts is the relation of identity.

Consider this claim: “Learning to speak is a process of automatizing the use (i.e., the meaning and the application) of concepts” (65). Unfortunately, this sentence is a little difficult to interpret. Had Rand said that learning to speak is the process of automatizing the use of concepts, we would know that automatizing the use of concepts is the same as learning to speak. But she said that learning to speak is a process of automatizing the use of concepts. This leaves open the possibility that there are other such processes of automatizing the use of concepts. However, Rand (1975) also says that “[a child] has barely acquired the ability to speak; he is not yet able to grasp the nature of this, to him, amazing skill, and he needs training in its proper use (i.e., training in conceptualization)” (196). With the use of 'i.e.,' Rand seems to be saying that learning to speak is the same thing as training in conceptualization.

Rand does discuss non-verbal identificatory experiences (which we will discuss below as perceptual judgments); they are made possible by implicit concepts. One of her interlocutors says:

Take the stage of concept-formation where a child regards certain entities as resembling each other. A child is observing these three notepads, and they are just entities [not yet conceptualized] so long as he does not show that he is treating them as distinct from other objects. . . .

But now he notices similarities and differences, and treats these as related together and distinguished from some other things. So these three objects are at this point units [i.e., entities which have become instances of a (perhaps implicit) concept]. . . .

Now at this point do I have the concept of “pad,” or do I still have something further to do . . . ?

Rand (1990) answers: “Yes. You have to give it a name” (167). She maintains this answer despite the example of a child who recognizes and responds differentially and appropriately to notepads because he has noticed that notepads are similar and can be treated the same. But, since he is treating the notepads differentially and appropriately, his treatment of them clearly is based on a concept, because a concept is the only mental entity (process, ability, or whatever) that can allow a subject to treat an open-ended category of non-essentially distinct things (like notepads) differentially and appropriately. But Rand simply views this as a preliminary to the possession of a concept: “what you are describing is the preconceptual stage. That is the mind in process. At the end of that process, he will be ready to grasp that a word names these objects” (169). I think that this claim makes Rand's view fairly clear. Nothing is a concept without having a word associated with it, and a concept gives us nothing other than the ability to use the word which is associated with it. Rand is pretty clearly running together concepts and words.

I want to consider four more passages that suggest the same interpretation.

1) . . . a concept is only a mental unit, a symbol, for a number of concretes of a certain kind. (175, emphasis added)

But it is obvious that words are symbols and that concepts, were they not words, would not be. So Rand seems to be referring to words with the word 'concept.'

2) A word which is not a proper name does refer directly to an indefinite number of concrete objects. A concept, in the form of a word, refers to them directly, not indirectly. (175)

Unfortunately, Rand does not make clear to what relationship she is referring with “in the form of.” But, in much Objectivist discourse, the form of a thing is the thing, processed by the subject so that it can become an object to that subject. [15] If Rand is employing the same usage here, then the word's being the form of a concept is the word's being identical with the concept and being the way the concept appears to introspection. This would contradict Rand's claims above that the word is not the concept, as well as her view of a word as a visual or auditory symbol, but it does make more sense in light of her view that a concept as such does not exist without being made into a mental entity by a word.

3) Of all the “anti-concepts” polluting our cultural atmosphere, “extremism” is the most ambitious. . . . To begin with, “extremism” is a term which, standing by itself, has no meaning. (Rand 1967, 177)

At one moment, “extremism” is an anti-concept, but at the next, it is a (meaningless) term. While hardly conclusive (since it is an anti-concept, not a concept), this equivocation again suggests that Rand identifies concepts with words.

4) Grammar is a science dealing with the formulation of the proper methods of verbal expression and communication, i.e., the methods of organizing words (concepts) into sentences. (Rand 1990, 37)

In this passage, the ability to use language is being explicitly identified with the possession and use of concepts.

Indeed, one can whip out a quick, clear, somewhat cheap deductive proof that Rand thought that concepts were words: recall that “as far as the process of concept-formation is concerned, the word is the result of the process” (164), and that “[t]he definition . . . refers to what a concept is—it refers to the product of the process” (167). So if a concept is the product of the process, and the process is concept-formation, and a word is the result of the process of concept-formation, then, assuming that results and products of processes are the same thing, words are concepts.

From this array of evidence, I conclude that, despite the occasional protest, Rand confused concepts with words by treating the possession of a concept as nothing but the mastery of a word. [16] The narrowness of this concern is unfortunate, because it means that her theory cannot deal with conceptual phenomena other than language.

5. A Critique of Rand's Confusion of Concepts and Words

Wallace Matson (1984) comes to the same conclusion for very much the same reasons. Motivated (apparently) by logical behaviorism, Matson is suspicious of inner inspectible mental entities that smack of a Cartesian soul. He thus proposes that, since Rand's concepts are just abilities to use words in the first place, we get rid of the concepts and keep only the words. On his revised version of Rand's theory,

We need only substitute for the troublesome (because subjective) mental state component the notion of an ability. . . . The test of a person's understanding of (the meaning of) a word is that he applies it to the right units . . . and, moreover, has the ability to do this as a causal consequence of his having recognized that just these units possess the essential characteristic(s) mentioned in the definition [of the word]. (33-34)

Matson's revised theory, however, fails to account for inner, mental dialogue because it does away with the inner and mental as such. Since Matson does not seem to believe in inner dialogue, this is fine; he prefers words to concepts specifically because they are not mental. To fully respond, I would have to go well beyond the bounds of this paper, so I will just lamely assert that I have inner dialogue and that Matson does, too.

But, assuming that we have inner dialogue, a Matson-like critique is still applicable. We can still do away with concepts and retain only words. We must simply retain not only the words we say, but also the words we think. Since Rand seems to identify words and concepts, and certainly doesn't think that a concept can be placed into discourse (inner or outer) without being associated with a word, it doesn't seem that concepts are necessary to account for either inner or outer dialogue. Moreover, it doesn't seem that concepts help account for them. However, a Matson-version of Rand's theory of abstraction, wherein abstraction concludes in a word rather than a concept, would still account for both kinds of dialogue.

This critique appears to make sense because Rand does not, in fact, appeal to concepts to explain anything but the use of words. [17] Rand says, of course, that it's our rational, conceptual faculty that explains virtually all of our activity. But she never offers the explanation. Indeed, because of her identification of concepts with words, concepts cannot account for a wide variety of conceptual functions—in fact, for every conceptual function other than the use of language. I want to discuss three such functions.

1) Perceptual judgments. Kelley (1986), following Rand, defines a perceptual judgment as “the conceptual identification of what is perceived” (208). If concepts are words, then a perceptual judgment must take the form of a sentence. But Kelley plans to consider mainly “judgments of a sort that are rarely made explicit, such judgments as that this is a desk. . . . These identifications are normally automatic for an adult; explicit judgments are made only in regard to less routine or obvious matters” (209). Rand (1975) is aware of such phenomena:

. . . you cannot perceive a table as an infant perceived it—as a mysterious object with four legs. You perceive it as a table; i.e., as a man-made piece of furniture, serving a certain purpose belonging to a human habitation, etc.; you cannot separate these attributes from your sight of the table, you experience it as a single, indivisible percept—yet all you see is a four-legged object; the rest is an automatized integration of conceptual knowledge which, at one time, you had to learn bit by bit. (192)

Rand is speaking of perceptual judgments, but she doesn't have the right to do so. If concepts are words—mental entities, as distinct from mental abilities—then the actual perceptual judgment, which is wordless, could not be conceptual. But of course it is conceptual.

2) Ready-to-hand. Martin Heidegger explores this phenomenon at some depth in Being and Time. [18] Something is ready-to-hand if we experience it (or, as Heidegger puts it, if it “reveals itself”) as a 'handy,' or useful, part of a larger context of handy things directed toward an end. For instance, I have a pad of sticky notes sitting to the right of my computer. As I pause in the writing to look up an important passage I've just recalled, I often reach for the sticky notes to mark the place. I, thus, experience the sticky notes as ready-to-hand. As I turn back to the computer, I experience it as ready-to-hand; in fact, I experience it as ready-to-hand as long as I am using it.

The ready-to-hand experience is similar to a perceptual judgment in that it is a conceptual identification of a perceived particular. It is also similar in that it is non-verbal. But it is different in that it happens in the context of action toward a goal and in the context of an array of related things that may become ready-to-hand. Thus, we are continually making perceptual judgments about everything in our surroundings, while we experience only some of these things as ready-to-hand. For instance, the cat on the floor to my left is not ready-to-hand but is identified in a perceptual judgment, while the glasses on my face are ready-to-hand (-to-eye) in addition to being identified in a perceptual judgment. This last example brings out the aggressively non-linguistic, background nature of both perceptual judgments and the experience of ready-to-hand.

Heidegger ([1926] 1996) suggests that the ready-to-hand experience is an “association which makes use of things” as distinct from a neutral and uninterested association such as my perceptual judgment that the thing to my left is a cat. He continues to say that this experience “is not blind, it has its own way of seeing which guides our operations and gives them their specific thingly quality” (69). This means that something is what it is, in terms of its place in a teleological context, in virtue of our assignment of function to the thing (while, of course, not just any assignment can be given to just any thing; physical nature imposes strict limits). [19] Heidegger is saying that the ready-to-hand experience is cognitive (“not blind”). He refers to this cognitive mode as 'circumspection.'

Heidegger does not talk about concepts, but once we acknowledge that ready-to-hand must be cognitive, we can see that it must be grounded in concepts (whatever Heidegger may think). Since (1) my experience of and subsequent treatment of the ready-to-hand thing is appropriate to an instance of the concept of which the thing is an instance, and (2) I typically have the ready-to-hand experience repeatedly with regard to an open-ended number of instances of the concept, and (3) the ready-to-hand experience disregards non-essential differences between instances of the concept, the ready-to-hand experience is a mode of awareness that treats an open-ended number of things the same so long as they are within a range along a respect of similarity and difference. These are all attributes possessed uniquely by the conceptual mode of awareness. However, Rand's theory cannot recognize that the ready-to-hand experience is conceptual, because for Rand, concepts are words and ready-to-hand is non-linguistic.

3) Know-how. Knowing how, as distinct from knowing that, is a notion developed by Gilbert Ryle. Ryle ([1949] 1984) critiques the idea that we always begin with an articulate theory and then act according to it. Surely he is right. I never learned, for instance, some theory of typing (as far as the activity of typing is concerned, it's irrelevant whether there is or ever was such a theory), and yet I am doing it right now. Like the previous two, this kind of knowing is often non-linguistic. Ryle says that “it would be quite possible for a boy to learn chess without ever hearing or reading the rules at all. . . . it should be noticed that the boy is not said to know how to play, if all that he can do is recite the rules accurately” (41). Language, then, is not necessary for the acquisition of at least some kinds of know-how and is incidental to the correct performance of the action one knows how to do.

Ryle also points out: “We all learned the rules of hunt-the-thimble and hide-and-seek and the elementary rules of grammar and logic in this way [the way the boy learned chess]” (41). Knowing how to use grammar and logic is prior to any familiarity with the rules of grammar and logic; indeed, those rules probably start by just articulating our prior know-how. Moreover, know-how must specifically be prior to the employment of language, because employing language is a matter of knowing how to use words. We must know how to use words before we can articulate any theory, including a theory to guide us in the use of words. So knowing how to use words precedes any knowledge that words must be used in such-and-such a way.

Know-how, like ready-to-hand, must be conceptual knowledge. If I know how to perform an action (like typing), I must be able to perform the action differentially and appropriately (e.g., I won't try to score a soccer goal while typing, but I will hit the 'return' key at the end of a paragraph). That is, I must be able to perform an instance of a concept because I possess the concept; I can perform some particular typing because I know how to type in general.

The example of typing is a bit interesting here, because Rand herself mentions it in her lectures on fiction-writing: “Physical skills like typing can also be taught. But to learn to type, more is required than merely listening to a factual lecture: you have to practice. . . . If an experienced typist were to ask herself, 'How do I do it?' she would answer, 'I just do it'” (Rand 2000, 51). [20] This seems right. But shall we say that the typist knows how to type? If so, then her knowledge must involve concepts. But if she cannot articulate how she types, then with what concepts is her know-how involved, since concepts are, for Rand, so intertwined with words?

The argument that Randian concepts cannot account for knowing how to do things is especially damaging to Rand's theory, because she opposes any radical distinction between theory and practice, suggesting that cognition has a practical purpose. But if Rand's theory cannot account for knowing how to perform actions, then practice as such has been left out.

6. Price, Method, and the Use of Concepts

Now we are in a position to see why I suggested that Rand's theory is in need of Price's as much as Price's is in need of Rand's. Rand has semi-identified concepts with words, and so cannot account for even a phenomenon such as perceptual judgment, much less ready-to-hand or know-how.

However, were we to borrow from Price to supplement Rand's theory, the synthesis would be able to account for some of these phenomena. If a concept is primarily a recognitional capacity, then perceptual judgments, which are recognitions of things as concept-instances, are capacities intrinsic in concepts; a concept is just the ability to perform perceptual judgments (among other abilities, of course). Moreover, ready-to-hand is accounted for insofar as it is a recognition of things as handy concept-instances and also recognizes the context of the handy entity in virtue of which it is handy.

However, performative abilities, which are differential and appropriate, go unexplained. Know-how and a certain feature of the ready-to-hand experience are both performative, not recognitional, in orientation. So treating concepts as a means of recognition is insufficient to explain these capacities. However, using Price gives us two things: (1) a hint about perceptual judgments, and (2) the notion that concepts should be treated as (recognitional) abilities, which then forms the paradigm for treating concepts as abilities of various kinds.

Rand (1990) has a valuable suggestion that she did not fully exploit but which should fill this explanatory gap. This is her category of concepts of method. Rand says:

A special sub-category of concepts pertaining to the products of consciousness, is reserved for concepts of method. Concepts of method designate systematic courses of action devised by men for the purpose of achieving certain goals. The course of action may be purely psychological (such as a method of using one's consciousness) or it may involve a combination of psychological and physical actions (such as a method of drilling for oil), according to the goal to be achieved. (35-36)

Unfortunately, Rand does not sufficiently distinguish concepts of method from other kinds of concepts; she treats them as being simply another genus, rather like furniture and pet are different genera, albeit lower ones in the conceptual hierarchy. This is true because the relation between concepts of method and instances of the appropriate method is the same ('designation') as that between concepts which are not of method and their concept-instances.

Nevertheless, the suggestion is a fruitful one, if we reconceive the relation between concepts of method and instances of the appropriate method. The basic relation between ordinary concepts (like pet and furniture) and their instances is the capacity to recognize.

The basic relation between concepts of method and their instances is the capacity to perform, according to the appropriate method. Typically, possessing a concept of method will also be a capacity to recognize others' performances, which are in accordance with the method. Also typically, concepts of both kinds will be associated with a word, and the concept will be a capacity to refer to its instances by means of that word; but contrary to Rand, this linguistic capacity is not basic to the possession of the concept.

This notion of concepts of method is especially valuable when it comes to language. Rand argues:

Grammar is a science dealing with the formulation of the proper methods of verbal expression and communication, i.e., the methods of organizing words (concepts) into sentences. Grammar pertains to the actions of consciousness, and involves a number of special concepts—such as conjunctions, which are concepts denoting relations among thought (“and,” “but,” “or,” etc.). . . . The purpose of conjunctions is verbal economy: they serve to integrate and/or condense the content of certain thoughts. (37)

I don't think that Rand is quite right to suggest that conjunctions denote relations among thoughts. Rather, let me propose that the concept conjunction is the capacity to conjoin thoughts. When those thoughts are being expressed linguistically, the concept conjunction is the capacity to conjoin the linguistic expressions of thoughts, typically by uttering such words as 'and' and 'but.' Likewise, the concept disjunction is the capacity to disjoin thoughts (and to utter 'or' by way of expressing the disjunction of thoughts that have been linguistically expressed). These concepts do not denote anything at all. Moreover, nothing is denoted by words like 'and' and 'or'; rather, by uttering these words we perform the acts of linguistically expressing conjunction or disjunction. (Of course, the words 'conjunction' and 'disjunction' will refer to all the acts of conjoining and disjoining, respectively.) Possessing the concept is the capacity to perform the appropriate act of creating a certain kind of logical relation, and typically to utter the words by means of which the act is linguistically expressed. Typically, these concepts of method will also allow us to recognize the performances of other people who follow the same method—to understand what it means for another person to conjoin or disjoin thoughts.

The linguistic expression of these concepts of logical method will sometimes involve the utterance of multiple words. For instance, the act of disjoining propositions is often performed not with the single word 'or' but rather with the two words 'either' and 'or,' located at the correct places in the sentence.

Consider also those concepts of logical method that are not elements of grammar, such as the laws of logical inference. Rand proposes that logic is “the fundamental concept of method, the one on which all the others depend” (36). It is possible that this is an exaggeration of the genetic importance of logic; while logic is a central concept of method for the abstract reasoning that Rand is seeking to explain and justify, there may be many concepts of method that precede it and, thus, cannot rely on it, because there are many methods that one can perform before one is capable of abstract reasoning. But on the present understanding of concepts of method, logic is the capacity to perform correct logical inferences; typically, logic will also be the ability to recognize correct logical inferences and logical errors.

There are also concepts of linguistic method that are not concepts of logical method. For instance, the act of making an assertion is a matter of correctly arranging all of the words in a sentence and correctly punctuating or intoning the sentence. The concept assertion, then, is the capacity to correctly arrange and intone or punctuate sentences, even though there is no single word in the sentence indicating that it is an assertion. People can perform asserting without using the word 'assertion,' so even concepts of linguistic method will often not be associated with a word in the mind of the concept-holder. Perhaps the present theory of concepts of method as performative capacities can be linked to the theory of speech acts through the concept of linguistic method, the capacity to perform kinds of speech acts.

The immediate question for this notion of concepts of method is: How are they acquired? We do not acquire the ability to perform according to a method by observing many instances of someone else's performances according to a method; we have to do it ourselves. Rand (2000) addresses these issues in her lectures on fiction-writing. This is most appropriate, because, in these lectures, Rand is trying to teach how to perform the activity of writing. She continues the typing example I mentioned earlier by saying: “First you learn to move your fingers and strike the keys—slowly and with conscious effort. Learning to type consists of automatizing this skill. . . . The same is true of . . . any physical skill. First it is learned consciously—and you are in command of the skill when it becomes automatic, so that conscious attention is no longer required” (51). Note that knowing how to type is something that becomes automatized through practice.

However, in discussing concepts that are not concepts of method, Rand says: “In order to perform its unit-reducing function, the symbol has to become automatized in a man's consciousness, i.e., the enormous sum of its referents must be instantly (implicitly) available to his conscious mind whenever he uses that concept” (Rand 1990, 64). Now, it's not at all clear what it means for the concept to become automatized, though Rand (2000) provides some hints in her fiction-writing lectures:

With the exception of proper names, every word is an abstraction. One way to have words come to you easily—words which express the exact shade of meaning you want—is to know clearly the concretes that belong under your abstractions. (12)

This is not immediately helpful, but note that the issue of words coming to you easily is the issue of automatization: “Before you sit down to write, your language has to be so automatic that you are not conscious of groping for words or forming them into a sentence” (1). Now, if automatization is what you have done to your concepts (words) when they come easily, and your words come easily just in case you know clearly which concretes belong under which abstractions—which things are the referents of which words—then automatization of concepts not of method, consists in clarity about the referents of the concepts (words). How is this achieved? “Fill your subconscious with as many concretes as possible under every abstraction you deal with—then forget about them. Your subconscious does not forget” (54). Having done this, you will “master the relationship of abstractions to concretes [and] you will know how to translate an abstract theme into action, and how to attach an abstract meaning to an action idea” (55). The trick to automatization of the use of words is tying as many concretes as possible to the word; having as many referents of the word as possible subconsciously in mind.

Now we can return to the question of the acquisition of concepts of method. Where concepts not of method become automatized (can be easily called upon) by having many referents connected to them in the subconscious, concepts of method can be automatized by practice. But the two processes seem to share a common structure: attending to referents of a concept (word) helps automatize that concept (word), while performing according to a concept of method helps automatize that concept and one's ability to perform according to it. In both cases, automatization builds on past experience.

7. Conclusion

The theory here proposed involves taking Rand's theory of abstraction and enriching it with Price's notion of concepts as abilities. Price's specific notion of concepts as recognitional capacities, as well as a revised version of Rand's notion of concepts of method, completes the theory.

The motivation has been explanatory. There are, I have argued, a number of concept-based abilities—perceptual judgment, ready-to-hand, and know-how—which do not involve words and, thus, are not accounted for by Rand's theory.

I want to discuss the motivational issue for a moment. Rand argues that the recognitional acts of an infant, who has formed the capacity to differentially and appropriately respond to an open-ended number of non-essentially different entities, are not based on concepts. Rather, they are based on a pre-conceptual phase referred to as an 'implicit concept,' where 'implicit' is a kind of temporally indexical negation, like 'potential.' The adjustment I wish to make to Rand's theory amounts to refusing to treat 'implicit' as a negation of any kind. Whereas Rand regards a wordless concept as not really a concept, I regard it simply as a wordless concept. Rand is wrong because a theory of objectivity has a wider explanatory burden than the one she recognizes. Rand is concerned exclusively with the objectivity of abstract theorizing and loses, for the time being, her insistence on unifying theory and practice. But day-to-day events such as perceptual judgment, ready-to-hand, and know-how are rational, concept-based activities even though they are not acts of reasoning. A theory of objectivity cannot come to grips only with the objectivity of word-use; it must also understand the objectivity of performance. The present theory is intended to move us toward such an understanding.

This can be phrased in another way. There are a number of human activities that can be arranged on a continuum. Rand wishes to abstract the use of language from other activities, despite the necessity for an abstract and universal capacity to underlie several of the excluded activities. But language is only one of the abilities that requires an abstract and universal capacity. All of these capacities are potentially rational and objective, even though only language-use can underlie explicit reasoning. Thus, the difference between language-use and, say, hammer-use is less essential than Rand thinks, so we ought to integrate where she differentiates. Reasoning is but one of many rational activities, to be treated, at a certain level of abstraction, as indistinct from the others. These rational activities are to be integrated around the fact that they are all based on concepts, and nothing else is.

Earlier, I critiqued Rand's definition of concepts, so I should now provide my own. Tentatively, I propose that 'concept' be defined as “an ability to recognize, perform, refer to, and/or infer about any member of an open-ended group of things (entities, properties, relations, events or actions), which are regarded as sharing some essential, differentiating trait(s).” Like Rand's definition, this does not take a genus-differentia form (unless we are rather generous about what a differentia can consist of), and I'm not certain that it is the best definition, but it does capture the essence of the theory here proposed.


I am pleased to note that Chris Furlong, Carolyn Ray, Chris Matthew Sciabarra, Jason Walker, and an anonymous reviewer commented informatively on a draft of this paper. I have benefitted tremendously from discussion of these issues with Chris Furlong; indeed, the project was conceived as a result of conversation with him. Any errors, of course, remain my own.



[Note 1] . To be sure, Rand says that she is rejecting nominalism. However, considered as an ontological hypothesis, nominalism is just a negative existential claim about real universals. So Rand is clearly a nominalist, ontologically speaking. Den Uyl and Rasmussen (1984b) argue that Rand is a moderate realist: “A concept signifies individuals that are something or other. Sheer extension does not suffice. There must be an intelligible feature in virtue of which classifications are made” (15). While this is true, it does not commit one to the real abstract existence of the 'intelligible feature.' Such features are simply consistent with nominalism. Rand's theory, as I understand it and will treat it here, is an attempt to answer the question of just how there can be such a feature without that feature being abstract.

[Note 2] . Of course, this is hopelessly optimistic. Many difficulties, including most especially the explanation of the intentionality of concepts (and words, beliefs, actions, and so forth) will remain. I hope to address these in the sequel, building on the present work.

[Note 3] . I am not fully comfortable with the term 'concept-instance,' because it suggests that the concept-instance has a relation (of instantiation) with a Form or essence. Price does not believe in non-mental universals, and neither do I, so I will go ahead and follow Price's terminology. But, for the record, by an 'instance of a concept' I mean one of the things that the concept is the capacity to recognize, perform, envision, infer about, refer to, or whatever else a concept allows us to do with things.

[Note 4] . To avoid a host of use-mention and related confusions, I will refer to words in the usual way (by placing them within single quotation marks) and to concepts by italicizing the word with which they are associated.

[Note 5] . One might also imagine brain-damaged patients who have an unusual combination of abilities and disabilities; a trained violinist with this kind of injury might be able to play the violin but would not be able to identify a violin. She would play out of habit, as it were. Whether such patients exist or not is an empirical question, but their existence would cast further doubt on Price's argument.

[Note 6] . The clearest presentation of Rand's theory, especially regarding its solution to these two objections, is Kelley 1984. I want to briefly mention a methodological issue. As we will see, Rand regards concepts as being the result of a certain process that begins with perception. Thus, by definition, there are no concepts that are not formed based on this process, and, thus, no concepts that are innate. I think that defining innate concepts away is a bad argument for empiricism, though I tentatively accept the conclusion for alternate reasons. The goal of philosophical empiricism, it seems to me, ought to be to allow one's philosophical theory of concepts to be consistent with a wide array of psychological results. A theory of concepts that has no philosophical commitment to innate concepts need not be rejected if psychologists discover that there are no such concepts. A theory of concepts that has a philosophical commitment to the non-existence of innate concepts similarly crosses into the psychologist's domain. I defend here empiricism of a sort that makes no commitments about innate concepts. To be sure, any theory of concepts must make some empirical predictions. But in lieu of any good philosophical reason to accept or deny innate concepts, I suggest that philosophers ought to do neither.

[Note 7] . On perception as the discrimination of entities from their background, see Kelley 1986, especially ch. 5.

[Note 8] . Unfortunately, I cannot come up with an alternative to the phrase 'experience as,' as in “Things are experienced as different.” The phrase 'experience as ' suggests a predicative or recognitional experience, which is specifically what the argument against Russell is saying we do not have to have with regard to difference. In this case, ordinary language will simply be misleading.

[Note 9] . Ordinarily (and, I think, correctly), we would say “Rand defines 'concept.'” But Rand (1990) claims that we define concepts rather than words (40), and I will not argue the point now.

[Note 10] . Another problem is that Rand does not tell us what it is to mentally integrate units that are non-mental. Typically, to integrate physical things is to bring them into some physical relationship with one another such that they become a larger or combined thing. Gluing two things together is integrating them, and putting a puzzle together integrates its pieces. But if the integration is mental, then the integration must be something on a totally different order from these examples. Frankly, Rand's definition just makes no sense as a definition and is useless beyond helping us focus on Rand's confusions.

[Note 11] . Unfortunately, Rand (1990) is very definite about this definitional practice. In the appendices to her epistemological work, she answers an interlocutor's question by saying: “The definition on page 10 [the one currently under discussion] refers to what a concept is —it refers to the product of the process” (167).

[Note 12] . The 'somehow' issue is the issue of the intentionality of concepts (and words).

[Note 13] . It has been suggested to me that, by 'concrete,' Rand is merely trying to distinguish between the members of a category and the concept for the category; so concreteness is related to abstractness. But I think that this is mistaken because, had Rand merely wished to contrast the abstract concept emotion with the concrete emotions, then she need not have said that the emotions are introspectively observable, because being concrete as opposed to abstract consists exclusively in being (at least potentially) a concept-instance for some abstract concept.

[Note 14] . Sciabarra (1995, 168-74 and esp. 210-15) suggests a parallel between Rand's 'implicit concepts' and Hayek's and Polanyi's 'tacit knowledge.' No such parallel exists, though Sciabarra's larger point about Rand's interest in articulating the tacit dimension of human functioning may hold. To say that a concept is implicit is to say that it is not a concept, while to say of knowledge that it is tacit is to specify what kind of knowledge it is. Since Rand views human knowledge as conceptual in nature, and concepts as necessarily explicit, she would reject the notion that there can be unstated knowledge.

To put it another way, Sciabarra seems to think that, because Rand says she wants to have theory-practice integration that she actually does some of the work necessary to achieve it. I disagree; it seems to me that the tenor of Rand's epistemology is focused exclusively on abstract theorizing and essentially ignores practice (except as a side-effect of theory). In this regard, note Johnson's critique of Rand's ethics as intellectualist (Johnson 1999).

Now, Rand appeals to unstated knowledge rather often. Nevertheless, she does not appeal to knowledge that employs implicit concepts—there can be no such knowledge. Moreover, it is hardly my contention that Rand is always consistent, so even if Rand does appeal to knowledge based on implicit concepts, the problem would not necessarily be with my interpretation, but might be with Rand. Thanks to an anonymous reader for calling this problem to my attention.

[Note 15] . See especially the discussion of perceptual form in Kelley 1986. But the usage is much older in the Objectivist tradition. See, for instance, Rand 1990, 279-81. Also, Branden's definition of emotion in Branden 1962 is of interest, especially when synthesized with some comments in Branden 1966. (Branden 1962 and Branden 1966 were published in newsletters that Branden and Rand co-published; Rand sanctioned everything in them, so we know that Rand agreed with Branden's definition.) Branden (1962) defines an emotion as the “psychosomatic form” (3) of a certain kind of judgment, while, later, he suggests that “[a]n emotion as such cannot be repressed; if it is not felt , it is not an emotion” (1966, 120). When combined, these two statements say that an emotion is the psychosomatic form of a judgment, but also that without the experience (the psychosomatic form) the emotion does not exist. This suggests that the experience and the emotion are identical, even though the experience is the form of the emotion.

We would expect that one thing's having another thing as a form would indicate that there is a relationship between them and that they are different things. But Objectivist usage consistently indicates that the relationship of being 'in the form of' indicates a slightly odd version of the identity relation. If one thing has something else as its form, that something else is, in fact, it.

[Note 16] . This does leave us with a question about different languages; if there is a certain concept associated with a word in one language, can a speaker of a different language have the same concept? I doubt Rand would have any difficulty here. We do speak of words in two languages as being 'the same word.' Likewise, they can be associated with the same concept. So, to be precise, we can say that, for Rand, the possession of a certain concept is nothing but the ability to refer, with some word in some language, to the instances of that concept.

[Note 17] . However, the process of concept-formation does account for things' being in categories. But the categories are just categories of referents of words, on the critique.

[Note 18] . Even unorthodox Rand followers are liable to be disturbed by the very mention of Martin Heidegger, much less an attempt to draw positively from his philosophy in an attempt to improve on Rand's theory. Actually, I believe that Rand and the early Heidegger may have a great deal in common. Consider, as an example, Rand's description of Howard Roark's walk back to the campus at the beginning of The Fountainhead : “He looked at the granite. To be cut, he thought, and made into walls. He looked at a tree. To be split and made into rafters. He looked at a streak of rust on the stone and thought of iron ore under the ground. To be melted and to emerge as girders against the sky” (Rand 1968, 16). It is instructive to place this beside a characteristic claim of Heidegger's: “The forest is a forest of timber, the mountain a quarry of rock, the river is water power, the wind is wind 'in the sails' ” (Heidegger [1926] 1996, 70 [standard pagination]). This isn't essential, and it doesn't prove anything, but it is suggestive.

[Note 19] . This is more than passably reminiscent of Rand's claim that things are objectively, not intrinsically, of value.

[Note 20] . Thanks to Chris Sciabarra for calling this passage to my attention. (I can't help but notice, by the way, that Rand refers to the hypothetical typist with 'she.' Either 'she' is gender-neutral, or 'typist' is feminine in gender—which it isn't.)


Branden, Nathaniel. 1962. Intellectual ammunition department: Reason and emotion.

The Objectivist Newsletter 1 (January): 3.

___. 1966. Emotions and repression, I. The Objectivist 5 (August): 120-29. Den Uyl, Douglas, and Rasmussen, Douglas, eds. 1984a The Philosophic Thought of Ayn Rand . Urbana: University of Illinois Press.

Den Uyl, Douglas, and Rasmussen, Douglas. 1984b. Ayn Rand's realism. In Den Uyl and Rasmussen 1984, 3-20. Emmet, E. R. 1954. The philosophy of resemblances. Philosophy 29: 146-51. Heidegger, Martin. [1926] 1996. Being and Time . Trans. Joan Stambaugh. Albany: State University of New York Press.

Johnson, Gregory. 1999. Liberty and nature: The missing link. The Journal of Ayn Rand Studies 1, no. 1 (Fall): 135-66.

Kelley, David. 1986. The Evidence of the Senses . Baton Rouge: Louisiana State University Press.

___. 1984. A theory of abstraction. Reprinted as a pamphlet by Institute for Objectivist Studies; originally in Cognition and Brain Theory 7: 329-57. Matson, Wallace. 1984. Rand on concepts. In Den Uyl and Rasmussen 1984a, 21-37.

Price, H. H. 1969. Thinking and Experience . 2nd ed. London: Hutchinson.

Rand, Ayn. [1943] 1968. The Fountainhead . 25th anniversary ed. New York: Signet. ___. 1990 . Introduction to Objectivist Epistemology . 2nd ed. Eds. Leonard Peikoff and Harry Binswanger. New York: Meridian. ___. 1975. The New Left: The Anti-Industrial Revolution . 2nd ed. New York: Signet.

___. 1967. Capitalism: The Unknown Ideal . New York: Signet. ___. 2000. The Art of Fiction: A Guide for Writers and Readers . Ed. Tore Boeckmann. Introduction by Leonard Peikoff. New York: Plume.

Russell, Bertrand. [1912] 1988. The Problems of Philosophy . Buffalo: Prometheus Books.

Ryle, Gilbert. [1949] 1984. The Concept of Mind . Chicago: University of Chicago Press.

Sciabarra, Chris Matthew. 1995. Ayn Rand: The Russian Radical . University Park: Pennsylvania State University Press.

Wittgenstein, Ludwig von. [1945-49] 1958. Philosophical Investigations . English text of 3rd ed. Trans. G. E. M. Anscombe. Englewood Cliffs, New Jersey: Prentice Hall.