Why Trust Science?

by Naomi Oreskes

  We swim in science every hour of every day, but we don’t talk much about it. The many roles of science in our lives are naturalized or black boxed. Many of those who question climate change or vaccines are more than happy to deploy drones as technologies of war, or for that matter to use Twitter. Drones depend historically on layered and clustered types of scientific theory and practice, going back many decades. Meanwhile those who promote creationism share their ideas on the web, which is a result of defense support for scientific research—in mathematics, electromagnetics, physics, and other fields. So both trust and mistrust of science are discriminating, selective, and biased. Political leadership in the United States controls the most powerful military in the world—power built by scientists and engineers. Yet this achievement of high technocratic reason, this weapons system that truly “works,” in fairly spectacular terms, does not seem to confer legitimacy on the enterprise of science-in-general.

  The late Carl Sagan is haunting me lately so I will quote him: “We live in a society exquisitely dependent on science and technology, in which hardly anyone knows anything about science and technology.” I agree with his claim that we live in a society that is exquisitely dependent on science and technology. But I am less sure about his claim that hardly anyone knows anything about it. People know the things that live inside their everyday lives, but they emphatically don’t see them as scientific.

  So, maybe we need an epistemology of frozen peas. For frozen peas, if interrogated historically, involve as many layers of science as do drones. As I have already noted, these layers include the geology of oil exploration, the development of monomer chemistry and polymerization, the impact of the evolutionary synthesis on agricultural breeding, the development of the new genetics and GMOs, the scientific understanding of conservation of matter in temperature change, developments in bacteriology, and even knowledge from the social sciences, in psychological theories of marketing, imagery, and persuasion that have reshaped the consumer experience in the twentieth century—helping manufacturers understand for example how to persuade people to buy and eat frozen peas.32

  You might not pay much attention to frozen peas, and I can’t blame you, but I use this commonplace and seemingly simple food technology to suggest just how invisible this scientific world-making has become—almost as though it were structured to be invisible. Scientists long ago began to systematically distance themselves from the technologies that their insights and understandings could produce. And their distancing has been successful. But people love and trust technology. The flow of prestige and legitimacy “down” from science to technology—the flow of trust, viability, proof of value, of “working”—should perhaps be transposed, for the good of science and the good of the world.

  Chapter 4

  WHAT WOULD REASONS FOR TRUSTING SCIENCE BE?

  Marc Lange

  The question of why we should trust science can easily induce a kind of dizziness and even despair. Suppose we try to argue that we should trust science because rigorous application of the scientific method has generally led to successful results: the discovery of many truths and the rejection of many falsehoods. This approach will not get us very far. It invites the reply: What is our basis for judging that the current verdicts of science consist of many truths? Apparently, our basis is that the current verdicts of science line up with our beliefs about what is true. But if (as is likely) we have arrived at our beliefs by relying on science, then we have only traversed a very small circle; we have tried to use science to vindicate itself, which cannot succeed.

  This kind of circularity is difficult to avoid. Suppose we try to argue that we should trust science because science has led to many accurate predictions and technological achievements. It has allowed us to predict accurately the outcome of adopting certain public health measures or of putting batteries and wires into certain configurations. As Professor Oreskes mentioned, the track record of science in these matters is rather good. So, the argument would conclude, we should trust science.

  But this reasoning was itself an instance of using science. The reasoning began by noting a pattern in our experience so far (that science has worked pretty well in the past). The reasoning then took these data as good evidence that the pattern will continue into the future. This is a good example of scientific reasoning. But to use scientific reasoning seems to beg the question, since our goal was to justify putting our confidence in scientific reasoning in the first place. (Thus there may be some concern about Comte’s suggestion, mentioned by Professor Oreskes, that we support science by studying scientists scientifically.)

  The same charge of circularity could be lodged against the idea, mentioned by Professor Oreskes, that we should especially trust science when science has employed peer review and other agreed-upon scientific practices, or that we should especially trust science when we are dealing with the verdict of recognized scientific experts working within their recognized field of expertise and conforming to recognized scientific procedures. Who is to recognize them? On what basis is someone to be judged an expert? By the endorsement of other experts, each of whom gets his or her status by the endorsement of still other experts? As I said, it is easy to succumb to a kind of dizziness at this point, with the looming threat of infinite regress or vicious circularity. It is also easy to imagine the American Enterprise Institute or young Earth creationists criticizing this incestuous pattern of experts vouching for other experts.

  This kind of extremely corrosive doubt has a venerable pedigree in philosophy. It can be traced back to David Hume1 in the eighteenth century, who is generally credited with having posed “the problem of induction.” (“Induction” denotes the form of reasoning by which hypotheses are confirmed by evidence in science.) Such pervasive doubt can be traced back even further—to René Descartes2 in the seventeenth century and to Sextus Empiricus in the first century, who famously wrote:

  Those who claim for themselves to judge the truth are bound to possess a criterion of truth. This criterion, then, either is without a judge’s approval or has been approved. But if it is without approval, whence comes it that it is truthworthy? … And, if it has been approved, that which approves it, in turn, either has been approved or has not been approved, and so on ad infinitum.3

  There are at least two recipes for combatting the dizziness and despair that this kind of wholesale skepticism tends to provoke. One recipe is to point out that to ask for a justification for science as a whole is to make an unreasonable demand. It is like asking someone to justify trusting in reason as opposed to faith, wishful thinking, or astrology. If you give a reason for trusting in reason, then you have presupposed what you are trying to show. On the other hand, if you give something that is not a reason for trusting in reason, then that is giving no sort of justification at all. The game is rigged. You should simply reject the demand to offer a reason when there is nothing that could possibly, even in principle, count as one. As Professor Oreskes mentioned, one of the most important features of science is that it is self-correcting; it is able to put in jeopardy any of its theories, scrutinizing its justification. But science cannot reasonably be expected to put all of its theories in jeopardy at once.4

  There is a second important recipe for combatting the dizziness and despair of the skepticism that we have been discussing. The question of why we should trust science might be put this way. Scientists begin by making a bunch of observations. That is supposed to be the first stage of scientific research—at least, first in logical order. In the second stage, scientists use their observations to confirm various theories that make predictions about what would be observed under various conditions—or perhaps to confirm various theories that purport to reveal the unobservable causes or mechanisms responsible for what we have observed. What justifies this second stage—this inductive leap from the safe ground of past observations to the risky business of predicting future observations or positing hidden mechanisms? The history of philosophy is littered with attempts to support th
is second stage given the first. But most of these attempts have ultimately been judged to be question-begging in one way or another.5

  If the challenge is to justify this second stage of scientific research given the first stage, then we can respond with the second recipe that I want to mention for combatting skeptical dizziness. The recipe is to point out that even at the first stage, we are already taking for granted that we are justified in engaging in the second stage; we are already presupposing that certain steps beyond our observations are justified. After all, if I take myself to be observing something to be the case, then I must believe that I am qualified to make that observation—that by dint of my training, I am now able to tell that something is the case simply by looking or hearing or smelling or whatever. I am not purporting to be infallible, of course. But I am purporting to be reliable enough that I am worthy of trust (in the absence of some specific reason to doubt my accuracy) in this case. Without this additional belief about myself, I cannot justly take myself to have genuinely observed that something is so.

  But how did I become justified in this additional belief about myself? The justification must derive from my own history of having purportedly observed things. My past behavior of saying that I saw various things, nearly always and only when those things were in fact there, justifies my belief that I will typically be reliable in the future when I say that I have seen those things. A generalization like that obviously goes beyond what we have already observed. So in making an observation, we must already allow that we are justified in believing various generalizations about matters that go beyond what we have already observed. As the philosopher Wilfrid Sellars says, in arguing for a similar point:

  The classical “fiction” of an inductive leap which takes its point of departure from an observation base undefiled by any notion as to how things hang together is not a fiction but an absurdity.… [T]here is no such thing as the problem of induction if one means by this a problem of how to justify the leap from the safe ground of the mere description of particular situations, to the problematic heights of asserting lawlike sentences and offering explanations.6

  What I have been discussing so far is the demand for a reason for trusting science as a whole, a demand that is problematic precisely because its target is science as a whole. This kind of wholesale demand for justification should be distinguished from a retail demand that asks why some particular scientific result should be trusted. That kind of question, unlike the wholesale question, can be answered without circularity—by appealing to other scientific findings. This is the approach implicit in the series of case studies that Professor Oreskes offers. In each of those cases, participants were called upon to present the epistemic credentials of various scientific hypotheses. In none of those cases were participants called upon to underwrite science as a whole.7

  However, a source of trouble for this wholesale/retail distinction comes from cases where very large bodies of theory are being called into question. As Professor Oreskes mentioned, the philosopher Thomas Kuhn8 was very influential in questioning the rationality of what he termed “scientific revolutions,” where an entire “paradigm” is called into question. A scientific revolution, according to Kuhn, is (in the phrase that Kuhn popularized) a “paradigm shift.” Rival paradigms disagree on what sorts of facts scientists can directly observe, what sorts of measuring devices are reliable, and how to interpret criteria of theory choice such as simplicity, accuracy to observations, explanatory power, and fruitfulness. As Professor Oreskes mentioned, Kuhn regarded rival paradigms as sharing only the most minimal standards for theory choice. So in a crisis, when the prevailing paradigm is in jeopardy and a justification is demanded for one of the new candidates for paradigm (or for sticking with the incumbent), the common neutral ground arbitrating among the rival candidates is too meager to sustain any powerful reasons favoring one candidate over another. This is one aspect of what Kuhn calls “the incommensurability of paradigms,” as Professor Oreskes mentioned.

  In a crisis, then, a retail challenge to a particular scientific theory becomes a wholesale challenge to the results of an entire scientific field. What sort of noncircular justification can there then be for one paradigm over another when scientific methods and scientific theories interpenetrate? In other words, Kuhn argued very successfully against a sharp boundary between scientific methods and scientific theories. What methods scientists believe to be reliable are informed by what scientists believe about what the world is like. But if theories and methods interpenetrate, then as scientific theories change, scientific methods change. There is then no permanent, neutral method to arbitrate between rival paradigms; although their common ground presumably includes mathematics, deductive logic, and the probability calculus, that is not enough common ground to serve as a neutral arbiter to decide the issue. This is the challenge that Kuhn launches at us.

  One promising reply to Kuhn’s challenge is to acknowledge that deductive logic, arithmetic, the probability calculus, and whatever else is permanently neutral common ground in science is not enough to decide in a crisis among the rival candidates for paradigm. Nevertheless, in a given crisis, there will be more common ground among the rivals than merely that which is common ground across every crisis. In different crises, this additional common ground will be different. Although it may be meager, creative scientists in various crises have found ways to take the common ground and extract from it powerful reasons supporting one candidate for paradigm over another.

  Let us look briefly at an example. Galileo managed to find a way to generate a powerful argument from the meager common ground available in the crisis of terrestrial physics occurring during his time. Galileo proposed that if a body falls freely to Earth from rest, then in each succeeding interval of time, the distance covered by the body grows as the odd numbers, so if 1s is the distance that the body traverses in the first time interval, then in the succeeding intervals, it covers the distances 3s, 5s, 7s, 9s, 11s.…9 Other scientists proposed rivals to Galileo’s “odd-number rule.” Honoré Fabri proposed that the distances traversed grow as the natural numbers (that is: 1s, 2s, 3s, 4s, 5s, 6s …). Pierre Le Cazre proposed that the distances grow as the powers of 2 (that is: 1s, 2s, 4s, 8s, 16s, 32s …). An experimental argument favoring one of these theories against the others would have been ineffective as long as there was no shared paradigm for terrestrial physics and so no agreement on which measuring devices were accurate for measuring time and distance and on when to blame disturbing factors (such as air resistance) for a theory’s failure to perfectly match with observation.

  Nevertheless, according to a 1627 letter from Gianbattista Baliani to Benedetto Castelli, Galileo had a powerful argument favoring his proposal over these rivals.10 His argument was that if one of these rivals holds for time intervals expressed in one particular unit (say, in seconds), then that proposal will not hold for time intervals expressed in another unit (say, in minutes). If we take the distances given by these proposals for successive one-unit intervals

  Fabri: 1s, 2s, 3s, 4s, 5s, 6s …

  Le Cazre: 1s, 2s, 4s, 8s, 16s, 32s …

  and switch to a new unit of time that is twice as long as the original, then we find the distances fallen in these new intervals to be

  Fabri: 3s (= 1s + 2s), 7s (= 3s + 4s), 11s (= 5s + 6s) …

  Le Cazre: 3s (= 1s + 2s), 12s (= 4s + 8s), 48s (= 16s + 32s) …

  These distances do not fit the proposals. For example, if the proposal is that the distances grow as the natural numbers, then that proposal is violated when we switch units since the distances 3s, 7s, and 11s do not stand in the ratio of 1 to 2 to 3.

  Thus, if one of these rivals to Galileo’s proposal holds in one system of units, then it will not hold in certain other systems of units. Galileo’s point, expressed in today’s terminology, is that neither rival to his proposal is “dimensionally homogeneous.” Roughly speaking, we can define “dimensional homogeneity” as follows:

  Relation R is “dimensionally homog
eneous” exactly when it is a broadly logical truth that if R holds in one system of units, then R holds in any system of units for the various fundamental dimensions (e.g., length, mass, time) of the quantities so related.11

  Of course, a relation can hold without being dimensionally homogeneous. For example, on a given date it may be that my son’s weight equals my age—but this relation holds only if my son’s weight is measured in pounds and my age is measured in years. The relation is therefore not between my son’s weight and my age themselves, but rather between their measures in a particular system of units.

  Part of the meager tacit background belief common to all parties during the crisis of terrestrial physics in Galileo’s time was that the relation among the distances traversed in successive equal time intervals by a body falling freely from rest is independent of the unit of measure. It is a relation among the distances themselves, not between their measures in some particular privileged unit. Presumably, neither Fabri nor Le Cazre ever bothered to say explicitly that they believed the relevant relation to be dimensionally homogeneous. But neither had to; it was understood. As far as I know, neither specified some particular units as having to be used to measure distance or time. Galileo argued that since neither of these proposals is dimensionally homogeneous, the relation in question cannot be given by either of them.