Nelson Goodman
Nelson Goodman; drawing by David Levine

A little over half a century ago philosophers and psychologists at Harvard shared a single department housed in Emerson Hall, its ground-floor corridor presided over by a seated, slightly frowning Ralph Waldo Emerson in bronze. The psychologists by then could scarcely wait to be free of their old-fashioned colleagues. That came soon enough. For what had psychology, armed with its shiny new tools of empirical inquiry, to do with metaphysics? Did one need philosophical speculation to study sensation, perception, and behavior?

Psychology even then had embraced the physicist Percy Bridgman’s “operationalism,” a philosophical position holding that scientific concepts, such as mass or “mind,” could only be defined by the experimental methods or “operations” that were used to establish their application to things or processes. Where psychologists were concerned, that was as much philosophy as was needed. IQ, accordingly, was simply what intelligence tests measured. Viennese logical positivism, moreover, defined the line between philosophy and psychology. According to its principles, only statements that were true by definition (as in logic or mathematics) or were “empirically verifiable” were meaningful and worthy of study. For the logical positivist, philosophers should be concerned with the “analysis” of concepts in ordinary language (such as “justice”) and the sciences (such as “gravity”); such analyses would clarify these concepts by offering improved definitions of them. The sciences—among them psychology—should investigate only statements that are “empirically verifiable.” The rest, issues in logic and language, are treated, in this view, by philosophy.

After World War II, the chief common concern of psychology and philosophy was “methodology”: like the businessman and his accountant, the philosopher told the psychologist the right way to talk about the numbers he had produced in experiments on the reactions of rats, say, or in compiling the results of intelligence tests. “Mind” remained a forbidden four-letter word in mainstream psychology, to be dealt with (if at all) in quotation marks. The “methodology” of “scientific psychology” grew stricter by the year; its terms and concepts had to have an objective basis and had increasingly to conform to stringent rules for defining scientific concepts by experimental procedures. Methods became a preoccupation: methods for making the subjective objective, the hidden overt, the abstract concrete.

Then, in the late 1950s, came what today is called the “cognitive revolution.” Psychologists like Herbert Simon and George Miller, and linguists like Noam Chomsky, devoted themselves not to their subjects’ overt, objective responses, but rather to what they knew and how they acquired knowledge. The emphasis shifted from performance (what people did) to competence (what they knew). And this inevitably led to the question of how knowledge was represented in the mind. Could you simulate what the mind knew with a computational program (as Simon was attempting to do) or with a theory of mental organization (as Piaget was doing)? “Mind” was being reintroduced into psychology and defined variously as ways in which knowledge was organized, or as a set of strategies for using knowledge in order to achieve intended results, and so on.

When Herbert Simon demonstrated that you could construct a computer program that could solve theorems in Whitehead and Russell’s Principia, or that could get the missionaries across the river safely in the puzzle about “Cannibals and Missionaries,” it was altogether appropriate to ask whether and in what degree these programs themselves worked like the mind. And if these programs presupposed that knowledge was represented in the mind, what sorts of representations were these and what was the mind that contained them? Is such knowledge organized according to the specific intentions that led to acquiring it, or is it an all-purpose and general knowledge? Moreover, how did we, or how did a program representing “us,” come to have its knowledge of the world? The new questions came in torrents, questions now interesting to both philosophers and psychologists.

The new cognitive psychology declared that the particular choice—the mental act—that guides an action is as real as the physical action that ensues; the principles underlying the choice therefore needed to be explained if mental actions were to be understood. But while overt actions can be observed and counted, the thoughts and rules that guide them are not “objective” in this sense. They are private. And there was the difficulty. Is a science of thinking not a science until it can be made “objective”? Or, to put it another way, was a philosophy of science that demanded such objectivity the only possible or correct one? Anglo-American philosophy of science, derived from writers like Rudolf Carnap and Carl Hempel, had taken nineteenth-century physics—not psychology—as its standard of “good science”; it had insisted that whatever is alleged to exist must be shown to be physical or, at least, reducible to what is physical. But the increased attention that psychologists were giving to cognitive processes—to such hard-to-objectify processes as thinking and knowing—challenged this position. So, after more than half a century of independence, psychologists began meeting with philosophers on something more than a social basis; the problems that the new psychology posed did not fit comfortably with the old philosophy of science; indeed they demanded a new philosophy of science.


Nelson Goodman is one of the major philosophers in the world today who address themselves to the solution of this tangled set of problems. His recent book, Of Mind and Other Matters, takes up where two earlier books, Ways of World-making and Languages of Art, left off and sets forth detailed replies to criticisms of these two earlier books.

Of Mind and Other Matters defends what Goodman calls a “constructivist” philosophy. It is at once a philosophy of science, a philosophy of art, and a philosophy of cognition—he ends by calling it “a philosophy of understanding.” Its central thesis, “constructivism,” is that, contrary to common sense, there is no unique “real world” that preexists and is independent of human mental activity and human symbolic language. For Goodman what we call the world is a product of minds whose symbolic procedures—whether in using words or numbers or pictures—construct that world. He argues that for some forms of mental functioning, as in perception for example, we already know a great deal about how mental processes do the work of construction:

The overwhelming case against perception without conception, the pure given, absolute immediacy, the innocent eye, substance as substrate, has been so fully and frequently set forth—by Berkeley, Kant, Cassirer, Gombrich, Bruner, and many others—as to need no restatement here.

The world of appearance, the world we live in, is something we ourselves create through the mind. The activity of “world-making” is, for Goodman, a diverse and complex set of activities of construction; however else it may express itself it involves “making not with hands but with minds, or rather with languages or other symbol systems.”

The “world” we create, he says, may arise from the cognitive activity of the artist (the “world” of Joyce’s Ulysses) or the scientist (whether the geocentric world view of the Middle Ages or that of modern physics) or the ordinary man (as in the common-sense world of trains, kitchens, and automobiles). Most people would claim that the world of modern science is privileged—that, for example, from astronomical observations and other procedures we know that the earth moves around the sun, not the other way around. Goodman disagrees. Once we give up the idea of “the pure given” and the innocent eye that sees things exactly as they are without “interpreting” them, we must conclude that the astronomer’s observations are dependent on the concepts he uses—such as “star,” “sun,” “orbit”—and that the world of stars he describes was itself created by our minds. Goodman writes that just as we

make constellations by picking out and putting together certain stars rather than others, so we make stars by drawing certain boundaries rather than others. Nothing dictates whether the skies shall be marked off into constellations or other objects. We have to make what we find, be it the Great Dipper, Sirius, food, fuel, or a stereo system.

The astronomer’s observations are highly coherent and seem to “work”—we use them to help us guide satellites. But Goodman says that it would be a mistake to hold that the world of astronomy is more “real” than all others or that it is the unique real world. In fact, he argues, the physical raw stuff of the astronomer is no more “real” than the figures depicted by Picasso or the characters of Ulysses; all of these are if anything less real than the psychological processes that created them. All of the “worlds” we deal with, he says, have been constructed out of other worlds that we have taken for granted. We are never in contact with some sort of aboriginal reality independent of our own minds or the minds of others who preceded us.

The constructivist view that what exists is a product of what is thought, was first worked out by Kant. Kant, in turn, said that his view owed much to Hume’s discovery that certain relations among things in the real world could not be attributed to events but rather were mental constructions projected onto an “objective world.” His principal case rested on the relationship of cause and effect. Hume had seen that causation was a mental construct imposed on a mere sequence of events. Goodman starts from Kant’s view that the world “out there” is made up of products of the mind.

But as we have already noted, Goodman refuses to assign any privileged status or any “ultimate reality to any particular world that mind may create.” Kant, on the other hand, argued that we all have certain knowledge, a priori, by virtue of having human minds. Such a priori knowledge, on Kant’s view, precedes all reasoning. In place of Kant’s a priori, Goodman offers a more relativistic notion. We do not begin with something absolute or prior to all reasoning, but, according to Goodman, begin instead with the kinds of constructions that lead to the creation of worlds. And these constructions have in common that they take certain premises for granted, as stipulations. What is “given” or assumed at the outset of our construction is neither bedrock reality “out there” nor a priori knowledge: it is always another constructed version of a world that we have taken as given for certain purposes. Any previously constructed version of a world may be taken as given for subsequent constructions. So, in effect, world-making involves the transformation of worlds and world versions already made.


Obviously, the idea of mind as an instrument of construction is (or should be) congenial to the developmental psychologist who observes that children assign different meanings to the same “event” at different ages. The clinical psychologist must always be impressed with the “reality” with which patients endow their rich and various narratives. And constructivism is nowhere more active than in the psychology of art and creativity. Blake, Kafka, Wittgenstein, and Picasso did not find the worlds they produced. They invented them.

The notion of stipulation that we find in Goodman’s work—of taking something as given—is also richly suggestive for cognitive psychologists. One immediately thinks of the importance of mechanisms like “recursion,” the process whereby the mind, or a computer program, loops back on the output of a prior computation and treats it as a given that can be the input for the next computation. Theories as divergent as Chomsky’s theory of grammar, Piaget’s account of the development of mental functions, and Simon’s idea of a “general problem solver” all depend on some process of recursion. Any formal theory of mind is helpless without recursion, for it is impossible without it to account for thoughts on thoughts, thoughts on thoughts on thoughts, up to whatever level of abstraction is necessary. Indeed, Philip Johnson-Laird, in his excellent Mental Models,* invokes the concept of recursion to account for how the mind turns around on itself to create the kind of summary of its capacities that might constitute something like a sense of “self.” One begins to get a glimmer from this work of how Goodman’s stipulations might be used in sequences, each transforming a previously created version of the world into a new one, the whole providing a basis for understanding not only single acts of cognition but also complex ones that have the look and feel of real world-making. So far, so good.

It would seem on the surface, then, that Goodman would become the immediate idol of cognitive psychologists. But though he has been taken to heart by some, others have been indifferent to his views. It is not that psychologists fail to appreciate Goodman’s stress on the active role of mind in creating “worlds”; nor do most of them doubt that we assign social reality to the pictures of a world we create. But “interpretive social science” of the kind represented, say, by Clifford Geertz in anthropology, which emphasizes the irreducibility of meaning and the creation of social reality through the construction of symbol systems, has not had much hearing in psychology. Psychologists (even cognitive psychologists) still like to think of the world that people create as “representing” a real or aboriginal world. Even Piaget, whose epistemological theory was a constructivist one—in which, as the child grows, more elaborated constructs encompass simpler ones—clung nonetheless to a residual naive realism. Constructions for him were representations of an autonomous real world which they had to match or “accommodate.”

Once we give up the idea of an aboriginal reality, we lose the criterion of correspondence between statement or hypothesis and “reality” as a way of distinguishing true from false models of the world. Under these conditions, what can protect us from the relativism that threatens to follow? Goodman rejects the standard alternative to a correspondence criterion of truth, a “coherence” theory of truth according to which a statement is true if it “fits” or coheres with the system of other statements that we take for granted. The radical relativism that a coherence theory implies is as unacceptable to him as it is to his critics. He writes:

We must obviously look for truth not in the relation of a version to something outside that it refers to but in characteristics of the version itself and its relationships to other versions….

When the world is [given up] and correspondence along with it, the first thought [of a way to establish truth] is usually coherence. But the answer cannot lie in coherence alone; for a false or otherwise wrong version can hold together as well as a right one.

The important point is that, for Goodman, some versions that hold together coherently are not true. Moreover, he claims, some truths conflict. For example, when we step off a moving bus, the earth we step onto is not moving. On the other hand, according to modern physics, the earth is moving. How can we reconcile such conflicting truths? “Usually,” he says,

we seek refuge in simple-minded relativization: according to a geocentric system the earth stands still, while according to a heliocentric system it moves. But there is no solid comfort here. Merely that a given version says something does not make what it says true; after all, some versions say the earth is flat or that it rests on the back of a tortoise. That the earth is at rest according to one system and moves according to another says nothing about how the earth behaves but only something about what these versions say. What must be added is that these versions are true.

Goodman claims he can formulate a criterion (or criteria) that will tell us what makes some world versions right (or true) and others not. This is no small order, and Goodman spends much of his new book trying to fill it.

To put his claim simply, right versions are versions that are true in some world. It is a right version of the everyday geocentric world that the earth is at rest when we step off a moving bus, though in the world of modern physics it is also a right version that the earth is always moving. But just as there are many right versions, each true in some world, there are many worlds. Now we have a kind of correspondence theory of truth in which every right version has a world corresponding to it. But unlike the usual correspondence theory, this one permits conflicting alternative accounts because apparently conflicting versions may each be true in some world, and there are many worlds. Or as Goodman puts it, since “there are conflicting true versions and they cannot be true in the same world,” there must be many worlds. These worlds do not occupy the same space or time. “In any world,” he says, “there is only one Earth”; and the several worlds—the world into which we step off the bus and the world of the astronomer—do not risk collision in the same space-time. Indeed, “space-time is an ordering within a world; the space-times of different worlds are not embraced within some greater space-time.” These plural worlds cannot be reduced by some maneuver to some single world, even to that of modern physics.

So Goodman has offered us a criterion that makes a version right if it is right in some world. But even if we are prepared to live with Goodman’s many right versions, has he really given us a workable means for distinguishing right versions from wrong ones? The answer depends upon whether one can clearly distinguish a version from the world of which it is true or false. At some points in his argument, Goodman goes a long way toward establishing the claim that versions and worlds can be kept distinct and checked for their correspondence with each other. He remarks, for example, that a “world is not the version itself; the version may have features—such as being in English or consisting of words—that its world does not.” Or again, “a version saying that there is a star up there is not itself bright or far off, and the star is not made up of letters.” This suggests that versions exist independently of the world they are versions of.

But then, at other points, he denies that worlds and versions can be distinguished. He says, “We make versions, and right versions make worlds. And however distinct worlds may be from right versions, making right versions is making worlds.”

Goodman wants to have it both ways; and this is the most puzzling part of his argument. It puzzled the philosopher Israel Scheffler, too. Goodman answers Scheffler on this point as follows: “We can have it both ways. To say that every right version is a world and to say that every right version has a world answering to it may be equally right even if they are at odds with each other”—i.e., at odds in the sense that the physical star is not made up of letters. We can “acknowledge,” he says, that a right version and its world are “different” in this sense. But the world in question is not independent of versions of it:

The objects themselves and the time and space they occupy are version-dependent. No organization into units is unique or mandatory, nor is there any featureless raw material underlying different organizations. Any raw stuff is as much the creature of a version as is what is made out of that stuff.

We may take it for granted that we speak on the one hand of linguistic versions of a world or, on the other, of worlds themselves (what the versions refer to). But for Goodman, we make this distinction as a matter of convenience or convention, not as a decision about objective fact.

Since it does not provide a universal criterion for determining a right version, Goodman’s canon of convenience and convention must remain ambiguous in practice, and it may well, alas, be an ambiguity that is built into any thoroughgoing constructivism. Nevertheless, it is a useful beginning. For despite its metaphysical ambiguity, his claim that we construct new worlds by applying symbol systems to a stipulated or “given” world is, from the point of view of cognitive psychology, correct, even if it does not yet provide a satisfactory constructivist philosophy of science. Interpreted in this way, Goodman’s views have powerful consequences for our understanding of such products of the human mind as scientific theory, art, and cognitive activity generally.

That is what much of Of Mind and Other Matters is about (as was Ways of Worldmaking before it). Let us take first Goodman’s efforts in the philosophy of science. Should we give up the idea that all the terms of science can be reduced to descriptions of physical things in favor of a thoroughgoing constructivism? As W.V. Quine remarked in these pages in his review of Ways of Worldmaking (November 23, 1978), physical theory is “ninety-nine parts conceptualization to one part observation,” and that makes “nature” as the physicists perceive it a poor candidate for the “real” world. Indeed, the intellectual vigor of modern physics is precisely its sensitivity in choosing appropriate theoretical descriptions to interpret particular observations. Some may claim that because he refuses to call any world or construction more “real” than all others, Goodman cannot capture within his philosophy the widespread belief that the theoretical constructions of modern science are uniquely successful in providing us with mastery over natural events. His pluralism seems to reduce science to the same level as any other “right” construction, whether of philosophy or painting.

This is to misinterpret Goodman’s intent. As we have seen, he does not deny that physics is superior to other constructions in explaining the behavior of atoms or the solar system. What he is urging, rather, is that we ask the hard but inevitable questions about the mental operations needed in constructing a world like that of modern physics or of everyday life. And then once physics is “entrenched”—becomes widely accepted and familiar—we may ask how it works as mental construction in making sense of the particular world it takes as given. This is what Abraham Pais did in his biography of Einstein, what Piaget did for the child’s conception of the world, and what Howard Gardner does in his efforts to understand children’s drawings.

The same goes for versions of the world created by the artist, the novelist, the patient in therapy. For Goodman, it will be clear by now, is a philosopher of mind who believes that science and art grow out of certain common constructional activities. They are guided in each case by different constraints for establishing rightness and different conventions that grow out of their “entrenchment.” The difference for him is not that the arts are “subjective” and science “objective.” Rather, each constructs its world differently and they cannot be distinguished correctly by calling one subjective and the other objective.

What is at issue, he proposes, is the difference in the constructional activities of the various arts and sciences, and particularly differences in the use of what he calls “symbol systems.” Goodman has devoted much effort to developing a theory of symbols, the most mature expression of which is to be found in his Languages of Art. In that book, as in the present one, he develops the proposition that “much of knowing, acting, and understanding in the arts, the sciences, and life in general involves the use—the interpretation, application, invention, revision—of symbol systems.”

The central notion of his theory of symbols is that of “reference.” As he puts it, reference is a term “covering all sorts of symbolization, all cases of standing for.” There are literal and nonliteral modes of reference, with simple and complex forms that provide a range and a subtlety that can be exploited in the making of worlds in both science and art. Even in the case of verbal denotation—naming, describing, predicating “where a word or string of words applies to one thing, event, and so on, or to each of many”—reference is dependent on context (as with words like “here” and “now”). Indeed, reference may be more or less vague, more or less dependent on the nature of the discourse in which it is embedded. Even in the allegedly simple case of pictorial denotation that takes resemblance to a physical object as its aim, Goodman says: “Resemblance is heavily dependent on custom and culture, so that whether and to what extent a symbol is ‘iconic,’ or faithfully depicts its subject, may vary without any change in the symbol.” The meaning of the symbol is given by the system of meanings in which it exists. A line can be the richly descriptive line that represents a hill in a landscape drawing, or the line standing for temperature on a thermometer in a system that lacks this discriptiveness.

Each system of symbols has its referential properties: fictive, figurative, and metaphoric denotations alter the referential distance they impose between a symbol and what it stands for. Hieronymous Bosch’s The Garden of Earthly Delights manages in its mode of depiction to be both fantastic and realistic. Both what is told and the mode of telling it enter into our conception of what a work of art is about. Wherever one looks at the creation of realities, one sees the complexity of symbol systems, the dependence of what they create on the discourse in which they are set and on the purpose to which the creation is to be put. Each symbol system transforms whatever input it accepts as its initial “given”: mathematics transforms the language of concrete objects in one way, poetry in another. Studying how this is done in such diverse activities as painting and literary interpretation and science is Goodman’s recommended program for the philosopher, a program that he proposes should supplant the false project of comparing works of art or science to a “real” world for their “truth” or “distortion.”

In the two earlier books we mentioned above, Goodman goes to some pains to explicate some of the “larger scale” ways in which worlds are made from previous versions. We compose and decompose worlds, impelled by different aims in doing so—practical as well as theoretical. We weigh and emphasize features of previous worlds in creating new ones, and “what counts as emphasis, of course, is departure from the relative prominence of the several features in the current world of our everyday seeing.” We impose order, and since all is in motion, the order or reordering we impose is also a way of imposing one kind of stability or another. In constructing the eight-tone modern musical scale, we supplement simpler scales, such as the five-tone scale, with extra tones; on the other hand we leave out everything that exists between C and C sharp. We deform the given that we took, and create caricature, the caricature itself being based on its own principles rather than being entirely fanciful. And we make such constructions not only in art but in science. There is, for example, a famous “map” constructed by the physiologist Lord Adrian depicting the nervous system of a monkey with each part of the body enlarged to correspond to its density of sensory stimulation—its lips and tongue in this caricature are grossly larger than its trunk and torso.

Goodman believes that there are practical consequences to the practice of philosophy, consequences particularly for the conduct of the arts and the sciences, and even for the way we carry out education. But he doubts that philosophers acting alone could get very far with these practical matters. In consequence, Goodman as much as any other modern philosopher makes common intellectual cause with artists, psychologists, film makers. There is no question that his philosophical views have had an impact on a generation of students of the arts as well as of the cognitive processes generally. He founded Project Zero at the Graduate School of Education at Harvard in 1967, where, together with others from a wide variety of backgrounds he has engaged in research on education for the arts. Howard Gardner’s Frames of Mind is a fine example of Goodman’s influence, for Gardner has been associated with Project Zero from the start and now directs it. Gardner’s effort to characterize the different modes by which intelligence expresses itself is deeply in the Goodman tradition. Characteristically, its central claim is that minds can become specialized in dealing with different forms of worldmaking—verbal or mathematical or spatial; in doing so, they are supported by symbolic systems provided by cultures which themselves construct different kinds of worlds.

Goodman puts the case well for the relevance of his views for the cognitive analysis of worldmaking through the arts:

“Cognitive” has been a battle cry in psychology and in philosophy of the arts for some decades. The movement it stands for, one of the most liberating and productive in this century, is often decried by behavioristically oriented theorists as non-empirical and unscientific, and widely thought by writers on art to be bent on analyzing the arts to death.

The trouble arises, I think, from a complex of confusions: confusion about cognition, about education, and about art and science. The cognitive approach to education for the arts must surely be condemned if cognition is contrasted with perception, emotion, and all nonlogical and nonlinguistic faculties; or if education is identified exclusively with lecturing, explaining, and providing texts and verbal and numerical exercises; or if art is looked upon as transient amusement for a passive audience, while science is taken as consisting of demonstrations founded upon observation and aimed at practical progress….

Cognition includes learning, knowing, gaining insight and understanding, by all available means…. Coming to understand a painting or a symphony in an unfamiliar style, to recognize the work of an artist or school, to see or hear in new ways, is as cognitive an achievement as learning to read or write or add….

The genuine and significant differences between art and science are compatible with their common cognitive function; and the philosophy of science and the philosophy of art are embraced within epistemology conceived as the philosophy of the understanding….

Since both science and art consist very largely in the processing of symbols, an analysis and classification of types of symbol systems…provides an indispensable theoretical background [for them both].

Goodman’s work is, in effect, a very serious effort to create, as he puts it, a philosophy of understanding. But it is a philosophy of understanding that is so pluralistic that its worth cannot be assessed fairly without considering its power in making sense of many particular worlds and symbol systems—the analysis of painting, of visual movement, of the ordering in pictorial narrative, of the structure of linguistic systems, of the creation of fictions like Don Quixote or postulational systems for defining points in space. After all, if reality is what one stipulates (rather than finds), the range of stipulation must be great, and what one makes of what one has stipulated is not something to be determined by quick intuition.

So whatever the limitations of Goodman’s proposals, he is a gadfly who has made clearer a notion of the mind not as something to be described by its properties but rather as an instrument for producing worlds. If the deep conceptual ambiguities we have noted in Goodman’s position can be sufficiently resolved to permit scientists to use some form of philosophical constructivism as a basis for research and theory, there may be Emerson Halls in the future where the mind’s constructive activities in all their variety are studied, where psychologists and philosophers (as well as artists and scientists) can appreciate one another’s work. If that happens, a bronze statue of Nelson Goodman ought to be installed in the main corridor in place of the seated Emerson, observing, coolly appreciating the varieties of construction going on around him.

This Issue

March 27, 1986