Realism can be said to be one of the major questions in the history of science and a problem that reaches far back and ramifies widely. It can at least be traced to the dispute between Plato and Aristotle, and it was also one of the main topics of medieval scholastic philosophy. In contemporary philosophy of science, as well as in epistemology, metaphysics, philosophy of mind, philosophy of religion, and so on, realism is also an inescapable theme. The definition of realism and anti-realism is likewise highly uncertain. What counts as realism from one angle may count as anti-realism from another. The treatment of realism in this material is rather too thin, or to put it bluntly, rather one-sided; it does not adequately show the complexity and richness of the topic. I will offer a few simple criticisms of this material, and then, in my own way, discuss my views on the relevant issues.
What follows is all written by me, and I take full responsibility for it. A few quotations will be cited with their sources, while some of the other background claims are simply what I have kept in mind; if there are objections, I can later add some sources. As for references, there are many. Recently, besides this English-language material, some worth mentioning are, for example, Newton-Smith’s edited The Philosophy of Science, several articles by Quine, Dummett’s The Logical Basis of Metaphysics, Goodman’s Ways of Worldmaking, and Putnam’s Realism with a Human Face, all of which I have flipped through. Also the Philosophy of Science, Metaphysics, and Introduction to Epistemology among the series of Western philosophy textbooks published by the People’s University Press; and some books on the philosophy of mathematics I read long ago, later Wittgenstein, Rosenberg’s Philosophy of Science, Roger Newton’s What Is Scientific Truth?, many popular-science books, and so on. Of course, there is also the Western Philosophy English-Chinese Dictionary, and so forth. All of the above books are worth recommending.
I will ignore the details in the material and look only at a few key points.
The first section, on “the problem of what realism is,” basically gives a fairly accurate summary of the author’s claim about scientific realism, namely: “the basic unobservable entities posited by mature contemporary scientific theories exist independently of the mind.” But one should note that this proposition rests on a very special philosophical standpoint, and that this foundation is obviously not universally accepted. We see that the author then mentions some stronger realism, which is realism about facts (fact) rather than realism about entities (entity); he says that for the sake of simplicity he will focus only on the weaker claim. The problem, however, is that the distinction between these two kinds of realism is by no means merely a matter of strength and weakness. In fact, it involves some more basic philosophical ideas, such as whether what constitutes the world is really “entities,” or “facts,” or something else. We know that Wittgenstein opens the *Tractatus Logico-Philosophicus* by saying: “The world is the totality of facts, not of things” (1.1), and that the essence of a thing lies in its being able to become a constituent part of an affair (2.011); “the existence and non-existence of states of affairs is reality” (2.06) … We need not delve into Wittgenstein’s philosophy. At the very least, we can see that in certain philosophical lines of thought, “realism about facts” is not a matter of being stronger than “realism about entities,” but rather a more basic consideration. Other approaches, such as process philosophy, make this even clearer. Such philosophies should not be simply lumped together as anti-realism or treated as stronger versions of this kind of realism.
Next the author begins with “common-sense realism.” He says: “I start with common-sense realism because, evidently, anyone who rejects it will reject scientific realism: if one doubts the independent existence of observable things, one will naturally doubt the independent existence of unobservable things.”
But this “evidently” is not evident at all. In the author’s account, it seems that doubting common-sense realism necessarily becomes extreme skepticism. In fact, that is not the case at all. To classify all idealists as extreme skeptics is obviously ridiculous. So-called extreme skepticism should mean a fundamental claim that knowledge and science in any sense of certainty are impossible; most idealists are obviously not so radical. They question common sense, but isn’t skepticism toward common sense precisely the starting point of philosophy and science? On the contrary, among the more typical skeptics in the history of philosophy—Hume, for example—it is precisely common sense (habit) that is most strongly emphasized.
In a sense, it is precisely scientific realism, rather than skepticism or idealism, that poses the greatest threat to common-sense realism. Descartes doubted his belief that he was sitting by the hearth, but in the end Descartes did not deny the reality of the hearth. Who, then, truly undermined the reality of the hearth? Precisely science! Since ancient times people have believed that “fire” is a real thing, even the most basic reality (the four elements, the five phases). Yet modern science tells you that fire is not an entity at all; fire is merely the phenomenon of a chemical reaction that emits intense light and heat. It is only because the light it gives off happens to fall in the visible spectrum, and because the reaction is sufficiently intense, that it looks to human eyes as though there really were some thing called “fire” there, when in fact there is none. So the “reality” of fire depends on human sensory capacities. Take “rainbows,” for example: we can all see rainbows clearly and unmistakably, so why not believe that a rainbow is genuinely “there”? But science tells you that there is no “one” rainbow truly out there. A rainbow is merely the refraction of sunlight, and what each person sees from a different angle is not the same “rainbow” but only the rainbow relative to that person.
In short, science as science lies precisely in its rebellion against common sense, not in its confirmation of common sense. Many philosophers have also recognized this point. Russell, for instance, said: “Naive realism leads to physics, and physics, if true, shows that naive realism is false. Therefore, if naive realism is true, it is false. Therefore, it is false.” (quoted in the *Western Philosophy English-Chinese Dictionary*, translation adjusted)
In any case, the author’s acceptance of common-sense realism and rejection of idealism are both far too hasty. I suspect that he has never seriously taken idealism—that is, the philosophy of the overwhelming majority of philosophers in the history of philosophy—into account. We see that his response to idealism is very childish. For example, he says: “…But how can we really construct dinosaurs and stars? That seems fantastic.” This sounds a bit like a middle-school textbook’s refutation of idealism. In short, idealists seem to be understood as wholly unreasonable and simply dismissed, without any serious and penetrating thought about what they actually claim.
The author then discusses the “success argument” that is mainly used to support scientific realism. Put simply, he thinks that the best explanation for the success of science is that its theoretical terms really have real referents. He says: “The world is observed to be as if it were S because it is approximately S. For example, why are all the observations we make exactly what we would get if there were atoms there? The answer is: because there really are atoms.” Later the author adds, “Assuming that the unobservable objects of science exist, we can give a good explanation of the behavior and properties of observable entities; otherwise behavior and properties remain inexplicable.”
There are quite a lot of issues involved here, some of which I will discuss later. For now, the first thing worth questioning is this: does the development of contemporary science really favor realism?
Last Friday at the Keke Forum, Bai laoshi spoke about underdetermination and hidden-variable theories. The topic was rather specialized and not very relevant to what I am saying today, and most of the students did not go, so I will not dwell on it here. But one question Bai laoshi mentioned in passing is worth bringing up—namely, why among contemporary Western philosophers of physics, are there so many supporters of Bohm’s hidden-variable theory, clearly far more than the rate of support among physicists? This is related to what we are discussing today. In my view, philosophers influenced by the tradition of logical empiricism are more likely to be drawn to hidden-variable theories, and there is some inner connection here. First of all, there is of course the attachment to “determinism,” but that is by no means all. More importantly, I suspect, is this very “realism.” Although so-called “scientific realism” was proposed only in relatively recent times, it is rooted in the tradition of logicism, that is, a certain kind of Platonic realism. This way of thinking insists that a theoretical term must have a referent, and that the object referred to is something determinate, independent of human beings, autonomous, and real. Thus “electron” refers to the object electron, and this object is objectively real: whether or not people observe it, it and its states “definitely” exist.
Quantum mechanics, however, breaks this kind of naive realism grounded in everyday experience—“electron” and “atom” are tools of theory, and as theoretical terms they only make sense when operated within the whole theoretical system; as explanations of objective reality, they can speak of some kind of truth only within an overall context that includes the operator and the process of operation. To speak independently of the reality of a particular electron is utterly meaningless.
Once one truly accepts such an anti-realist way of thinking, the paradoxes of quantum mechanics disappear without a trace. The reason quantum mechanics seems so incomprehensible is that the language of our everyday thinking is ultimately a kind of realist language. Quantum mechanics is not absurd, and it may not even be indeterministic; the key is not to search for some other substitute science, but to reflect on the limits of our language. As for quantum mechanics, if I have time later I can give everyone more of an introduction.
Apart from Bohm’s hidden-variable theory, the mainstream of the entire frontier of modern physics in fact no longer provides support for that kind of “realism.” Many heavyweight scientists have also stood on the side of anti-realism.
Take Einstein first. Although Einstein opposed orthodox quantum mechanics, he did so mainly out of a principle of sufficient reason, rather than out of realist considerations. And on the question of realism, Einstein tended toward conventionalism. Einstein repeatedly mentioned that among two possible theories, namely a priori theory and conventionalism, he chose conventionalism rather than a priori theory. In other words, he did not even consider the other doctrines. And these two, at least in this author’s classification, are clearly typical forms of anti-realism. Einstein put it very clearly: “Physical concepts are free creations of the human mind, and are not, however it may seem, uniquely determined by the external world. In our endeavor to understand reality, we are somewhat like a man trying to understand the mechanism of a closed watch. He sees the face and the moving hands, and even hears the ticking, but he has no means of opening the case. If he is clever he may form some picture of the mechanism which is responsible for all the things he observes, but he can never be quite certain that his picture is the only one which can explain his observations. He can never compare his picture with the real mechanism, and he cannot even imagine the possibility or the meaning of such a comparison. In *The Meaning of Relativity*: the world of ideas appears incapable of being logically deduced from experience and, in a certain sense, is human mind’s creation; without this creation science would be impossible. Nevertheless, this world of ideas is to a lesser degree independent of the nature of our experience, as clothes are to a lesser degree independent of the shape of our body.” (see M. Klein, *Mathematics and the Search for Knowledge*, pp. 241–242)
Bohr said: “What is it that we human beings fundamentally depend on? … We depend on our words. … Our task is to communicate experiences and ideas to others. We must continually strive to extend the range of our descriptions, … ‘Reality’ is also a word, a word we must learn to use correctly. … There is no quantum world. There is only an abstract quantum-mechanical description. It is wrong to think that the task of physics is to find out how nature is. Physics concerns what we can say about nature.”
Heisenberg wrote that some people, in a kind of euphoria, wanted to “return to the real concepts of classical physics, … or return to the ontology of materialism. … However, this is impossible. … The wish to state how atomic phenomena ought to be cannot be our task; our task is merely to know them.” Later he wrote: “In experiments concerning atomic events, we must deal with things and facts, that is, with phenomena as real as any phenomenon in everyday life. But atoms or elementary particles are not regarded as real; they form a world of potentiality and possibility rather than a world of things or facts.” In his view, “the ontology of materialism is based on illusions such as existence, and the direct ‘reality’ surrounding our world can be extrapolated into the atomic domain. However, this extrapolation is impossible.” (Bohr and Heisenberg quoted in *What Is Scientific Truth?*, pp. 179–180)
Besides the three big names above, one could cite many similar remarks from Pauli, Feynman, and others, but there is no need to list them extensively here. Of course, this does not mean that because great physicists say such things, philosophers of science cannot object. One can certainly criticize Einstein and Bohr and say that they were all mistaken, that they did not understand philosophy. But at least one thing is clear: the development of contemporary physics does not, as a matter of course, support realism; rather, the opposite is true. In fact, it is precisely anti-realism that provides a more reasonable and more acceptable explanation of the success of modern physics. It is realism, not anti-realism, that will have to face the situation of contending with the mainstream trend of contemporary science as a whole. The author of this material seems to take the progress of modern science in a rather self-evidently naive way. Scientific realists claim that unless one believes in realism, science cannot explain natural phenomena. But we can see that quantum mechanics shows exactly the opposite: unless one believes in anti-realism, you cannot understand quantum mechanics and quantum phenomena. Scientific realists can only evade quantum mechanics; this is probably why there are so many supporters of hidden-variable interpretations among philosophers of science.
The author’s discussion of anti-realism next is even more “self-talk”: he neither gives due attention to modern science, nor gives sufficient attention to the many and varied anti-realist doctrines, but instead helps anti-realism construct a mistaken line of argument in his own way.
After fiddling around with EE1 through EE3 for quite a while, the author finally says: “we will hardly get any basis for concluding that any two genuine competitors, as EE3 requires, are empirically equivalent in the absolute sense.” Fine—let us grant that his fiddling is meaningful, and look at EE3. Are we really unable to find theories equivalent in such a strong sense? In fact, not only is it theoretically possible, but we have long since found such an example in practice—
We know that quantum mechanics has three equivalent theoretical formulations: first, the matrix mechanics founded by Heisenberg; second, the wave mechanics founded by Schrödinger; and third, Feynman’s path integral. Feynman proved that these three descriptions are mathematically equivalent. What does mathematically equivalent mean? It means that, as theoretical tools, they are empirically equivalent in the absolute sense: the empirical results derived from using them are equivalent at all times. Not only can no currently available empirical evidence distinguish them, but no possible empirical result could distinguish them. Yet mathematical equivalence does not mean the same so-called “physical meaning.” \((a\times b)\times c\) and \((b\times c)\times a\) are mathematically equivalent, but their physical meaning is different. These three descriptions explain the world in radically different ways, at least on the surface: one treats the electron as an indeterminate particle, one treats the electron as a wave function, and one treats the electron’s path as the sum of all possible paths (and in Feynman’s case, the whole universe only needs one electron running back and forth to be enough). In what sense can you say that these three sets of theories describe the same “reality”? Do you mean that the real world is both this way and, at the same time, that way? Then why not simply admit that none of these descriptions is a genuine, literal depiction of reality, but merely a set of computational tools? Isn’t anti-realism the more natural conclusion?
There is no need to say much about the final section on “meta-induction.” One point is worth mentioning: although the claim of scientific realism can indeed be formulated as saying that “most” of the terms in contemporary theories are true, the corresponding anti-realist claim is not necessarily preoccupied with “most” versus “a few.” In fact, typical anti-realism does not accept the “truth” of any theoretical term, or else demands a new interpretation of the word “true.” Even what the author thinks is obviously real, such as “atoms,” is, in anti-realist terms, also a product of construction.
Let me now leave this overly one-sided material and reorganize what a form of anti-realist claim might mean:
First we must make clear that what we are discussing now is the problem of “realism and anti-realism,” not “knowledge and anti-knowledge,” or “science and anti-science,” and so on. The target of anti-realism is “realism,” not science or knowledge. Realists may claim: the possibility of knowledge and science depends on some realist presupposition, belief, premise, or the like. But anti-realists disagree. What they disagree with is this: such a presupposition or assumption is not necessarily needed for certain knowledge and science to be possible; rather, the source or guarantee of such certainty does not have to rely on some realist belief, but can, for example, appeal to something like innate ideas, a priori forms of perception, pragmatism, instrumentalism, and so on. This is the real point at issue between realism and anti-realism,
More precisely, the focus of their dispute lies in the “meaning of theoretical terms.” Realists claim that a theoretical term succeeds because it “really exists,” whereas anti-realists say the opposite: a term seems as though there really is such a thing because of its success. It succeeds, so it is taken as true, rather than being successful because it is true. Then how can it be successful? Perhaps theories such as idealism, apriorism, pragmatism, instrumentalism, and so on can better explain this.
We know that common sense understands a noun by evoking a sense impression, including sight, hearing, touch, and so on. The so-called terms for “unobservable entities” in science, by contrast, cannot evoke a sense impression that points directly to them; they can only be understood through mathematics. The question is: which is more real? Is it possible to reduce unobservable entities to sense experience as well? Or, when they cannot be directly reduced to sense experience, how are we to understand their meaning? Of course, we can design experiments and then indirectly experience them through the results those experiments output. Realists then interpret this as: “If atoms exist there, we will test for such-and-such observable outputs in the experiment.” And when predictions like these often succeed, the most natural explanation is: “Atoms really do exist there”; otherwise, wouldn’t it be a miracle? So realism is reasonable.—And yet the key point here is how to translate scientific terms! You may understand this scientific proposition as “if atoms exist there, then…,” but I can also translate it as “if the conventional concept of ‘atom’ is indeed useful, and our design and arrangement of this experiment are also accurate, then we will observe…” In that case, the success of this prediction need not be explained by appealing to “atoms really exist”; at most it can only support the claim that our convention regarding the instrumental concept of the atom is indeed accurate and practical. As for how we establish the meaning of the concept atom in the first place: for example, according to operationalism, the meaning of a scientific term lies in describing a certain set of repeatable operations. Of course, one can ask further: how is a successful operational definition possible? Is it not because it truly “exists”? But here one must be very careful—what exactly does “exist” mean? It is precisely because concepts such as existence and reality, and the like, are unclear (and not empirically verifiable) that approaches such as operationalism came into being. There are also many explanations for the repeatability of experiments—for instance, the uniformity of nature, or appeals to innate forms of perception, innate ideas, God’s guarantee, and so on. Even if one appeals to the uniformity of nature, that is not the same thing as “scientific realism.”
For example, what is a “constellation”? A constellation is an accurate definition that can be observed repeatedly, because the starry sky is uniform; thus, in astronomical prediction, we can usually really see the constellation at the angle where we said we would see it. But is a “constellation” “real”? Is the Great Bear a reality? In a sense, the meaning of theoretical terms as understood by instrumentalism is similar to terms like “the Great Bear.” The external structure they refer to depends on human construction; nature itself does not contain any such thing. You can group these seven stars together, and I can group these five and those six together. In short, nature itself does not have some objective mode of division that exists independently of human beings. The distribution of the stars is originally chaotic and unordered, and the appearance of the starry sky depends entirely on where human beings are situated. In fact, most of the stars we see on Earth may already have died by the time their light reaches Earth, and the stars within a constellation may never even have existed simultaneously at the same moment. In short, human beings, out of their own subjective preferences, establish a certain order in the starry sky as they see it, dividing it into distinct “objects” here and there. Then people reach agreement through “convention,” for example, the Western zodiac and so on; the division of constellations is fixed and widely recognized. And the reason it is divided this way rather than that way depends first on the cultural background of the members of the community and their mutual compromises in communication, and second on whether such a division is “convenient and practical.” For instance, if you divide the ecliptic into twelve equal signs, certain astronomical records and descriptions become more convenient; for example, once someone says “the Leonid meteor shower,” it is easy to find the corresponding direction. But if you divided it into three parts or into two thousand parts, that would obviously be very inconvenient… In short, we can acknowledge that nature is uniform. But the problem is that in itself it is a chaotic “one,” or, one might say, a “many.” And the parts we carve out within it and call “the Great Bear,” the parts we delineate and call “rainbow,” the parts we define and call “atom,” and so on—all these are human conventions, not structures inherent in nature itself.
Anti-realism does not necessarily have to claim that science has nothing to do with “truth.” The key is that “truth,” “true,” “real,” and so on are, first of all, just concepts. What realism and anti-realism must argue about first of all is precisely the meaning of these concepts. You say science seeks truth—fine, I can agree with you. But the crucial thing is: then tell me, what does “truth” mean? You say truth is correspondence with the external world—then what do you mean by “external world”? And what does “correspondence” mean? … In the end, one finds that realists often can only explain one empty concept with another empty concept, and after all the back-and-forth they never really get anywhere. You say the success of science can only be explained by the truth of theoretical terms; but why should we believe in and proclaim the truth of a certain theoretical term? Is it not precisely because of its success? This is like the proposition “survival of the fittest”: why did this species survive?—Because it was adapted. Then how do we determine whether a species is adapted?—Look at its ability to survive. Saying “the true succeeds” is similar. It sounds as if it is saying something profound, but in fact it is nothing more than fitting one concept into another; it says almost nothing at all. Anti-realism today is not satisfied with speaking emptily about “truth,” nor does it engage in a circular game between truth and success. I can offer a more meaningful explanation: for example, the reason we believe a certain term to be “true” depends on its “success”; and the reason it can succeed is that it is more practical, more operational, more coherent and self-consistent, more concise, and so on. An explanation like that feels like it has some substance.
My own inclination is toward an anti-realism of a conceptualist kind, or in a Kantian sense, or in an intuitionist sense. According to Quine, contemporary disputes over the foundations of mathematics can be seen as a revival of medieval debates: medieval “realism” corresponds to contemporary logicism, “nominalism” corresponds to formalism, and a “conceptualism” corresponds to intuitionism.
Simply put, realism in the Platonic sense maintains that there is a real world of Ideas. In Plato, that world of Ideas stands above the sensible world of reality, whereas in a certain sense modern science, through the mathematization of the world, has in effect fused that world of Ideas—namely the mathematical world—together with this everyday world. Nominalism, by contrast, holds that there is no such thing as “universals”; conceptualists think there are universals, but that they are human constructions.
Dummett supports an intuitionist anti-realism, and he believes that the watershed between realism and anti-realism lies in whether one accepts the law of bivalence. Although I must point out that intuitionism does not directly oppose the law of bivalence, in a certain sense this characterization is still accurate.
Put simply, realism says that, for example, an electron is either in a certain place or not in that place; the situation has only two values, “yes” or “no.” Even if we make no observation of it at all, we can still say: “That electron is either there or not there.” This is the bivalent logic of realism.
Intuitionist or constructivist anti-realism, however, does not agree with such logic. It does not directly oppose bivalent logic; rather, it has a different view of the objects about which logic can speak. Intuitionism holds that only those objects that can in principle be constructed in a finite number of steps are discussable. In mathematics, this appears as opposition to actual infinity; in physics, the constructivist attitude is this: when we observe some object, the state it presents is indeed in accordance with bivalent logic. That is to say, when we observe an electron, we can indeed expect such an outcome: it either passed through here, or it did not pass through here. But when we do not observe which slit the electron passed through, and instead only watch the result on the screen, we cannot say, “The electron that reached the screen either passed through this slit or did not pass through it.” Because we have not constructed an observation of that electron passing through the double slit. To put it more provocatively, we have not constructed that electron as it is in the process of passing through the double slit, because the thing called “electron” is not real but a concept constructed by human beings. So we cannot claim, “That electron either passed through here or did not pass through here.” Therefore, in response to the apparently paradoxical character of quantum mechanics, I think only intuitionist anti-realism provides the most reasonable explanation, whereas logicism and realism fail here.
(Depending on the time and the response to the talk, one can finally add some impromptu popular introduction to intuitionism or quantum mechanics.) More than two years ago I wrote an article, “An Apologia for Intuitionism.” Although it was still rather immature at the time, I am confident that it offered a fairly accessible outline of the basic aims of intuitionism, better than most domestic writings I have seen; as for a popular understanding of quantum mechanics, you can also discuss it with me privately, and you may also refer to an even earlier article of mine, “Loose Talk on Science pp,” in which the second half uses the EPR paradox to display the strangeness of the quantum world. I would also warmly welcome students to continue the discussion on KKBBS.
Gu Wu
December 28, 2008
Translated from the Chinese original with AI assistance. The original text is authoritative.
Leave a Reply