Why Trust Science?

by Naomi Oreskes

  I am no fan of the American Enterprise Institute. With my colleague Erik M. Conway I have shown how they (along with other think tanks promoting laissez-faire approaches to social and economic issues) have persistently mispresented or mischaracterized scientific findings on climate change, as well as a variety of public health and environmental questions. (They are no fans of mine, either. Their scholars have attacked my work on scientific consensus.)11 But the question raised is a legitimate one. Should a scientific analysis be viewed as ex ante authoritative? Is it reasonable to take the default position that the scientific community can in general be trusted on scientific matters, but the petroleum industry (to use his example) cannot?

  Science in North American universities and research institutes is generally well funded and respected—typically much more so than the arts and humanities—but outside those hallowed halls something very different is transpiring. The idea that science should be our dominant source of authority about empirical matters—about matters of fact—is one that has prevailed in Western countries since the Enlightenment, but it can no longer be sustained without an argument.12 Should we trust science? If so, on what grounds and to what extent? What is the appropriate basis for trust in science, if any?

  This is an academic problem but one with serious social consequences. If we cannot answer the question of why we should trust science—or even if we should trust it at all—then we stand little chance of convincing our fellow citizens, much less our political leaders, that they should get their children vaccinated, floss their teeth, and act to prevent climate change.

  Scholars’ views on the answer to this question have changed dramatically and more than once in the past century. Moreover, some of the answers that scientists offer are manifestly contradicted by historical evidence. It is routine, for example, for scientists to insist that their theories must be correct, because they work. How else, they argue, would planes fly or medicines cure disease?13 But utility is not truth: we can identify many theories in the history of science that worked and later were rejected as wrong. The Ptolemaic system of astronomy, the caloric theory of heat, classical mechanics, and the contraction theory of the Earth explained observed phenomena and made successful predictions, and are now on the scrap heap of history. Many scholars in the history and philosophy of science and science studies have, however, recently converged on a new view that does hold up to scrutiny: of scientific knowledge as fundamentally consensual. This consensual view of science can help us address the current crisis of trust.

  The Dream of Positive Knowledge

  Throughout the eighteenth and the early nineteenth centuries, most scholars located the authority of science in the authority of the “man of science.”14 The results of scientific investigations were trustworthy to the extent that the people who undertook them were. This is one reason why scientific honor societies, such as the Royal Society or the Académie des Sciences, were created: to acknowledge and identify the “worthies” whose opinions on scientific matters should be sought, trusted, and heeded.15 These societies served to identify the individuals whose work was considered worthy of acceptance. In the United States, this ideal was instantiated in the creation of the US National Academy of Sciences during the Civil War to advise President Lincoln. Identifying these “great men” of science would enable the president to get the reliable advice he needed.

  However, in the mid-nineteenth century, a substantive intellectual shift occurred, driven to a significant extent by the work of Auguste Comte (1798–1857), variously credited as the founder of sociology, the founder of philosophy of science in its modern form, and the founder of the philosophical school known as positivism.16 Comte’s work is abundant and complex and has been subject to various considerations and reconsiderations, refutations and restorations, but the most important aspect, for our purposes, is his commitment to the idea of positive knowledge. Science, Comte believed, was uniquely able to provide positive—which is to say reliable—knowledge. While the term “positive knowledge” is no longer much used apart from academics discussing it, most often as a discredited concept, the idea persists in our linguistic conventions. We still retain the notion of something being “absolutely, positively true.” In English we can ask: “Are you positive?” To which you may reply: “Yes, I’m positive.”

  For Comte, the key element in the concept of positive knowledge is method, which he contrasted with doctrine—whether religious, superstitious, or metaphysical. The doctrines of religion and metaphysics, he argued, were forms of bias and blinkering that impeded intellectual and social progress, which the method of science, by contrast, could provide. By applying method to the pursuit of knowledge, science had the potential to liberate men and women from the shackles of religion and superstition.

  Comte’s philosophy (like many in the nineteenth century, including famously Marxism) was teleological: he saw human history as being characterized by three stages: the theological or fictitious, the metaphysical or abstract, and the scientific or positive. These were not necessarily sequential—they might coexist within a society or even with a person—but overall the direction of progress was from theology to science, with metaphysics serving as a necessary transition.17 In the “positive stage” of human development, theology and metaphysics are replaced by scientific reasoning. And scientific reasoning is rooted in observation.

  It has been argued that Comte was seeking to replace conventional religion with a new religion of science, and there is some justice to this claim. Teleology is a common feature of many religions. He accepted that people had a need for moral principles but thought those principles could be found in the humanistic ideals of truth, beauty, goodness, and commitment to others. He also believed that people had a need for ritual and proposed to replace the veneration of Christian saints with a set of positivist heroes. In his own life, he set aside time for meditation and affirmation of his central values.18 But whether his views were quasi-religious or not, the key point for our discussion is that for Comte—and generations of those who followed him, knowingly or not—science was reliable because of its commitment to method. This leads one to ask: what is that method?

  Comte was sensitive to the variety of scientific disciplines that were developing at that time. He did not assert that their practices were uniform, but he did believe that they shared a fundamental characteristic of the “positive” state of human existence. He wrote:

  In the positive state, the human mind, recognizing the impossibility of obtaining absolute truth, gives up the search after the origin and hidden causes of the universe and a knowledge of final causes of phenomena. It endeavours now only to discover, by a well-combined use of reasoning and observation, the actual laws of phenomena—that is to say their invariable relations of succession and likeness. The explanation of facts, thus reduced to its real terms, consists henceforth only in the connection established between different particular phenomena and some general facts, the number of which the progress of science tends more and more to diminish.19

  In stressing the importance of empirical regularities, Comte was making an argument similar to the British empiricists, particularly David Hume.20 He acknowledged his debt to British empiricism, particularly the work of Francis Bacon, writing, “All competent thinkers agree with Bacon that there can be no real knowledge except that which rests upon observed facts.”21 But he was not the “naïve positivist” that some later commentators made him out to be. He was a sophisticated thinker who recognized that our theories structure our observations as much as our observations structure our theories:

  If we consider the origin of our knowledge, it is no less certain that … [as] every positive theory must necessarily be founded upon observations, it is, on the other hand, no less true that, in order to observe, our mind has need of some theory or others. If in contemplating phenomena we did not immediately connect them with some principles, not only would it be impossible for us to combine these isolated observations and, therefore, to derive a
ny profit from them, but we should even be entirely incapable of remembering the facts, which would for the most part remain unnoted by us.22

  We can understand, therefore, why primitive humans had need of religion, superstition, and metaphysics: these early concepts were a step toward apprehending the world around us. We need not disdain or disparage these early stages in human development, we simply need to recognize and accept that to move forward—to identify the true laws that govern nature—our thinking needs to be grounded upon observation. In his words: “we must proceed sometimes from facts to principles [and] at other times from principles to facts,” but ultimately we will establish “as a logical thesis that all our knowledge must be founded upon observation.”23

  Comte was also a fallibilist: he recognized that our views would grow and change and that his own vision would in time be modified. (Indeed, if his basic concept was correct, then the progress of knowledge would necessarily modify our views, and we might note that the persistence of religion has falsified a key element of his teleology.) But, to his credit, Comte was consistent insofar as he insisted that future change in our thinking would be the outcome of our observations.

  Comte was also reflexive, recognizing that the practices of observation must themselves be subject to observation. An improved knowledge of positive method must come, therefore, not by theorizing it but by studying it; we must observe science in order to understand it. Comte thus anticipated Bruno Latour and his anthropological studies of laboratory science by more than a century when he held: “When we want not only to know what the positive method consists in, but also to have such a clear and deep knowledge of it to be able to use it effectively, we must consider it in action.”24

  Comte’s key move was to insist that science is reliable not by virtue of the character of its practitioner, but by virtue of the nature of its practices.25 We need to attend to these practices by studying them empirically. The key questions, then, for those who took up the Comtean program were: What exactly are those practices? Is there a scientific method?

  Varieties of Empiricism

  For twentieth-century empiricists, which we have come to call logical positivists or logical empiricists, the answer to the question of the method of science was the principle of verification.26 The concept was developed most extensively by a group of German-speaking philosophers and scientists, known as the “Vienna Circle.” The most famous English language articulation of the verificationist program came from the Oxford philosopher A. J. Ayer (1910–89). In his 1936 book, Language, Truth and Logic, which is still in print, Ayer summarized the principle by framing it in terms of the problem of meaning: A statement can be considered meaningful if and only if it can be verified by reference to observation. Put another way, “some possible observation must be relevant to the determination of [the statement’s] truth or falsehood.”27 Science is the practice of formulating meaningful statements, and using observations to judge whether a meaningful statement is correct.

  Verification gives us the basis for evaluating what is or is not justified true belief. If a claim can be verified through observation, and if it has in fact been so verified, then we are justified in believing it, which is to say, justified in accepting it as true. If a claim cannot be so verified, then it is meaningless and need not detain us further. Thus, in one fell swoop did Ayer dispense with religion, superstition, and various forms of political ideology and theory that were unverifiable. The principle of verification provided a means of demarcating scientific knowledge from non-scientific knowledge: scientific claims were verifiable thorough observation; claims that were not verifiable were not scientific.

  Like Comte, Ayer was ambitious but not naïve. He understood that in practice any observation necessarily entails background assumptions. But, like his Vienna Circle colleagues Rudolf Carnap and Otto Neurath, he insisted that verification through observation was the key component to meaning, hence the moniker verificationism. In order to test a statement, one had to be able to deduce an observable consequence from it and express that deduction as a statement, and that deduction had to be specific to the statement under investigation for the verification to be dispositive. Ayer wrote: “A statement is verifiable, and consequently meaningful, if some observation statement can be deduced from it in conjunction with certain other premises, without being deducible from those other premises alone.”28

  Ayer and his colleagues recognized that any program that foregrounded observation necessarily faced the problem of induction: namely, how many observations are needed to conclude that a statement is true? Following Hume, his answer was that inductive knowledge was necessarily probabilistic, and he suggested that one needed to allow for weak and strong forms of verification, based on the quantity and quality of available relevant observations. These sorts of concerns underpinned research on the character of scientific observation, which quickly led to various complications regarding the formulation of observation statements, the meaning of terms, and the identification of what, precisely, was being verified by any particular observation or set of observations.

  These issues detained many logical empiricists for the rest of their lives. Carl Hempel, in particular, paid attention to the role of hypothesis in generating testable observation statements; Carnap focused on the observation statements and the language in which they were rendered, and famously argued with Willard Van Orman Quine over whether observations could really confirm or refute beliefs. (Quine concluded they could not, a point we will take up.) This work did not resolve the issues it entailed.29 For our purposes, the important point is that the logical empiricists sustained the central Comtean idea that the core of scientific method is verification through experience, observation, and experiment.

  Challenges to Empiricism

  While logical empiricism is often attacked as the ruling dogma of twentieth-century philosophy of science, it never really ruled. Even in its heyday, several important challenges were already underway.30

  Karl Popper and Critical Rationalism

  The most well-known critic of logical empiricism is Karl Popper (1902–94). Popper rejected several key tenets of logical positivism. First, he denied that induction was the method of science. Second, he argued that what distinguishes science from other forms of human activity is not its activities, but its attitude. Great scientists are notable for the critical attitude they take toward their work, which is an attitude of skepticism and disbelief. Third, he insisted that the goal of science is not to prove theories—since that cannot be done—but to disprove them. He introduced his now-famous notion of falsifiability, concluding that what distinguishes a scientific claim from a non-scientific one is not that there is some observation by which it can be verified, but that there is some observation by which it can be refuted.

  These three ideas are linked in the following way. There may be habits or practices or even principles of induction, but there is no rational rule of induction. Inductive inferences cannot be justified based on any purely logical rule, and therefore cannot be established with logical necessity. This is what nowadays is referred to as the black swan problem. I may have observed one hundred swans, or one thousand, or ten thousand, and found that they have all been white, as have all the swans observed by my scientific colleagues. Therefore, my colleagues and I conclude (seemingly with robust warrant) that all swans are white. Yet, one day I travel to Perth, Australia, where I see a black swan.

  Thus, we see that observations cannot prove that a theory is true, no matter how extensive or comprehensive. Refutation may be lurking around the corner (or the antipodes). If science is to be a rational enterprise, induction therefore cannot be its method.

  Because observation alone cannot give us logical grounds to support inductive generalizations, verification cannot be the basis of scientific method. However, the observation of the black swan did prove that my inductive generalization was false, so there is a logic of refutation. There is a logical asymmetry between verification and falsification: veri
fications are always necessarily provisional, whereas falsifications (Popper held) can be dispositive. Given this, as a scientist I should not be seeking observations that confirm my theory, but observations that might refute it. The method of science, Popper therefore concludes, is neither generalization from observation nor verification by observation, but falsification. Put another way, the key activity of science is not the gathering of observations, but the formulation of conjectures and the pursuit of specific observations that may refute them. Thus the title of his famous collection of essays and lectures: Conjectures and Refutations.

  Even more urgently than his logical positivist colleagues, Popper held science to be the model of rationality, insisting that critical rationality is not only the appropriate basis for intellectual inquiry, but also for politics and civil society, as it empowers resistance to authoritarianism of both the right and the left. Therefore he labeled his approach critical rationalism. His project was both epistemological and political: he sought an epistemology that would enable not just scientific rationality but also political rationality in democratic forms of governance. Among other things, Popper sought to refute Marxism by showing that “scientific socialism” was an oxymoron, because problems in Marxist theory were never taken as refutations but only as elements to be explained or accounted for in some way.31

  Popper’s critical rationality ironically opened the door for a form of radical skepticism that he abhorred. Popper pushed fallibility further than his predecessors, insofar as he insisted that refutation is not merely an inevitable feature of science, but the goal of it; it is through refutation that science advances. But if our scientific views are not only soon to be refuted, but should be refuted, then why should we believe any of it?32 Popper’s answer was to develop the notion of corroboration: that we can have good reason to believe theories that have passed severe tests, such as the deflection of starlight as a test of the general theory of relativity. Successful empirical tests corroborate theories, even if they do not prove them. In making this move, Popper helped to explain why theory testing plays such a major role in scientific practice, but he also radically weakened the otherwise strict tenor of his work: we are now left with having to make subjective judgments as to what constitutes a “severe” test and how many such tests we need.