THE EXPERIENCE OF OBJECTS AND THE OBJECTS OF EXPERIENCE

Lakoff (1987) has been a major force in the recent effort to redefine the study of concept formation, semantics, metaphor, and ultimately all of scientific psychology. In place of the traditional "objectivist" account of these domains, Lakoff offers "experientialism," a position that has become increasingly popular in a wide array of psychological subspecialties over the last several years. We believe that Lakoff's account of the relation between experientialism and objectivism is fundamentally flawed. The primary sources of the problem are an equivocation in his account of objectivism, and a misunderstanding of the relation between classification schemes and truth. Moreover, we argue that a suitably sophisticated form of objectivism can subsume experience under its aegis (and that psychology might be impossible otherwise). In any case, we try to show that the alternative account of semantics he provides fails because it falls prey to precisely the same criticism he considers crucial to the refutation of traditional semantics.

George Lakoff's book Women, fire, and dangerous things (1987) has been a major force in the recent effort to redefine the study of concept formation, semantics, metaphor, and ultimately all of scientific psychology. Starting with the celebrated work of psychological and anthropological researchers such as Berlin and Kay (1969), Kay and McDaniel (1978), Brown (1958), Ekman (1971), and Rosch (1973, 1975, 1978), Lakoff strives to show that traditional accounts of cognition, semantics, metaphor, and even metaphysics are fundamentally misguided. The book ranges through a stunning array of disciplines and findings, and to critique it fully, point by point, would require a document at least as long as Lakoff's--some 600 pages. Instead, we choose to concentrate our attention on only those of Lakoff's arguments that we believe are crucial to his overall position, that lend themselves to a succinct formulation, and that cannot be maintained given the current state of available evidence.

In place of traditional "objectivist" views of cognition, semantics, and metaphysics, Lakoff offers "experientialism", a position that has become increasingly popular in a wide array of psychological, philosophical, and linguistic subspecialties over the last several years. This position holds, inter alia, that:

(1) thought is embodied; the structures used to put together our conceptual systems grow out of bodily experience,

(2) thought is imaginative in that those concepts that are not directly grounded in experience employ metaphor, metonomy, and mental imagery,

(3) thought has gestalt properties, and is thus not atomistic; concepts have an overall structure that goes beyond merely putting together conceptual "building blocks" by general rules,

(4) thought has an ecological structure; it is more than just the mechanical manipulation of abstract symbols. (Lakoff, 1987, pp. xiv-xv)

Principally, Lakoff opposes this position to what he calls "objectivism". Objectivism is said to hold that:

(1) thought is the mechanical manipulation of abstract symbols,

(2) the mind is an abstract machine manipulating symbols essentially in the way a computer does,

(3) symbols get their meaning via correspondences to things in the external world,

(4) thought is therefore abstract and disembodied,

(5) thought is atomistic,

(6) thought is logical in the narrow technical sense.

(Lakoff, 1987, pp. xii-xiii)

It is only fair to note at the outset that Lakoff does not believe that experientialism and objectivism share no features in common at all. According to Lakoff, they both recognize:

(1) a commitment to the existence of the real world, (2) a recognition that reality places constraints on concepts,

(3) a conception of truth that goes beyond mere internal coherence,

(4) a commitment to the existence of stable knowledge of the world. (p. xv)

We believe that Lakoff's account of the relation between experientialism and objectivism is fundamentally flawed. The primary sources of the error are (1) a crucial equivocation in his account of objectivism, and (2) a misunderstanding of the relation between classification schemes and truth. Moreover, we believe (3) that a suitably sophisticated form of objectivism can subsume experience under its aegis (and that psychology might be impossible otherwise), and (4) that, in any case, the alternative account of semantics he provides fails because it falls prey to precisely the same criticism he considers crucial to the refutation of traditional semantics. It is our primary objective in this paper to demonstrate these claims.

According to Lakoff, one of the fundamental tenets of objectivism is what he calls the "Independence Assumption". This amounts to the claim that,

existence and fact are independent of belief, knowledge, perception, modes of understanding, and every other aspect of human cognitive capacities. No true fact can depend upon people's believing it, on their knowledge of it, on their conceptualization of it, or on any other aspect of cognition. Existence cannot depend in any way on human cognition. (p. 164, italics added)

For Lakoff, the crucial importance of this claim is clear: if objectivism sanctions only facts that are in no way dependent on human cognition, then it can never sanction what Lakoff dubs "institutional facts." These are facts that depend upon the way a person or culture conceptualizes the world, such as facts about marriage, business, school, or art. In Lakoff's words,

Since institutions are products of human cognition, institutional facts must depend on human cognition, which violates the Independence Assumption, which states that no fact can be dependent on human cognition. (p. 170)

Lakoff has conflated several distinct senses of the term "dependence" in this argument. We will outline four such senses which we call logical, cultural, idiosyncratic, and radical dependence, respectively.

The first form of dependence is the trivial one that cognition has on itself. One cannot study cognition unless cognition exists to be studied. This kind of dependence, far from undercutting the objective study of cognition, however, is one of the conditions of possibility of such study. In this sense, the "facts of cognition" are dependent upon cognition in exactly the same way that the "facts of physics" are dependent upon the physical. This would be so trivially true as to be hardly worth mentioning, except for the fact that it shows Lakoff to have gone too far when he says, as cited above, that "existence cannot depend in any way on human cognition." The existence of thought, for instance, depends on there being cognition per se, for there could not be one without the other. This form of dependence is purely syntactic--one might call it "logical dependence"--but it is a kind of dependence nevertheless.

Another sense of dependence--one that Lakoff makes use of--is the kind of dependence that cultural institutions have on cognition. If there were no cognitive agents, then institutions such as universities, marriage, and art could not exist. This might be called "cultural independence." Lakoff argues that because such institutions are created by the interpersonal agreements of people, such institutions are not "objective" and are therefore beyond the boundaries of objective study. This simply does not follow. It is a perfectly objective fact, for instance, that in the West there are institutions called "universities" that are organizations dedicated to the development of knowledge. Anyone who said there were no universities would simply be uttering a falsehood. The same is true of marriage. It is simply true that our marriage rites consists of certain things being said and believed by certain people. As before, anyone who denied these facts would simply be wrong.

Lakoff, we think, has confused "objectivity" with a particular construal of what it means to have a "science." Because institutions such as the university and marriage are of human invention, it is difficult to envisions a successful science of them aimed at the discovery of laws of such institutions, in the sense that one discovers laws of physics. Any empirical generalizations put forward by scientists of such institutions could be falsified not simply by the discovery of counterexamples, as is usually the case in natural science, but by the intentional invention of such counterexamples. For instance if it were claimed, as a law of universities, that they all have buildings, a contrary-minded groups of radicals could establish an "open-air" university in which all classes and administrative functions were carried out in a local park, all for no other reason than to disprove the theory. This does not happen in natural science. Such possibilities perhaps exclude institutional facts from being captured by scientific laws, but they do not render the facts themselves somehow inherently "subjective."

It is important not to confuse two independent meanings of "subjective" here, as we are afraid that Lakoff has. On the one hand, there is the sense in which something is "subjective" because it implies the operation of a mental "subject," purely in the Cartesian sense. In this sense, all thoughts are subjective in that they presume the existence of a thinker. Such subjects are related to the "objects" of their thoughts (though Descartes used the terms in almost the opposite manner, much to the chagrin of many first-year philosophy students), but none of this has any bearing on the issue of the objectivity of subjectivity of knowledge in the epistemological sense that Lakoff seems to be discussing. It is purely an ontological matter with respect to thoughts per se. The existence of thought requires the existence of a thinker; a mental subject.

Lakoff's epistemological claims are strictly a matter of the other meaning of "subjective;" viz., pertaining to the universality, or at least the generality, of the claim being made. That one thinks Manet to have been a good painter, for instance, is a "subjective" opinion, in this sense, unless one gives reason and evidence that are likely to make others believe it too. Such reasons make the judgment more "objective." The important point is that the epistemological subjectivity of my opinion about Manet is completely independent of the ontological subjectivity of its being a product of a Cartesian subject.

This leads us to two other species of dependence on cognition. The third is the kind of dependence expressed by Kant's immortal phrase, "de gustibus non est disputandum;" roughly, "matters of taste are not up for debate." If I like lima beans, it's as simple as that. You cannot argue me out of my liking. There are no facts you can bring to bear against it. This might be called "idiosyncratic" dependence. This is the kind of dependence on cognition that objectivism blocks. For all this, there is still a sense in which the very fact of my liking lima beans is an objective fact. If you assert that Hortense likes lima beans, you may be wrong.

Finally, there is a fourth kind of dependence on cognition, and this is one that also runs contrary to what has traditionally been called objectivism. It is the kind of dependence in which an individual predicates cetain properties to objects for which no criterion can be articulated. We call this "radical" dependence. The classic example is "good." If one says that lima bean are good (rather than just that I like them), and can give, in principle, no natural criterion for their goodness (e.g., it is not a matter of their tasting good, or being nutritious, or growing into giant stalks that lead to places of great wealth), then that opinion of lima beans is dependent on cognition. This is subjectivity in a very radical sense. It is not simply that no criterion is given; it is, rather, that no adequate criterion is possible in principle.

The reason this sort of dependence is seen as being opposed to objectivism is that it was claims such as these that the most influential objectivists of this century--the logical positivists--were trying to eliminate when they declared in the 1930s that every contingent statement that is not empirically verifiable is nonsense. This criterion turned out to be untenable (viz., because no propositions of the form "All x are y" are, strictly speaking, verifiable), and the logical positivists rejected it themselves before 1940. The continuing aim of the logical positivist project, however, was to exclude from consideration statements that are dependent in this radical sense. With this in mind, we can begin to see who Lakoff's theory is really targeted at: the straw-man of 1930s logical positivism. Lakoff, we believe, would like to paint more modern objectivists with the same brush, but no one still holds to the verificationist criterion, and so the point seems to be moot.

To summarize the claim, Lakoff is mistaken in asserting that objective facts cannot be about human cogntion itself. It is a perfectly objective fact, for example, adequately independent of the cognition of any human observer, that we are now presenting a research paper in an academic journal. Under Lakoff's formulation of the Independence Assumption, however, this is not an objective fact because notions such as "present," "research paper," and "academic journal" only have meaning relative to our particular cultural institutions, which are, in turn, dependent upon the beliefs and desires of (some of) the members of our culture. Put another way, Lakoff claims that science demands that facts be ontologically independent of human cognition; that they continue to be true even if there are no minds at all. Clearly cultural institutions cannot exist if there are no members of the culture to bring them into existence as part of their efforts to realize their values and desires. There can be no journals without editors, publishers, and subscribers; no universities without professors, students, (and administrators, alas); no marriage without spouses; no conventions without conventioneers.

This, however, is not the demand of science. What science demands is only that facts only be epistemologically independent of cognition; that they continue to be true in the absence of any particular observer. That is, facts cannot depend upon the whims and biases of a single person. For instance, science does not accept the "fact" that lima beans are good just because one person happens to believe it to be true. Science requires objective criteria of justification for the claim of goodness; it does not require that goodness is itself independent of human cognition. That is, it would require an explication of "good" (e.g., nutritionally, gustatorily, ecologically) that could be scrutinized, criticized, and applied by any suitably equipped apprehender. In this sense, institutional facts are fully independent and objective. For instance, we continue to be employed by the universities at which we teach regardless of who is, or is not, in the business ascertaining that fact. The fact may be contingent on a cultural consensus about what universities are and what employment is, but certainly not on the beliefs of a particular apprehender of the fact.

To put the matter very succinctly, Lakoff argues:

P1) Objectivism recognizes all and only facts that are independent of cognition.

P2) Institutional facts are not independent of cognition.

C) Objectivism does not recognize institutional facts.

We reply that he has equivocated on "independent," and reformulate the argument:

P1) Objectivism recognizes all and only facts that are epistemologically independent of cognition.

P2) Institutional facts are epistegwlogically independent of cognition (though they are not ontologically independent of cognition).

C) Objectivism recognizes institutional facts.

The crucial fact is that the existence of cognition just is an objective fact of the world. We believe that it is only the implicit conviction that mind and world belong to separate realms that leads to the problems that are "discovered" and the subsequent attempts to "solve" them. This is a sort of dualism that we believe Lakoff would explicitly reject, but is implicit in his position nonetheless.

Lakoff claims that a variety of traditional categorization schemes are false. This is said to be because such schemes are grounded in the principles of objectivist metaphysics and a host of associated views. To detail the alleged difficulties, it is necessary to quote Lakoff at some length.

OBJECTIVIST METAPHYSICS: All of reality consists of entities which have fixed properties and relations holding among them at any instant.

Objectivist metaphysics is often found in the company of another metaphysical assumption, essentialism.

ESSENTIALISM: Among the properties that things have, some are essential; that is, they are those properties that make the thing what it is, and without which it would not be that kind of thing. Other properties are accidental-- that is, they are properties that things happen to have, not properties that capture the essence of the thing.

The classical theory of categories relates properties of entities to categories containing those entities.

CLASSICAL CATEGORIZATION: All the entities that have a given property or collection of properties in common form a category. Such properties are necessary and sufficient to define the category. All categories are of this kind.

In the standard view, every entity either does or does not have a given property. As a result, categories have well- defined rather than fuzzy boundaries. However, one could extend the standard objectivist position to allow a category to have a property to some degree....

Given that properties have objective existence, and that properties define categories, it can make sense to speak of categories as having objective existence.

THE DOCTRINE OF OBJECTIVE CATEGORIES: The entities in the world form objectively existing categories based on their shared objective properties.

If one adds essentialism, one can distinguish a special kind of objective category--one based on shared essential properties, as opposed to shared incidental properties.

THE DOCTRINE OF NATURAL KINDS: There are natural kinds of entities in the world, each kind being a category based on shared essential properties, that is, properties that things have by virtue of their very nature. (Lakoff, 1987, pp. 160-161)

Armed with these relatively unobjectionable definitions, Lakoff goes on to argue that perfectly ordinary taxonomies--such as the zoological, botanical, and mineralogical--disprove not only this view of categorization, but the underlying metaphysics itself. This stunning conclusion, however, is made possible only by sliding from a fairly weak objectivist claim to a very strong one that is not entailed by the weaker one, and is held in the strong form by no important current school of thought.

For instance, as cited above, Lakoff says that, "objectivist metaphysics is often found in the company of another metaphysical assumption, essentialism" (p. 160, italics added). This is true, assuming a reasonable explication of "often." Later in the book, however, he claims that, "the concept of natural kind plays an absolutely crucial role in objectivist metaphysics" (p. 187, italics added). This dramatic increase in the strength of the claim is neither justified nor, as far as we can see, justifiable. Lakoff's slide from the weaker position that objectivism is "often found in the company of" essentialism to the stronger position that essentialism is "absolutely crucial" to objectivism seems to stem from the belief that, under objectivism, there must be one and only one correct classification scheme for a given domain (p. 186). Hacking (1991a) has called this the "uniqueness principle" (UP). It is true that if one were not only possessed of the belief that there are natural kinds, but also that our knowledge is imperfect unless it reflects all and only these natural kinds, then one would indeed be committed to the UP, and Lakoff's argument would run. This is hardly the modal view of objectivists, however. Belief in the UP is a decidedly optional part of the natural kind armamentarium, and one that only few advocates of natural kinds actually espouse (see Hacking, 1991a).

One might believe, for instance, that there are natural kinds, but that they are neither the exclusive nor the exhaustive route to knowledge. In such a case, one would not be at all committed to the belief that there is one, and only one, correct categorization scheme. Russell (1948) and Quine (1969), for instance thought that natural kinds are a reasonable place to begin science, but that science seeks precisely to refine and recast these distinctions as it matures.

To make matters even worse (from Lakoff's perspective), the belief in natural kinds is independent of the belief that there are necessary and sufficient conditions for such kinds. That is, essences, even if they exist, need not be easily articulable (As noted by Hacking, 1991b; Boyd, 1991; Keil, in preparation).

Further still from Lakoff's suggestion, one could perfectly consistently hold that categories are objective, but that there are no natural kinds at all. For instance, one could claim that there are facts about reality that make picking out tigers something other than personal arbitrary choice--i.e., others could equally come to the same categorization because of facts about the world--but that does not mean that this objective division of the world exemplifies or exhausts all of its facts. Other divisions, equally epistemologically objective, could reveal other salient facts about reality. Knowledge obtained would be relative to the conceptual scheme used, but the facts revealed would not be epistemologically dependent on human cognition for their existence. Thus, again, only the weaker of Lakoff's claims about the relation between objectivism and natural kinds could plausibly be considered true, and even that depends on how one explicates the phrase, "is often found in the company of."

Even more unsettling than Lakoff's conflation of objectivism and the doctrine or natural kinds, however, is his apparent belief that categorization schemes themselves can have truth values. One of the most conspicuous examples comes in his analysis of biological taxonomy, where he says,

Classical categories and natural kinds are remnants of pre-Darwinian philosophy. They fit the biology of the ancient Greeks very well, and even the biology of local naturalists such as Linnaeus. But they do not accord with the phenomena that are central to evolution--variation within species, adaptation to the environment, gradual change, gene pools, etc. Whatever one's choices are in the styles of contemporary biology, objectivist semantics and cognition and, to a large extent, even objectivist metaphysics are in conflict with post-Darwinian biology. I'd put my money on biology. (p. 195)

Categorization schemes are nothing more than what are usually called "conceptual schemes" in the philosophical literature (Davidson, 1973-74). Conceptual schemes, in and of themselves, are not claims about the world, they are just ways of dividing the things of the world, both real and imagined, into categories. They cannot, therefore, be shown by empirical means to be true or false. Consider, for example, the category "tyrannosoup". A tyrannosoup is anything that is either a Tyrannosoarus rex, or a can of soup. This is a perfectly good classical category. It has necessary and sufficient conditions. It even has an essential feature, though the feature is disjunctive. We doubt, however, that anyone would think it a natural kind. Notice that in itself it is neither true nor false. It simply divides the world into those things that are a Tyrannosaurus rex or a can of soup, and those things that are not. A contigent proposition containing the category, such as, "There are tyrannosoups," may be true or false (this one happens to be true), but even the falsity of this proposition would not somehow falsify the category itself; it would just imply that it is empty. Empty categories are not false, they are just empty. It is important to make this distinction. Some empty categories are very important to us; e.g., the category of people who have been to Mars, the category of people who start World War III. We have to work very hard to either fill, or maintain as empty, such categories.

To return to biological taxonomy, Lakoff argues that the competing claims of all three major theories of species are false because none of the schemes fit all that is known about species. These three positions are cladism, which holds that species are identified by their evolutionary histories, pheneticism, which holds that species are identified by similarities in physical appearance, and evolutionary systematism, which attempts to integrate the claims of both cladism and pheneticism. More important for Lakoff, however, is the belief that the argument among these three is moot because there need not be one and only one definition of species. Only a person wedded to the doctrine of natural kinds would care to argue over which of these approaches gives the "one true" explication of what a species is. If one abandons the doctrine of natural kinds, one need not accept only one approach to biological taxonomy. One can see that they all have merit.

Lakoff may be absolutely correct about this. Where Lakoff is incorrect is where he takes this to be a crucial nail in the coffin of objectivism. As argued above, one's position on objectivism implies nothing about one's position on natural kinds. One can simultaneously be a radical objectivist, and a radical opponent of the doctrine of natural kinds perfectly consistently. Such a person would be able to accept the relative merits of each of the approaches to biological taxonomy without having to give up objectivism.

But what sort of objectivism would it be? How can an objectivist claim to see the merits of various approaches to a problem, and still live up to what Lakoff calls the "spirit" of objectivism? Here again we see the influence of the equivocation on "dependence" discussed above. Recall that objectivism need only require that the facts be epistemologically independent, not ontologically independent. That is, it need not be the case that there is a single true definition of species; it need only be true that the definition one uses be open to criticism and application by others. Whether one is a cladist, a pheneticist, or an evolutionary systematist, one's scheme is still epistemologically independent, and therefore objective.

So, the question turns to how an objectivist can account for the variability of not only concepts such as "species", but also of Lakoff's favorite examples, such as "mother" and "balan". As we have claimed, both the cladist and pheneticist explications of "species" are perfectly objective. The question remains, "whence did they come?" Lakoff argues that they represent different "folk theories" of biological taxonomy. In a sense, we agree with him, but we do not believe this has the profound consequences he does. When we began classifying animals it is likely that we divided them into various types according to various perceptual and functional characteristics. Among the perceptual characteristics, we noticed that some could fly, some could run, some could hop; among those that could hop some had long ears and short tails, some had short ears and long tails, etc. Among functional characteristics, some could be eaten, some could eat us, some fell into both categories, some fell into neither. One of the characteristics that was noticed was the seeming relatedness of animals to each other. This relatedness may have initially been primarily perceptual (i.e., certain animals "looked" alike), but there was undoubtedly the suspicion that it indicated familial similarity as well. Thus, a relatively primitive notion of species may have developed. Though a reasonable guess that worked in many instances, this assumption turned out not to be completely true, and thus the concept of species became differentiated into two explications--one along lines of ancestry (cladism) and one along the lines of similarity of appearance (pheneticism).

Lakoff would have us believe that the simple fact of cross-categorizability--the fact that we can classify an animal both by its means of locomotion and its edibility--falsifies objectivism. We see no merit to this at all, unless one awalsely) believes that all objectivists are exclusively and exhaustively wedded to an essentialist doctrine of natural kinds, as Lakoff does.

We believe the differentiation model of concept development, outlined above, to be essentially sound, however. Consider the concept "mother". Lakoff argues that "mother" is a "radial concept," which means that there is a core concept of mother--something like a stereotypical mother--surrounded by a set of specialized "mother" concepts, corresponding to things such as the "genetic" mother", the "birth mother", the "nurturing mother", etc. Radial concepts such as these are important to Lakoff's position because he invokes them as evidence against traditional classification schemes. By contrast, we see the generation of specialized mother concepts not as a refutation of traditional approaches to categorization, but rather, as nothing more than straightforward cases of differentiation in the face of new experience. To elaborate a little, at first most people have a relatively undifferentiated concept of mother. Since the same person serves in the genetic, birthing, nurturing, etc. mother roles, there is no need for it to be differentiated. As one meets, or hears about, an adopted child, the question arises as to whether the "biological" or the "nurturing" mother is the "real" one. The question is resolved--or, rather, dissolved--by differentiating the concept of mother into these two types. (The argument over which one is the "real" mother would be as pointless as the question of which is the "real" species.) Probably later, one finds out about surrogate mothers, and wonders whether the woman who donated the ovum, or the one who carried the fetus to term is the "real" biological mother. Again the notion is differentiated into "genetic" and "birth" types. The process continues as one finds more separable functions that were originally unreflectively attributed to mothers. Although Lakoff might claim that this account is compatible with his, it is important to note that there is nothing here that refutes objectivism. Each new differentiation is defined in epistemologically objective terms, and can be studied in its own right perfectly objectively and scientifically. There are no radical consequences for theories of cognition, semantics, metaphor, or metaphysics at all.

Next we come to the question of the categories derived from embodied experience that are central to Lakoff's main project. If objectivism is not refuted by institutional facts, and objectivist classification schemes are not made false by the implications of natural kinds, by empirical facts, or by the existence of interesting conceptual structures, perhaps the admittedly important facts of embodied experience can show objectivism to be inadequate. We argue that the answer to this supposition is "No." As with institutional facts, we believe that the facts of embodied experience can be rendered epistemologically independent, even if they are inherently ontologically dependent.

Consider, for instance, the concept of heat (the experience, not the molecular explication). The concept of heat (and its opposite, cold) are defined relative to our sensory apparatus. If we were built differently, we might consider 0° C. to be hot, or 100° C. to be cold. In fact animals from extreme climates (e.g., tropical birds and fish) seem to have different criteria for hot and cold than humans. Lakoff believes that facts such as these show objectivism to be false, but they do not. Although the concept of heat may be relationally defined, it is not non-objective for this reason. As there was an equivocation on the notion of independence before, so there is one on that of subjectivity here. The sensation of heat is subjective inasmuch as it can only be directly felt by a single person. That is, one cannot share his or her experience of heat with another person (N.B., even if the two people are both touching the same hot object, their respective experiences of heat are distinct), but it is still an objective fact of the world that the relation between a state in the physical world and one's bio-psychological constitution gives rise to the purely subjective aspects of the experience. Lakoff seems to have conflated the claim that a concept is dependent on a relation with respect to a particular subject with the claim that the concept is epistemologically subjective. Thus, categories derived from embodied experience are objective enough for traditional science.

The structure of our concepts is said to derive from the "preconceptual" structure of our experience; i.e., of our interactions with the things of the world. Before moving on to examine the ramifications of this view, we feel it is important to make a preliminary point. Lakoff claims that his experientialism is a form of internal realism, à la Putnam (1981) (p. 265). Yet he also states that a crucial component of his view is the belief that meaning is ultimately dependent on the preconceptual structuring of experience (p. 271). But in Putnam's explication of internal realism, which Lakoff quotes, Putnam explicitly states that internal realism denies that there are any experiential inputs "which are not themselves to some extent shaped by our concepts" (p. 263, Putnam's italics). We find it difficult to reconcile that claim of the priority of the preconceptual structuring of experience with Putnam's denial of any inputs that are not structured to some degree by our concepts. Lakoff seems here to have stepped into a full-fledged contradiction, and must explain how these various elements of his position hang together.

As we have said, Lakoff's position involves the claim that the structure of concepts is dependent on the preconceptual structuring of experience. However, Lakoff repeatedly argues that the "real" world has no inherent conceptual/categorical structure. How, then, could our interactions with it have structure? If this structure is derived exclusively from our bodies, then it is innate structure. If there really were a sharing or intermixing or integration of bodily and worldly structures that gives rise, ultimately, to our conceptual structures, then the world itself must be structured prior to our interactions with it, so that it can contribute to, or at least constrain, the structure of our experience and, ultimately, of our concepts. At this point, it seems that the incipient dualism implicit in Lakoff's position, discussed above, begins to shade off into a kind of full-blown idealism, his stated commitment to a "real" world--called "basic realism"--notwithstanding.

Having brought these preliminary considerations to light, we now move specifically to the issues of semantics. Lakoff states that the fundamental principle of semantic theory--the criterion that all semantic theories must meet--is that there cannot be a change in the parts of a sentence without there being a resultant change in the meaning of the whole (p. 230). A theory that fails to meet this criterion is said to fail as a theory of meaning. To show that traditional truth-conditional theories of meaning actually violate the fundamental criterion, Lakoff then makes use of an argument about a problem with linguistic reference found in Hilary Putnam's Reason, truth, and history (1981).

Briefly, what Putnam shows is that it is possible to construct two sentences that share exactly the same truth conditions, yet whose component parts obviously mean different things. By the truth conditional theory of meaning such sentences should have the same meaning, even though the component parts have radically different meanings. Hence the fundamental criterion is violated. Lakoff thus concludes that any attempt to base meaning on truth, or on any other structurally defined notion, is doomed to failure (p. 256). He claims that his theory of meaning, based on what he calls "idealized cognitive models" or ICMs (see Vervaeke & Green, in press), does not fall into this trap, but never explicitly shows this to be true.

Consider the following example. The Mohawks are said to divide their nation into two groups: the animal clans and the bird clans. Thus, by "composing" one's cognitive model of the animal clans with one's cognitive model of the bird clans, one should get a model of the whole Mohawk nation. It is important to note that here Lakoff gives up strict compositionality for a looser "culturally motivated" compositionality.

Consider, now, that the Mohawks regard deer as part of the domain of birds, rather than that of animals. Thus their model of the bird clans will differ from ours, and their model of the animal clans will differ from ours. However, our combined models --representing the whole of the Mohawk nation--will be identical. Thus, if Putnam is right about the fundamental principle of semantic theory (viz., that there cannot be a change in the parts of a sentence without there being a resultant change in the meaning of the whole) then Lakoff's cognitive model theory can be no more adequate an account of semantics than the formal model theory, for it equally violates the fundamental principle.

Lakoff might respond that the two composite models (the Mohawk's and ours) are not identical (i.e., do not have the same meaning) because they do not contain the same division between clans. If he takes this tack, however, his statement boils down to nothing more interesting than the claim that similar wholes containing different parts are still different from each other just in that they contain different parts. This is just a tautology. In any case Lakoff correctly points out that simply fixing reference will not solve the problem (p. 238). If so, there is no reason to believe that fixing the grounding of terms in particular ICMs fares any better. Thus, if the "fundamental criterion" really is fundamental, Lakoff's theory of semantics fails exactly where he claims traditional theories of semantics are said to fail as well.

We have tried to show that Lakoff's critique of objectivism contains some important conceptual confusions. We have also tried to show that the experientialist alternative he proposes does not seem to be an adequate alternative. It seems to be right just to the extent that it is subsumable under a suitably sophisticated version of objectivism. Where objectivism seems to fail, however, experientialism seems to fail as well. That is, it does not escape Lakoff's own fundamental criticisms of objectivist theories of cognition, meaning, or metaphysics. Thus, we conclude that although Lakoff has directed attention to aspects of cognition, semantics, and metaphysics that require intensive investigation, he has failed to show that their traditional recalcitrance has any fundamental or specific connection with objectivism.

Berlin, B. & Kay, P. (1969). Basic color terms: Their universality and evolution. Berkeley: University of California Press.

Boyd, R. (1981). Realism, anti-foundationalism and the enthusiasm for natural kinds. Philosophical Studies, 61, 127-148.

Brown, R. (1958). How shall a thing be called? Psychological Review, 65, 14-21.

Davidson, D. (1973-74). On the very idea of a conceptual scheme. Proceedings and Addresses of the American Philosophical Association, 67, 5-20.

Ekman, P. (1971). Universals and cultural differences in facial expressions. Lincoln: University of Nebraska Press.

Hacking, I. (1991a). A tradition of natural kinds. Philosophical Studies, 61, 109-126.

Hacking, I. (1991b). On Boyd. Philosophical Studies, 61, 149- 154.

Keil, F. (in preparation). Coming to explain the properties of living things -- A case study of how we come to have causal understanding of natural kinds.

Kay, P. & McDaniel, C. (1978). The linguistic significance of basic color terms. Language, 54(3), 610-646.

Lakoff, G. (1987). Women, fire, and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press.

Putnam, H. (1981). Reason, truth, and history. Cambridge: Cambridge University Press.

Quine, W. V. O. (1969). Natural kinds. In Ontological relativity and other essays. New York: Columbia University Press.

Rosch, E. (1973). Natural categories. Cognitive Psychology, 4, 328-350.

Rosch, E. (1975). Cognitive representations of semantic categories. Journal of Experimental Psychology: General, 104, 192-233.

Rosch, E. (1978). Principles of categorization. In E. Rosch & B. B. Lloyd (Eds.) Cognition and categorization. Hillsdale, NJ: Lawrence Erlbaum.

Russell, B. (1948). Human knowledge: Its scope and limits. London: Allen & Unwin.

Vervaeke, J. & Green, C. D. (in press). Women, fire and dangerous theories: A critique of Lakoff's theory of categorization. Metaphor and Symbolic Activity.

Correspondence concerning this paper should be addressed to either Christopher D. Green, Department of Psychology, York University, North York, Ontario, CANADA, M3J 1P3 (e-mail:christo@yorku.ca, WWW: http://www.yorku.ca/dept/psych/ people/faculty/cgreen/homepage.htm) or to John Vervaeke, Department of Philosophy, University of Toronto, Toronto, Ontario, CANADA, M5S 1A1 (e-mail: jvervaek@epas.utoronto.ca).

This paper was written in part while the first author held an Individual Research Grant from the Natural Sciences and Engineering Research Council of Canada. We would also like to acknowledge the assistance many members of the University of Toronto Department of Psychology, who discussed and debated these issues with us at some length. Most notable are Dan Chiappe, Michael Gemar, and especially Philip Groff who provided us with the example concerning Mohawk clan organization.