1 Scientific Method and the Problem of Demarcation
If there is one thing that most people think is special about science, it is that it follows a distinctive “scientific method.” If there is one thing that the majority of philosophers of science agree on, it is the idea that there is no such thing as “scientific method.”
If you are one of those people who has saved his or her textbooks in astronomy, physics, chemistry or biology, I invite you to go find one now and open it to the first page. This is typically the page that is never assigned by the professor and never read by the students, but it is nonetheless de rigeur, because it purports to provide the basis for why the rest of the claims in the book are to be believed. Often this page gives an account of “scientific method.” There are different renderings, but here is a simple version of the classic so-called five step method:
(1) Observe.
(2) Hypothesize.
(3) Predict.
(4) Test.
(5) Analyze results, revise hypothesis, and start again.1
Is this how scientific discoveries are in fact made? Few would claim this. The way that scientific theories are produced is an often messy process that involves serendipity, failure, blind alleys, heartache, dogged determination, and the occasional stroke of luck. But that is not what is supposed to be special about science. The quirky way that scientists sometimes get their ideas is the stuff of legend. One thinks of August Kekulé in front of the fireplace, dreaming of a snake swallowing its tail, leading to the benzene ring, or Leo Szilard stepping off the curb as the traffic light changed from red to green, realizing in that moment that it was possible to split the atom.2 Inspiration in science, as in art, can come from a diversity of sources. Yet many hold that the results of science have a special claim to be believed because of the distinctive way that they can be rationally reconstructed after the fact. Thus it is not the way that scientific theories are found that gives them such great credibility, it is the process by which they can be logically justified.
Science textbooks provide a cleaned-up version of history. They give us the result of many centuries of scientific conflict and make us feel that the process inevitably led to our present enlightened understanding. Historians of science know this to be inaccurate, but it remains immensely popular, because of the great convenience that this account of “scientific method” provides in supporting not only the claim that the content of science is especially credible but also the idea that the process of scientific explanation can be emulated by other disciplines that wish to make their own empirical discoveries.3
Yet even if the classic five-step method proves too simple to get the job done, there are other ways that philosophers have sought to characterize what is distinctive about science, and some of these focus on methodology. Here it is important not to get confused. The claim that there is no universal one-size-fits-all “scientific method”—where we put in sensory observations at one end and get scientific knowledge at the other—does not contradict the idea that there could be some unique methodological feature of science. To say that there is no recipe or formula for producing scientific theories is a very different thing than to claim that scientists have no methods whatsoever. This is to say that even if most philosophers of science are willing to reject the idea of a simple “scientific method,” many still think there is enormous benefit to analyzing the methodological differences between science and nonscience, in search of a way to justify the epistemic authority of those scientific theories that have already been discovered.
The Relevance of the Problem of Demarcation
One benefit of focusing on the methodology of science is that it purports to provide a way to demarcate between what is science and what is not. This is the so-called problem of demarcation, and it has been of enormous concern to the philosophy of science at least since the time of Karl Popper at the beginning of the twentieth century. In his essay “The Demise of the Demarcation Problem,” Larry Laudan claims that the problem of demarcation goes all the way back to Aristotle—who sought to differentiate between knowledge and opinion—and surfaced again in the era of Galileo and Newton—who pushed science into the modern era by using empirical methods to understand how nature worked. By the beginning of the nineteenth century, Laudan asserts, Auguste Comte and others began to hone in on the claim that what was distinctive about science was its “method,” even if there was as yet no widespread agreement about what that actually was.4 By the beginning of the twentieth century, philosophers were ready to sharpen this analysis and attempt to solve the problem of demarcation by creating a strict “criterion of demarcation” that could differentiate science from nonscience.
The Logical Positivists tried to do this on the basis of the allegedly special “meaning” of scientific statements. In contrast to other kinds of assertions, scientific statements were accepted as making a difference to our experience in the world, which meant that they must in some way be verifiable through sensory data. If scientists said that the planet Venus went through phases, we had better be able to see that through a telescope. Statements that could not do this (other than those used in logic, which is deductively valid and so already on firm ground) were branded “cognitively meaningless” because they were unverifiable; they were dismissed as nonsense and unworthy of our time, because there was no procedure for determining if they were right or wrong. If a statement about the world purported to be true, the positivists claimed, it must be verifiable by experience. If not, then instead of being scientific it was just “metaphysics” (which was the pejorative term they used to cover huge swaths of knowledge including religion, ethics, aesthetics, and the vast majority of philosophy). To pull off such a hard and fast distinction, though, the Logical Positivists needed to come up with a “verification criterion” by which one could sort meaningful from meaningless statements. And this, owing to technical reasons that ultimately came down to the problem that they couldn’t get the sorting right, led to their undoing.5
The problem of demarcation was then taken up by perhaps its greatest champion, Karl Popper. Popper understood—even before Logical Positivism had formally failed—that there were problems with pursuing the verification of scientific statements. The positivists had based science on inductive inference, which undercut the idea that one could prove any empirical statement to be true. David Hume’s famous problem of induction prevented the sort of logical certainty that the positivists coveted for scientific statements.6 Even if they could not be proven, however, weren’t scientific assertions nonetheless uniquely meaningful, given the fact that in principle they could be verified? Popper thought not, and regarded the pursuit of “meaning” as another mistake in the positivists’ approach. What made science special, he felt, was not its meaning but its method. Popper thus set out, in the winter of 1919, to try to solve the problem of demarcation in another way—by renouncing both verification and meaning, focusing instead on what he called the “falsifiability” of scientific theories: the idea that they must be capable of being ruled out by some possible experience.
What concerned Popper was the difference between statements like those in astrology—that seemed compatible with any evidence—and those of science, that take some risk of being wrong. When an astrologer produces a personalized horoscope that says “You are sometimes insecure about your achievements and feel like an imposter” this can feel like a stunning insight into your inner-most thoughts, until you realize that the same horoscope is used for all clients. Contrast this with what happens in science. When a scientist makes a prediction, it comes with an understanding that if her theory is correct you will see what was predicted. And if you do not see that result, then the theory must be flawed.
Popper used this sort of contrast to think about the possible methodological difference between science and nonscience. He was searching for a way to forgo the impossibly high standard which said that scientific statements must always be proven by their evidence, but would still allow evidence to count. And then it hit him. If the Logical Positivists and others were searching for a way to differentiate science from nonscience—but were blocked from being able to say that scientific statements were verifiable because of Hume’s problem of induction—why not instead follow the path of deductive certainty that was already enjoyed by logic?
Those who have studied formal logic know that the simplest and most famous deductively valid inference is modus ponens, which says “If A, then B. And A. Therefore B.” No problem here. No need to check to see whether it “makes a difference to our experience.” Deductive arguments are and always will be valid because the truth of the premises is sufficient to guarantee the truth of the conclusion; if the premises are true, the conclusion will be also. This is to say that the truth of the conclusion cannot contain any information that is not already contained in the premises. Consider the following valid argument:
If someone was born between 1945 and 1991, then they have Strontium-90 in their bones.
Adam was born in 1963.
Therefore, Adam has Strontium-90 in his bones.7
The problem with scientific statements, however, is that they don’t seem to follow this form. For hundreds of years before Popper, they were accepted as being inductive, which meant that the reasoning looked more like “If A, then B. And B. Therefore A.” For example:
If someone was born between 1945 and 1991, then they have Strontium-90 in their bones.
Eve has Strontium-90 in her bones.
Therefore, Eve was born between 1945 and 1991.
Obviously, this kind of argument is not deductively valid. The fact that Eve has Strontium-90 in her bones is no guarantee that she was born between 1945 and 1991. Eve might, for example, have grown up near a nuclear reactor in Pennsylvania in the late 1990s, where it was found that Strontium-90 was present as a result of environmental contamination. Here the form of the argument does not guarantee that if there the premises are true, the conclusion will be true. With inductive arguments, the conclusion contains information that goes beyond what is contained in the premises. This means that we will have to engage in some actual experience to see if the conclusion is true. But isn’t this how we do science? Indeed, when we are engaged in reasoning about empirical matters, we often seek to go beyond our firsthand experiences and draw inferences about those situations that are similar to them. Even though our experience may be limited, we look for patterns within it and hope to be able to extrapolate them outward.
Suppose we are interested in a straightforward empirical issue such as the color of swans. We’ve seen a lot of swans in our life and they have all been white. We may therefore feel justified in making the assertion “All swans are white.” Is this true? We’ve made our observations and have come up with a hypothesis, but now it is time to test it. So we make a prediction that from now on every swan we see will be white. Here is where it gets interesting. Suppose that this prediction turns out to be fulfilled. We may live our whole lives in North America, and, as it turns out, every single swan we ever see is white. Does this prove the truth of our assertion? No. It is still possible that someday if we go to Australia (or just open Google), we will see a black swan.
When we are trying to discover empirical truths about the world, we are hampered by the fact that our experience is always finite. No matter how long we live, we cannot possibly sample all of the swans who have lived or ever will live. So we can never be certain. If we wish to make blanket statements about the world—sometimes instantiated in scientific laws—we face the in principle worry that some future piece of evidence may come along to refute us. This is because the form of the argument that we are using here is inductive, and inductive inferences are not deductively valid. There is just no way to be certain that the rest of the world will conform to our limited experience.
Science nonetheless works pretty well. Although it may not guarantee the truth of our assertions, we are at least gathering evidence that is relevant to the warrant for our beliefs. And shouldn’t this increase the likelihood that our general statements are true?8 But why settle for this? Popper was bothered by the inductive form of inference used by positivists and others as the basis for science. But if that is its logical foundation, how can we possibly demarcate science from nonscience? To admit that “we could be wrong” doesn’t sound like much of a distinction. Popper sought something stronger. He wanted a logical basis for the uniqueness of science.
Popper didn’t have to look far. The inductive argument that we used above has a name—“affirming the consequent”—and it is a well-known fallacy of deductive logic. But there are other, better forms of argument, and one of the most powerful—modus tollens—is deductively valid. It works like this. “If A, then B. And not B. Therefore, not A.”
If someone was born between 1945 and 1991, then they have Strontium-90 in their bones.
Gabriel does not have Strontium-90 in his bones.
Therefore, Gabriel was not born between 1945 and 1991.9
This was Popper’s insight: this, he felt, was the logical basis for scientific inference. Just because science seeks to learn from empirical facts about the world, this does not mean that it is doomed to the problems of inductive inference. For if we look at the argument above, we can see that it is possible to gather empirical evidence and learn from it in a negative way, such that if our test does not work out, we must revise our general assertion. Like the Logical Positivists, Popper was still relying on empirical evidence. But now, instead of that evidence making a difference to our experience so that it could be used for verification, evidence counted because the theory at hand was capable of being refuted by it.
Remember that black swan? If we’d seen one, it would have caused us to revise our hypothesis that “All swans are white.” A single counterexample has the power—through modus tollens—to affect our blanket statements about the world. And that provided a way, Popper thought, for giving up on the idea of verification in science. If we wish to demarcate science from nonscience, we have to ask a simple question: is the general statement that we have made about the world capable of being refuted by some possible experience, even if we have not had and may never have that experience? If the answer is no, then it is not scientific.
Fortunately for Popper, a real-life example of good science was close at hand. In fact, it may have been what inspired his theory. In May 1919, Arthur Eddington set out on an expedition to take photographs of the stars during a total solar eclipse. This was crucial for the confirmation of Einstein’s theory of general relativity. As Popper explains:
Einstein’s gravitational theory had led to the result that light must be attracted by heavy bodies (such as the sun), precisely as material bodies were attracted. As a consequence it could be calculated that light from a distinct fixed star whose apparent position was close to the sun would reach the earth from such a direction that the star would seem to be slightly shifted away from the sun; or, in other words, that stars close to the sun would look as if they had moved a little away from the sun, and from one another. This is a thing which cannot normally be observed since such stars are rendered invisible in the daytime by the sun’s overwhelming brightness; but during an eclipse it is possible to take photographs of them. If the same constellation is photographed at night one can measure the distances on the two photographs, and check the predicted effect. … Now the impressive thing about this case is the risk involved in a prediction of this kind. If observation shows that the predicted effect is definitely absent, then the theory is simply refuted. The theory is incompatible with certain possible results of observation—in fact with results which everybody before Einstein would have expected.10
In other words, the falsifiability of Einstein’s theory was a prime example of the proper way to do science. In one fell swoop, Popper claimed to have simultaneously solved the problem of demarcation and the problem of induction. That is, since science is not based on induction, it no longer mattered. He had now found a way for empirical observations to make a direct difference in testing our general assertions about the world. And, through modus tollens, this was deductively valid. It is important to understand that Popper was not claiming that his criterion of falsifiability was a way of demarcating meaningful from meaningless statements. Unlike the positivists, Popper did not need to use meaning as a proxy for verifiability, since he had found a direct way to tell the difference between scientific and nonscientific statements.11 It is worth pointing out that falsifiability thus purported to identify not only what was special about science but also what was wrong with those inquiries that were merely pretending to be scientific.
We have already mentioned the example of astrology—which goes back to Popper’s day—but consider here a more contemporary example. In 1981, the state of Arkansas passed Act 590, which required that public school teachers give “balanced treatment” to “creation science” and “evolution science” in the biology classroom. It is clear from the act that religious reasons were not to be offered as support for the truth of creation science, for this would violate federal law. Instead, the curriculum was expected to concentrate only on the “scientific evidence” for creation science. But was there any? And, how precisely was creation science different from creationism?
The act held that the existing situation could not stand, since teaching evolution alone could be taken as a violation of the separation between church and state, to the extent that this would be hostile to “theistic religions” and would tend to favor “Theological Liberalism, Humanism, Nontheistic religions, and Atheism in that those faiths generally include religious belief in evolution.”12 The strategy here was clear: not only were the proponents of creation science attempting to show that it was not religion, they were suggesting that evolution very nearly was religion. But since it was unacceptable to fight this battle in a religious venue, creation science advocates held that they merely wanted an equal chance to offer their views as a scientific contender to Darwin’s theory of evolution by natural selection.13
The fate of this particular piece of legislation—and the lawsuits that followed—will be discussed later in this chapter, and revisited in chapter 8, with intelligent design theory, which took a second swing at trying to get creationism into the public schools. For now the question is a philosophical one: could falsification identify what might be wrong with creation science? Some felt that it could, for just as with the earlier claims of astrology, it seemed that the main thesis of creation science—that God created the universe and all of the species within it—was compatible with any evidence. Didn’t the discovery of 65-million-year-old dinosaur fossils conflict with the 6,000-year timeline in the Bible? Not really, the creation scientists contended, for surely an omnipotent God could have created the entire fossil record! I hope it is clear from our earlier consideration of the problems with astrology that this sort of tendency to explain away any contrary evidence is not a shining example of falsifiability. Whereas true science goes out on a limb to test its theories against experience, creation science refused to change its theory even when there was evidence against it. Add to this the fact that creation science had precious little to offer as positive evidence in its favor, and many were willing to dismiss it as nothing more than pseudoscience.14
The virtues of falsification are clear. If Popper had found a way to solve the problem of demarcation, philosophers and scientists now had a powerful tool for answering the question of what is special about science. They also had a mechanism for dismissing and criticizing those practices—such as astrology and creationism—that they did not want to accept as scientific; if they were not falsifiable, they were not scientific. An added benefit of Popper’s approach was that he had found a way for a theory to be scientific without necessarily having to be true.15 Why did this matter? In seeking a criterion of demarcation, it mattered a great deal to those who were versed in the history of science, and understood that some of the greatest scientific minds of the last few millennia had said things that later turned out to be false. It would be wrong to think that they weren’t scientists. Even though Ptolemy’s geocentric theory was later replaced by Copernicus’s heliocentric one, this did not mean that Ptolemy wasn’t a scientist. He had based his theory on empirical data and had pushed things forward as far as he could. What mattered is that his claims were falsifiable, not that they were later falsified.
It would be easy to imagine that Popper’s new criterion of demarcation was also a vindication of the idea of “scientific method,” but that would be far from true. In fact, Popper was one of the earliest and harshest critics of the idea that there was such a thing as “scientific method.” In his most definitive statement on the subject, appropriately titled “On the Non-Existence of Scientific Method,” Popper wrote “As a rule, I begin my lectures on Scientific Method by telling my students that scientific method does not exist.”16 Elsewhere, he writes:
The belief that science proceeds from observation to theory is still so widely and so firmly believed that my denial of it is often met with incredulity. … But in fact the belief that we can start with pure observations alone, without anything in the nature of a theory, is absurd; as may be illustrated by the story of the man who dedicated his life to natural science, wrote down everything he could observe, and bequeathed his priceless collection of observations to the Royal Society to be used as inductive evidence. This story should show us that though beetles may profitably be collected, observations may not.17
It is important here to remember the distinction between saying that there is a “scientific method” and saying that there is some methodological difference—such as falsifiability—between science and nonscience. Although Popper is unequivocally rejecting the idea of “scientific method,” he still believes that we can have a criterion of demarcation and even one that is methodological in nature.18
This opinion was not shared by some of Popper’s critics, notably by one of his most famous, Thomas Kuhn, who felt that although Popper was correct to abandon the idea of scientific method,19 one should probably also give up on the idea that there is any distinctive methodological difference between science and nonscience. Note that this does not necessarily mean that one is giving up on the idea that science is “special” or even that there is a way of distinguishing between science and nonscience. Kuhn was not yet ready to do this (though many of his later followers were); instead, he merely pointed out that the process by which scientists actually work has much more to do with nonevidential “subjective” factors in theory choice, such as scope, simplicity, fruitfulness, and the ability to fit a theory with one’s other beliefs, and much less to do with any formal method. And surely this must have an impact on justification.
It is important to understand that Kuhn was not an opponent of science. He was not—although he has been blamed for it—one of those who later claimed that science was an “irrational” process, no better than any other way of knowing, nor did he believe that the social factors that sometimes influenced scientific theory choice undermined its claim to produce credible theories. Instead, Kuhn was at pains to make sure that we understood science for what it really was, feeling that even if we did so it would be no less wonderful. While Kuhn never took it upon himself to try to provide a criterion of demarcation, he did nonetheless feel himself to be a champion of science.20
What of Popper’s theory? Despite its virtues, it was severely criticized—by Kuhn and others—as offering too simple a picture of scientific theory change, especially given the fact that most science did not work precisely as the heroic example of Einstein’s prediction indicated. There are few such crucial tests, involving risky predictions and dramatic successes, in the history of science. Most of science actually grinds along fairly slowly, with tests on a much smaller scale and, tellingly, widespread reluctance to give up a workable hypothesis just because something has gone wrong.21 Yes, evidence counts, and one cannot simply ignore data and insulate a theory from refutation. Yet many philosophers, embracing the Duhem–Quine thesis (which says that it is always easier to sacrifice a smaller supporting hypothesis or make an ad hoc modification than to give up a theory), were skeptical that science worked as Popper said it did. Even though Popper maintained that his theory dealt only with the logical justification of science, many felt that there was a growing credibility gap between the way that scientists actually worked and the way that philosophers justified their work, given the sorts of social factors that Kuhn had identified. As Kuhn demonstrated, we are occasionally given to engage in a scientific revolution, but it does not happen nearly often enough for this to be accepted as the basis for demarcating between science and nonscience.
The upshot of all this is that by the 1970s, there was fairly wide agreement among most philosophers of science not only that the classic five-step scientific method was a myth, but also that there was no genuine methodological distinction between science and nonscience. This had great importance for the idea that science was special. Can one defend the idea that science is distinctive without also believing in scientific method or at least some other criterion of demarcation? Many said no.
Once Kuhn had opened the door to examining the workaday details of how scientists did their business—through “puzzle solving” and the search for accommodation to the dominant paradigm through “normal science”—the critics seemed unstoppable. To Kuhn’s horror (he did after all agree with Popper and other defenders of science that evidence counts and that the revolution from one scientific theory to another on the basis of this evidence is the hallmark of science), his work was often cited as support by those who no longer believed that science was special. Sociologists of science, relativists, postmodernists, social constructivists, and others took their turns attacking the idea that science was rational, that it had anything to do with the pursuit of truth, or indeed that scientific theories were anything more than a reflection of the political biases about race, class, and gender held by the scientists who produced them. To some, science became an ideology, and facts and evidence were no longer automatically accepted as providing credible grounds for theory choice even in the natural sciences.
Paul Feyerabend went so far as to claim that there was no method in science at all. This was a radical departure from merely giving up on the scientific method. Along for the ride went any claims about methodology (such as objectivity), a criterion of demarcation, and even the idea that scientific beliefs were privileged.22 Many wondered whether philosophy had now given up on science all together.
This is not to suggest that all philosophers of science felt this way. There were many who followed the ideas of Logical Empiricism (the successor to Logical Positivism), which held sway contemporaneously with Popper’s theory of falsification right through the Kuhnian revolution. Here the focus was on defending the special method of science—even picking up on the earlier positivist idea of a “unified science” whose method could be extended to the social sciences—not through falsifiability (or meaning), but through close examination of how one might build more credible and reliable theories even in the face of the problem of induction.23 Even here, though, it was necessary to modulate the full-throated defense of science, and certain concessions had to be made.24
By 1983, Larry Laudan, one of the most prominent philosophers of science, was ready to pull the plug on the idea that one could have a criterion of demarcation. Laudan’s work was not so radical as to suggest that science wasn’t important. He was one of the post-Kuhnians who looked for a way to uphold the idea that science could make “progress,” though certainly not toward “true” theories or in any way that suggested the hegemony of science over other ways of knowing. In his earlier-cited article “The Demise of the Demarcation Problem,” Laudan argued that there was no possible solution to the problem of demarcation, largely on the grounds that if it could be solved it would have been by now. By the time Laudan entered the picture, it goes without saying that there is no scientific method, but even the idea of finding another way of distinguishing between science and nonscience now seems dead.
Note that this does not necessarily mean that there is no difference between science and nonscience. One could even believe (as I think Laudan does) that science is uniquely explanatory. It’s just that we are not going to be able to find a workable device for demarcation. Even if we all agree in our bones what is science and what is not, we are not going to be able to create a hard and fast way to distinguish it. The technical reason for this, Laudan tells us, is that philosophers have not been able to come up with a set of necessary and sufficient conditions for science. And to him this seems to be an absolute requirement for fulfilling a criterion of demarcation.
What will the formal structure of a demarcation criterion have to look like if it is to accomplish the tasks for which it is designed? Ideally, it would specify a set of individually necessary and jointly sufficient conditions for deciding whether an activity or set of statements is scientific or unscientific. As is well known, it has not proved easy to produce a set of necessary and sufficient conditions for science. Would something less ambitious do the job? It seems unlikely. Suppose, for instance, that someone offers us a characterization which purports to be a necessary (but not sufficient) condition for scientific status. Such a condition, if acceptable, would allow us to identify certain activities as decidedly unscientific, but it would not help “fix our beliefs,” because it would not specify which systems actually were scientific. … For different reasons, merely sufficient conditions are equally inadequate. If we are told, “Satisfy these conditions and you will be scientific,” we are left with no machinery for determining that a certain activity or statement is nonscientific. … Without conditions which are both necessary and sufficient, we are never in a position to say ‘this is scientific: but that is nonscientific.25
What is the problem with giving only a necessary condition? It is too strict. By aiming to exclude all that is not science, we may also keep out some things that we would want to include. Suppose our necessary condition were that a science must be capable of performing controlled experiments. Doesn’t that rule out geology? Astronomy? All of the social sciences? Suppose, on the other hand, that we abandon this and instead aim at providing only a sufficient condition for scientific investigation: for instance, that it must be concerned with seeking truth based on empirical evidence. Here the concern is that we may have been too inclusive. Haven’t we now allowed as scientific the search for Bigfoot? By trying to include everything that is science, we may also let in those things that we would surely want to keep out.26 Thus, Laudan tells us, to have an adequate criterion of demarcation between science and nonscience, we need to specify a set of individually necessary and jointly sufficient conditions for science.27
There is perhaps no better illustration of the difficulties presented by adhering to such a lofty standard than Karl Popper’s own proposal of falsifiability. Is this intended to be a necessary standard for science, a sufficient one, or both? The search for an answer is maddening. In some places, Popper writes as if he intends his account to do any one of these. Consequently, his criterion has been criticized both for excluding legitimate science (such as evolutionary biology) and for allowing some of the pseudosciences (such as astrology) to seem scientific.28 Laudan in particular takes Popper to task for the latter, when he writes that Popper’s criterion “has the untoward consequence of countenancing as ‘scientific’ every crank claim which makes ascertainably false assertions.”29
This last claim would surely enrage Popper (who designed his criterion specifically to keep things like astrology out of the scientific pantheon). So perhaps falsification should be interpreted as providing only a necessary condition?30 But, as we have seen, there is a weakness in this approach as well.31 So maybe it is best to accept the idea that Popper intended to meet the highest standard by providing an “individually necessary and jointly sufficient” set of criteria. Late in life, Popper did at one point say “a sentence (or a theory) is empirical-scientific if and only if it is falsifiable.”32 In philosophy of science, those are the magic words: “if and only if” commits him to providing both necessary and sufficient conditions. For the reasons already given, however, falsification alone will not do, yet one looks in vain in Popper’s writings for a definitive statement of what other conditions might apply. Frank Cioffi, in his important essay “Psychoanalysis, Pseudoscience, and Testability,” comes close, arguing that in addition to falsifiability, Popper had intended to include the requirements that “energetic attempts are made to put the theory to test” and that “negative outcomes of the tests are accepted.”33 But even here, one runs into the previously encountered problem, identified by Kuhn and others, that negative outcomes are not always accepted as overthrowing a theory.
If even Karl Popper gets caught up in the problem of providing the necessary and sufficient conditions that Laudan requires for offering a criterion of demarcation, some may wonder whether the rest of us should just give up. That is precisely what occurred for almost three decades after Laudan’s essay, when many were compelled by his reasoning to abandon the attempt to provide a criterion of demarcation between science and nonscience. This is not to say that they necessarily gave up on the idea that science was special. Remember that, like Laudan, one might believe that science could be defined by ostension. Surely many, like Laudan himself, were not ready to give up on the idea that science was worth defending, even if it could not be done through demarcation. With Laudan’s positive account that it was still possible for science to make “progress” along Kuhnian lines, many pulled back from some of Feyerabend’s and the social constructivists’ extreme assertions that science was just another way of knowing. The lament instead was that even if we know science (and pseudoscience) when we see them, we are prevented from having a good way to say what defines them. Even if many had given up on the problem of demarcation, they had not given up on science.
This strategy, however, had its costs. The low point was reached in 1982, when Act 590 was challenged on constitutional grounds in the case of McLean v. Arkansas. The prominent philosopher of science Michael Ruse was called to testify as an expert witness and, when backed into a corner about the definition of science, gave a version of Popper’s theory of falsifiability. This ended up convincing the judge, who quoted liberally from Ruse’s testimony in his opinion that creationism was not science, and therefore had no business in the science classroom. Although Ruse had done the best he could—and was in my view unfairly criticized for being bold enough to appear in court and prevent the travesty of seeing creationism accepted as legitimate scientific theory—the academics’ disapproval was swift and direct. Laudan, who surely agreed with Ruse that creationism was a farce, decried his use of Popper’s theory in the judge’s decision.
In the wake of the decision in the Arkansas Creationism trial. … the friends of science are apt to be relishing the outcome. … Once the dust has settled, however, the trial in general and Judge William R. Overton’s ruling in particular may come back to haunt us; for, although the verdict itself is probably to be commended, it was reached for all the wrong reasons and by a chain of argument which is hopelessly suspect. Indeed, the ruling rests on a host of misrepresentations of what science is and how it works.34
Laudan’s worry can be summed up as follows:
Laudan replied that creationist doctrine itself is science by [Popper’s] criterion. It is obviously empirically testable since it has already been falsified. To be sure, its advocates have behaved in a nonscientific manner, but that is a different matter. The reason [creationism] should not be taught is simply that it is bad science.35
One imagines that if such scrupulously faithful adherence to the scholarship of the day had been allowed to hold sway in court the creationists would have been thrilled: yes, teach creationism as “bad” science, but teach it nonetheless.
Thus we see that failure to demarcate between science and nonscience can have real-world consequences outside philosophy. For one thing, the issue of teaching creationism in the public schools did not simply disappear in 1982, but instead has morphed and grown—partially as a result of philosophers’ inability to defend what is special about science—into the current claim that “intelligent design (ID)” (which I have elsewhere referred to as “creationism in a cheap tuxedo”) is now a full-fledged scientific theory that is ready for its debut in biology classrooms.36 This was again put to test at trial in 2005 in Kitzmiller v. Dover Area School District, where another judge—in a stinging rebuke that was reminiscent of the McLean decision—found that intelligent design is “not science” and ordered the defendants to pay $1 million to the plaintiffs. This may give pause to future ID theorists, but sadly this story is still not over, as current “academic freedom” bills are pending in the state legislatures of Colorado, Missouri, Montana, and Oklahoma, modeled on a successful 2012 Tennessee law that defends the rights of “teachers who explore the ‘scientific strengths and scientific weaknesses’ of evolution and climate change.”37
Figuring out how to demarcate between science and nonscience is no laughing matter. Being able to say, in public and in a comprehensible way, why science is special seems a particular duty for those philosophers of science who believe in science, but have not been able to articulate why. As climate change deniers begin to gear up, taking a page from the earlier battles of the creationists (and the tobacco lobby) in fighting scientific conclusions that they don’t like through funding “junk science,” then spreading it through public relations, isn’t it time that we found a way to fight back?
Of late, this is precisely what has happened. In 2013, philosophers Massimo Pigliucci and Maarten Boudry published an anthology entitled Philosophy of Pseudoscience: Reconsidering the Demarcation Problem, in which they self-consciously seek to resurrect the problem of demarcation thirty years after Laudan’s premature obituary. The papers are a treasure trove of the latest philosophical thinking on this issue, as the profession tries to steer its way out of the ditch where Laudan left it: where we believe that science is special, but can’t quite say how. It is disappointing—but certainly understandable—that after all this time philosophers are a little unsure how to proceed. Perhaps resurrecting the traditional problem of demarcation is the answer. Or perhaps there is another way.
It is no small thing to dismiss the problem of demarcation, which has been the backbone of the philosophy of science since its founding. The attractiveness of using its structure and vocabulary as a way to understand and defend the distinctiveness of science is obvious. Perhaps this is why virtually all previous attempts to say what is special about science have involved trying to come up with some criterion of demarcation. But there are many pitfalls to resurrecting this approach.
In Pigliucci’s essay “The Demarcation Problem: A (Belated) Response to Laudan,” he rejects the “necessary and sufficient conditions” approach, preferring instead to rely on Ludwig Wittgenstein’s concept of “family resemblance.” Pigliucci thus claims to rescue the problem of demarcation from Laudan’s “old-fashioned” approach (which may be conceived of as challenging Laudan’s “meta-argument” over what is required to solve the problem of demarcation).38 Instead Pigliucci’s idea is to treat learning the difference between science and pseudoscience as a kind of “language game,” where we come to learn the clusters of similarity and difference between different concepts by seeing how they are used. The goal here is to identify the various threads of relationship that do not fall neatly along the lines of necessary and sufficient conditions but nonetheless characterize what we mean when we say that some particular inquiry is scientific. Two of these threads—“empirical knowledge” and “theoretical understanding”—appear to do most of the work. As Pigliucci writes, “if there is anything we can all agree on about science, it is that science attempts to give an empirically based theoretical understanding of the world, so that a scientific theory has to have both empirical support and internal coherence and logic.”39 As a result, Pigliucci thinks that—among other things—we will have discovered a “Wittgensteinian family resemblance” for the concepts of science and pseudoscience that provides a viable demarcation criterion to “recover much (though not necessarily all) of the intuitive classification of sciences and pseudosciences generally accepted by practicing scientists and many philosophers of science.”40
This account, however, seems quite nebulous as a criterion of demarcation. For one thing, what is its logical basis? At various points Pigliucci refers to the use of “fuzzy logic” (which relies on inferring degrees of membership for inclusion in a set) to help make his criterion more rigorous, but it remains unclear how this would work. As Pigliucci admits, “for this to actually work, one would have to develop quantitative metrics of the relevant variables. While such development is certainly possible, the details would hardly be uncontroversial.”41 To say the least: one imagines that the central concepts of empirical knowledge and theoretical understanding may be equally as difficult to describe and differentiate from their opposites as the concept of science itself. Has Pigliucci solved the problem of demarcation or merely pushed it back a step?
Others who have pursued a “post-Laudan” solution to the problem of demarcation have encountered a similarly rocky path. In the same volume, Sven Hansson pursues an extremely broad definition of science. Apparently wary of the implications of classifying disciplines like philosophy as nonscience, he instead expands the scope of science to mean something more like a “community of knowledge” and then proceeds to demarcate this from pseudoscience. For all of the alleged advantages of rescuing the humanities from the realm of nonscience, however, the cost is quite high, for now he cannot say that the problem with pseudoscience has anything to do with its bastardization of empirical standards (since at least part of what he now classifies as science is not empirical either).42
Maarten Boudry takes a similarly questionable step in saying that he thinks there are really two demarcation problems—the “territorial” and the “normative”—instead of just one. The former dispute he dismisses as sterile. It is just a matter of “turf” that concerns separating science from legitimate but nonempirical epistemic endeavors like history and philosophy. According to Boudry, the real dispute is between science and pseudoscience; this is where the normative issue arises, because this is where we face those disciplines that are just pretending to be sciences.43 Yet this bifurcation of the problem of demarcation reveals a basic confusion between saying that a discipline is nonscientific and saying that it is unscientific. Does Boudry mean to identify the “territorial” dispute as one between fields that are scientific and nonscientific? If so, that is a highly idiosyncratic and misleading use of the term. The dispute that he appears to be searching for when he talks about the territorial problem of demarcation seems to be between science and what may be called “unscience.” Yet why would this be the proper alternative to the normative dispute? The more traditional interpretation of the demarcation debate—revealed in most scholarship in the field—is that between science and nonscience, or between science and pseudoscience. These are the terms of art used by Popper, Laudan, and most everyone else.44 Instead, Boudry seems to be creating a new demarcation problem, while saying nothing about why we should ignore the classic problem of demarcating between science and nonscience. But why does Boudry think that he can make a case for the normative battle between science and pseudoscience, when he has not legitimately dispensed with the larger issue of science versus nonscience? The straw-man “territorial” distinction between science versus unscience (history, philosophy, etc.) does not do the job.45
This struggle to explain whether the problem of demarcation should be between science versus nonscience—or between science versus pseudoscience—may seem like a mere terminological dispute, but it is not. For if we are attempting to distinguish science from all that is not science, it may lead to a very different criterion of demarcation than if we seek to distinguish science merely from its imposter. The important point here is to recognize that, according to most scholars, the category of nonscience includes both those fields that are pseudo-scientific and those that are unscientific. An inquiry can be nonscientific either because it is merely pretending to be scientific (in which case it is pseudoscientific) or because it concerns matters where empirical data are not relevant (in which case it is unscientific).46 (See figure 1.1.)
This failure to be specific about what one is differentiating science from, however, not only exists in the “post-Laudan” essays by Pigliucci, Hansson, and Boudry, but seems to reflect a deep equivocation in the literature that goes all the way back through Laudan to Karl Popper himself. Remember that in The Logic of Scientific Discovery, Popper says that he is demarcating science from math, logic, and “metaphysical speculation.”47 By the time he gets to Conjectures and Refutations, however, his target is pseudoscience. In Laudan’s essay, he too slides back and forth between talking about nonscience and pseudoscience.48
What difference does all of this make? It turns out to be crucial. Later we will revisit the issue of necessary and sufficient conditions and learn that the entire question of what is special about science may hang in the balance. We will see that the project of trying to solve the problem of demarcation is hamstrung unless we can specify precisely what it is that we are trying to define (science, nonscience, pseudoscience, or unscience) and, as we have seen, there has as yet been no definitive answer to that. My goal will be to provide a way to say what is distinctive about science without getting tripped up on the problem of providing both necessary and sufficient conditions—or trying to solve the problem of demarcation—because I do not think that these problems can be solved. Yet we still need a way to defend science.
First, however, let us deal with the problem of those who have misunderstood how science works.