Scientific Methodology
17
Scientific Methodology
At the outset it should be stated that there is no “scientific method,” no formula with five easy steps guaranteed to lead to discoveries. There are many methods, used at different stages of inquiry, in widely varying circumstances.
Ian Barbour, Issues in Science and Religion
1—Introduction
Undoubtedly the most important influence shaping the modern world is science. People who lived during the American Civil War had more in common with Abraham than with us. From space travel and nuclear power, anesthesia and organ transplants, to DNA research and lasers, ours is a world of modern science.
If Christians are going to speak to the modern world and interact with it responsibly, they must interact with science. And if believers are going to explore God’s world by means of science and integrate their theological beliefs with the results of that exploration, they need a deeper understanding of science itself. What is science? Is there such a thing as the scientific method, and if so, what is it? Are good scientific theories at least approximately true, or are they merely useful fictions? How should theology and science interact? Is creation a science or a religion? What about theistic science? Queries of this sort are philosophical, and the next four chapters take us on a philosophical journey that explores science. This chapter begins that journey by focusing on scientific methodology. First, we will investigate the nature of philosophy of science and its relationship to science itself; second, we will examine key issues involved in clarifying scientific methodology.
2—Science and the Philosophy of Science
Philosophy is, in part, a second-order discipline that studies the assumptions, concepts, and argument forms of other disciplines including science. By contrast, science is a set of first-order disciplines. In philosophy of science, we investigate questions of philosophy about science. Philosophers and historians of science, and not scientists themselves, are authorities trained to deal with these types of questions. In science, we investigate questions of science about a specific realm of scientific study. Here, scientists are the authorities.
Here are some of the types of first-order questions scientists formulate: What is a covalent bond and how does it work? What is the structure of a methane molecule? What makes an ecosystem stable? How does the holding relationship between mother and young infant affect later childhood development? Where will the moon be on November 1, 2062? By contrast, here are some second-order questions—philosophical questions—about science: What is science and are there clear necessary and sufficient conditions that some intellectual activity must have for it to count as science? Is there such a thing as the scientific method and, if so, what is it? How do scientific theories explain things? How do observational data confirm a theory? If a scientific theory is a good one (e.g., makes accurate predictions, harmonizes with what we observe), does that mean that the theory is at least approximately true and that the unseen, theoretical entities (e.g., electrons) postulated by that theory really exist?
Some questions are more complicated. Formulating and answering these questions defy simple characterization and are best seen as a joint venture between science and philosophy. For example, the question of whether electrons exist is, first of all, a philosophical question. As we will see in chapter eighteen, many scientists and philosophers today and in the past have embraced some form of antirealism. Some antirealists deny the proposition that the theoretical entities postulated in good scientific theories are real; other antirealists merely claim that the success of a scientific theory does not provide us with sufficient justification for thinking that the theoretical entities in that theory are real, even though they may actually exist. The notion of an electron, they claim, may merely be a pragmatically useful fiction, allowing us to make predictions or develop technology. On the other hand, if scientific realism is adopted, then we should believe in the existence of the entities postulated in good theories. If this is our view, then we should believe in electrons if electron theory is judged by scientists to be a good theory. And this largely will be a scientific matter.
In sum, great care must be given to distinguish issues in the philosophy of science from those in science. And for issues that directly involve both disciplines, we should try to clarify the scientific and philosophical aspects of those issues.
There are two very different and competing approaches to the philosophy of science. The first is called an external philosophy of science (EPS). In this view, science itself is the object of study, and one applies a general philosophical understanding of reality (metaphysics), knowing (epistemology), and logical structure to episodes of science, evaluating the episodes as good or bad science. Philosophy is seen as a normative discipline that justifies the presuppositions of science and evaluates certain scientific claims in light of what we already have reason to believe from metaphysics and epistemology. For example, advocates of EPS deny that epistemology can be naturalized (e.g., reduced to descriptive studies in psychology) by claiming that there is a difference between asking what the psychological processes are that we go through to form our beliefs and asking the normative question about what justifies a particular belief as being rational. This is the traditional approach to the philosophy of science, and it recognizes that (1) claims about reality and knowledge within science already presuppose that there is knowledge and reality in the first place, (2) scientific assertions that some proposition is true or rational must conform to and not conflict with general features we already know about rationality and reality from philosophy, and (3) philosophy is primarily a normative discipline and science is primarily a descriptive discipline.
By contrast, a recent view in the philosophy of science, developed by thinkers like W. V. O. Quine and Wilfrid Sellars, is called an internal philosophy of science (IPS). In this view, philosophy is a branch of science. There is no difference in kind between philosophical and scientific questions but only one of degree—usually, philosophical questions (e.g., What is reality in general?) are broader than scientific ones (e.g., What is matter?). For example, epistemology is a branch of psychology, evolutionary biology, and neurophysiology. Moreover, science is its own justification and does not need evaluation from some alleged higher viewpoint. In short, science is the measure of all things. The task of philosophy of science is to describe carefully just how scientists do their job and to clarify scientific language and activity. Philosophy of science is primarily linguistic and descriptive.
The inadequacies of IPS have kept most philosophers from accepting it. IPS assumes that we can already recognize the difference between good and bad science in our attempt to describe science, but such a recognition will involve distinctively philosophical assessment. Moreover, IPS begs the question against skeptics that ask why we are justified in accepting the cognitive authority of science in the first place. IPS merely asserts the epistemological authority of science, and this is question-begging. Finally, IPS breaks down the distinction between normative and descriptive issues. Descriptive scientific questions about how we do, in fact, form our beliefs are very different from and presuppose answers to normative philosophical questions about how we are justified in trusting beliefs in general.
In addition to the differences between science and philosophy of science, as well as the debate between external and internal philosophies of science, there are three main areas within philosophy of science as a branch of study. First, there is the epistemology of science, which investigates the process of discovering scientific laws and theories, how we use those laws and theories to explain things, and how laws and theories receive confirmation from various sources like successful predictions. Second, there is the ontology of science, which focuses on the realism-antirealism debate. Should good scientific theories be interpreted as true or approximately true descriptions of the theory-independent world, and/or should we believe in the existence of the theoretical entities postulated in those theories? Or should we interpret the success of good scientific theories in ways that do not require commitment to the existence of the theoretical entities in those theories? Finally, there is the philosophy of nature: given that we accept scientific realism, how should our scientific beliefs about what is real factor into our broad worldview about reality in general? The first of these areas occupies the remainder of this chapter; the second will be covered in chapter eighteen; the third was one of the topics surveyed in chapter two, the section on metaphysics, and will also be among the concerns in chapter twenty.
3—The Epistemology of Science: Scientific Methodology
There is a fairly widespread belief that there is something called the scientific method that can be characterized in a clear manner and that separates science from other disciplines. For example, the following statement, taken from a widely used high school biology text, is typical: “Scientists use the scientific method in attempting to explain nature. The scientific method is a means of gathering information and testing ideas. . . . The scientific method separates science from other fields of study.”
In this section, our exploration of scientific methodology will lead us to two discoveries: First, there is no such thing as the scientific method, but rather there is a cluster of practices and issues that are used in a variety of contexts and can be loosely called scientific methodologies. Second, various aspects of scientific methodologies are used in disciplines outside science. These discoveries become apparent when we take a philosophical look at scientific practice. We will begin by examining and rejecting a view of scientific methodology called inductivism. Then we will look at different aspects of a more adequate, eclectic view of scientific methodology.
3.1 Inductivism
Inductivism is a view of the scientific method made popular in the nineteenth century and usually associated with the ideas of Francis Bacon (1561–1591), J. F. W. Herschel (1792–1866), and John Stuart Mill (1806–1873), though the actual description of scientific methodology by these three figures is much more complicated than the sort of inductivism to be described in this section. Inductivism is an entire view about scientific methodology, and should not be confused with induction itself, which is a form of inference wherein the truth of the premises does not guarantee but only supports the truth of the conclusion to one degree or another. As it came to be understood by the middle of the twentieth century, inductivism is a view of scientific method wherein scientists are seen as starting with unbiased observations of facts, progressively piling up more and more facts by means of those observations, generalizing them by enumerative induction into laws, combining these generalizations into broader and broader generalizations by piling up more facts, and, finally, arriving at various levels of scientific laws whose contents are nothing but the facts in general form. Inductivism pictures the scientific method as in figure 17.1.
Figure 17.1. The inductivist view of scientific method
As figure 17.1 shows, an inductivist view of the scientific method has two main components: the psychology of discovery and the epistemology of justification. The psychology of discovery refers to a description of the temporal, psychological process that an individual scientist or community of scientists goes through in a step-by-step fashion to form laws and theories. This is how scientists should do their work if they are to follow the scientific method according to inductivism. The scientist starts with observations and experiments and proceeds to the formation of laws and theories by enumerative induction, that is, by inductively deriving generalizations from past observations.
The epistemology of justification refers to the normative, logical structure by which a scientist or community of scientists justifies scientific laws and theories. Inductivism implies that a scientific law or theory is justified only if the evidence in favor of that law or theory fits the inductive scheme already mentioned. Scientists form and test laws and theories by (1) starting with observations without any bias or prior guesswork as to what is important or unimportant to be observed, (2) observing and analyzing the facts gathered in step 1 so as to classify them in different ways, (3) inductively deriving generalizations from this classification of facts, and (4) testing these generalizations by further observations and experiments and forming higher-order generalizations. Scientific knowledge is a conjunction of well-attested facts that grows by the addition of new facts that usually leaves previous facts unaltered. One’s belief in the plausibility of a law grows in proportion to the number of observed positive instances of the phenomenon described in it.
An alleged example of inductivism is the work of Gregor Mendel (1822–1884), the scientist-priest who studied the laws of inheritance by focusing on peas. Mendel studied different traits of pea plants that appeared in two different forms: seed texture (smooth/wrinkled), seed color (yellow/green), and stem length (long/short). He and his assistants grew pea plants, crossbred them, and counted the number having certain characteristics in each generation. Some of Mendel’s results appear in table 17.1.
Table 17.1. Some of Mendel’s results
Parental Characters
1
2
2
Smooth × Wrinkled seeds
all smooth
5474Sx:1850W
2.96:1
Yellow × Green seeds
all yellow
6022Y:2001G
3.01:1
Long × Short stems
all long
787L:277S
2.84:1
*F1 and F2 are the first and second generations, respectively.
On the basis of these and other observations, Mendel allegedly derived, by pure induction, the following law: In the second generation, the ratio of the most frequent to the less frequent of two traits that correlate with each other is 3:1. How would Mendel or any other scientist derive and justify this law according to inductivism? First, one would simply start with a sample of peas, call it sample X, grow and crossbreed them, and observe the results with no prior guess as to what to observe or not to observe. After recording the results of observing sample X, one would do the same for samples Y, Z, and so on. Next, one would observe the results of X, Y, Z, and other samples and make a generalization for the type of peas used in these specific samples. Third, one would repeat this procedure for peas of another type, and then for other types of vegetables, and finally for other types of living organisms, until one reached the broadest generalization possible covering all living organisms and all cases of two observable traits that correlate together into the 3:1 ratio.
Although inductivism continues to persist in the popular conception of science and even in the minds of many scientists, very few, if any, philosophers of science accept it. It fails to grasp the variegated texture of science, and the objections against it are severe.
First, one cannot merely start with observations without some guiding hypothesis or background assumptions, however tentatively they are held, to guide in deciding what is and is not relevant to observe. Pure, presuppositionless observations are a fable in science, and scientists almost never start with observations. Usually, they start with a problem to be solved and a set of assumptions and hypotheses about what is and is not relevant to observe. In the Mendel case above, Mendel and his associates did not observe and record the position of the moon, the weather in Boston, or the color of shoes they wore as they did their experiments. These and a vast host of other factors were not observed because the experimenters had enough of an idea about pea breading to know that these factors were irrelevant but others (the color and texture of seeds, the length of stem) were relevant. These judgments were brought to the task of observing; they were not simply derived from observations.
Moreover, this same point can be made about classifying and arranging observations of different samples of peas and other living organisms. Without some framework about what is going on, there is no way to decide what factors should serve as the basis on which to classify particular facts.
Third, scientific laws are not formed, accumulated, or justified by the progressive accumulation of brute, uninterpreted observational data. Rather, these processes involve an interpenetrating mixture of observation and theory in ways that defy simple characterization. In fact, sometimes a shift in theory can turn seeming facts into falsehoods, as Rom Harré has pointed out:
For instance, consider the history of the determination of the atomic weights. What were the facts? Under the influence of Prout’s hypothesis some chemists considered that the discrepancies between integral values for the atomic weights of the elements [e.g., chlorine is approximately 35.5] were errors, since Prout had maintained that all elemental atoms were combinations of whole numbers of complete hydrogen atoms, and hence their atomic weights had to be integral numbers by comparison with hydrogen. Those who did not accept or had abandoned Prout’s hypothesis were inclined rather to suppose that the non-integral weights were the facts, that is a genuine measure of a natural phenomenon. What the facts were depended in part on whether one held or did not hold to a particular theory.
In the case just cited, what counted as a fact depended on the prior acceptance of a theory, and this type of fact-theory interaction is not allowed by inductivism. But since reasoning of this type has been an appropriate part of science during its history, inductivism must be judged inadequate as a total account of scientific methodology.
Fourth, from a finite set of observational data, there will always be a potentially infinite number of general laws consistent with those data. Theories are underdetermined by observational data. The underdetermination of theories by data refers to the fact that those data, considered by themselves, do not unambiguously select one and only one law. But if we must operate within the confines of inductivism, how is one to choose among these alternative laws? One way to do so is to pick the most simple generalization consistent with the data. But this strategy is not part of inductivism (simplicity is not derived from observations but brought to them as a background assumption). Moreover, there are different, rival understandings of what simplicity means; resolving this debate is part of spelling out scientific methodology, and these issues go beyond the picture of science contained in inductivism.
Several other problems have been raised against inductivism: the problem of induction (what justifies the inference from “All observed A’s are B” to “All A’s whatsoever are B”); the difficulty of deciding between accidental generalizations about a phenomenon that merely happens to be true (e.g., plants grow from the sun’s warmth) and lawlike generalizations that express real necessities in nature based on a background theory of the true nature of the phenomenon in question (e.g., plants grow from the sun’s light by photosynthesis); and the fact that scientists try not merely to describe phenomena by generalizations but also to explain them with theories about underlying mechanisms, often unobserved or even unobservable, that account for observational generalizations.
In sum, inductivism is one view about the scientific method, but it turns out to be inadequate. A better way of picturing scientific reasoning is called the hypothetico-deductive method, advocated by Carl Hempel, among others. Roughly, this view sees scientists as, in one way or another, forming and putting forth a hypothesis, deriving test implications from it (along with what are called boundary conditions), then seeing if observations corroborate with the hypothesis. For example, if we accept the ideal gas equation (PV = nRT, where P is pressure, V is volume, T is temperature, n is the amount of gas in terms of the number of moles of gas, and R is a constant), and if we have as boundary conditions (particular values for P and V that may be plugged into the equation) P1 and V1 for a specific amount of a gas n1, then we derive that T1 ought to be its temperature. We can then make observations to see if this is supported by the data we gather. But as we shall see, even the hypothetico-deductive method does not capture everything that a scientist does. In order to probe more deeply into scientific methodology, let us turn to an examination of a more eclectic model of scientific practice.
3.2 An Eclectic Model of Scientific Methodology
There are seven aspects of the proposed eclectic model to explore: (1) the formation of scientific ideas, (2) the nature of scientific questions and problems, (3) the use of scientific ideas and scientific explanation, (4) the nature of scientific experiments, (5) the testing of scientific ideas (scientific confirmation), (6) the nature of scientific ideas (laws and theories), and (7) the aims and goals of scientific ideas.
Before we look at each of the seven areas, an initial observation should be made. If we are going to state a view of scientific methodology in a detailed and comprehensive way, then we will need to describe positions regarding various areas of debate within these seven aspects. As we will see, there are different ways scientific ideas are formed, different problems scientists try to solve, different ways scientific ideas are used to explain things, and so forth.
Further, major debates exist within each of these seven areas. For example, not everyone agrees about how scientists should use their laws and theories to explain things. This means that several different views of scientific methodology are possible, if those views are thorough enough to offer positions in these seven areas of discussion. While some views we will look at are not as plausible as others, nevertheless, each is logically consistent with the practice of science, and this supports an eclectic model of scientific methodology; it shows that scientific methodology is a cluster of different methodologies and not one single method worthy of the title “the scientific method.” In order to get a feel for the variegated texture of scientific methodology, we will examine each of these seven aspects of an eclectic model of scientific methodology in order.
1. The formation of scientific ideas. Area one is sometimes called the psychology of discovery and refers to the process by which individual scientists or communities of scientists discover and form their ideas. It is generally agreed that there is no formalized method, no step-by-step procedure that characterizes the process of scientific discovery. Sometimes scientists discover things by accident. On other occasions they generate their ideas in more bizarre ways. It is well known, for example, that F. A. Kekule (1829–1896) came up with the hexagon formula for the benzene ring by having a trancelike vision of a snake attempting to chase its own tail and thus curving into such a ring!
More frequently, however, scientists form their ideas in one of two ways. First, scientists generate ideas by a creative process of educated guesswork known as adduction or abduction. Adduction refers to the process of inventing a theory to explain observed facts. Science is a craft, and after a scientist has worked in an area for a while, this personal involvement allows the scientist to develop savvy about that area, a sense of tacit know-how. Part of this know-how is the ability to see things in a certain way, to intuit patterns of phenomena, and, by the use of creative imagination, to adduce a conceptual web to explain those patterns. Often a scientist cannot say how it was that he or she came up with a theory. This same sort of tacit knowledge is used by auto mechanics, judges, biblical exegetes, and others who use their knowledge of a field to weigh things and adduce a solution to a problem.
There is one difficulty with this way of generating a scientific idea. Sometimes, after working in an area for a long time, a scientist is indoctrinated so heavily into a prevailing way of seeing a realm of data that he or she can distort data by reading things into them and cannot see things in another way. That is why Thomas Kuhn pointed out that, often, discoveries in science are made by newcomers or by people working in another branch of science or another field altogether, because they have the cognitive distance from the prevailing paradigm necessary to see things differently.
Adduction is a method of discovery, as it were, “from below,” that is, from a sense of savvy with data in a certain realm. There is a second way scientists have frequently discovered their ideas: a method “from above.” Here, scientists are guided in their search for a theory from a broad picture of how the world is or how certain data should turn out. This broad picture can come from metaphysics, mathematics, theology, or other domains. For example, some scientific discoveries have been made in light of the scientist’s prior commitment to the idea that correct equations should be mathematically beautiful or elegant. Again, Copernicus was uncomfortable with Ptolemy’s model of planetary motion because it departed from the Platonic ideal of uniform circular motion around a center. As a Platonist, Copernicus was guided “from above” by the geometrical idea that motion that is circular is more perfect and thus the planets seek to move in circular motion. The point is not that he was correct, but merely that he was guided in the process of discovery in this way.
Sometimes metaphysical views guide a scientist. Some scientists have had a metaphysical commitment against action at a distance and this has guided them in a search to find some intervening mechanism (e.g., an exchange of gravitons) between two bodies that allowed gravity to act on contact and not at a distance. Sometimes theological beliefs guide research and help generate scientific ideas. Leibniz (1646–1716) derived Snell’s law (a ray of light traveling obliquely, i.e., neither parallel nor perpendicular, from one optical medium into another is refracted at the surface in such a way that the ratio of the sines of the angles of incidence and of refraction is a constant for any media pair: n1sin θ1 = n2sin θ2) from his metaphysical principle that nature always selects the easiest, most direct course of action given a set of alternatives. And he derived this principle from his theological conviction that God created the world such that maximum simplicity and perfection should be realized. Linnaeus and other early taxonomists were guided in their taxonomic work by the theological idea that God created organisms in distinct, classifiable types. James Clerk Maxwell derived his field picture of light, in part, from his theological convictions about the Trinity and the incarnation. Thus fields like mathematics, philosophy, and theology can enter into the very fabric of scientific methodology by providing criteria for guiding the psychology of discovery from above.
2. The nature of scientific questions and problems. Often, scientists try to solve problems by answering at least three types of questions. First there are “what” questions: What does the fossil record look like? What is the half-life of uranium? Here scientists try to establish facts even if they cannot, in principle, explain those facts. For example, scientists could try to establish what the rest mass is of some alleged ultimate particle, even if they do not believe there is a further explanation for why the rest mass is some specific value because the particle is taken as ultimate. Second, scientists answer “why” questions: Why do metals expand when heated? Here the focus is on stating the cause for some phenomenon (e.g., the efficient or final cause). Third, scientists answer “how” questions: How does light dislodge an electron from the surface of a metal? “How” questions are requests for a description of how it is that some cause accomplishes an effect.
Two things should be noted about these questions. First, disciplines outside of science ask and answer very similar types of questions. Second, scientific methodology is not exhausted by a search for answers to “how” questions. They also answer “what” and “why” questions.
Scientists ask questions and try to solve problems. Generally speaking, these problems come in two broad types: empirical problems and conceptual problems. Scientific theories contain two types of terms: observational terms, which refer to what is directly observational (e.g., “is red,” “floats,” “is longer than”), and theoretical terms, which express theoretical concepts (e.g., “gene,” “electron,” “kinetic energy”). Empirical problems focus on observational difficulties with a theory; conceptual problems focus on theoretical, conceptual difficulties with a theory.
In general, an empirical problem is anything about the world of observations that strikes us as odd and in need of explanation. Why do the tides move as they do? Why are there gaps in the fossil record? These are examples of empirical problems. Suppose we have two rival theories that are competing for our allegiance. Such a situation gives rise to three types of empirical problems: unsolved problems (those not adequately solved by any rival theory), solved problems (those that are solved by all rivals, perhaps in different ways), and anomalous problems (those solved by one rival but not another). Of the three, anomalous problems carry the most epistemological weight. However, though important, anomalies are not necessarily decisive against a theory. The strength of anomalous problems is a function of several things: How many of them are there, and how important or central are they for the theory in question? How well does the rival solve them? How strong are the various rival theories in light of factors other than their ability to solve these anomalies? It is worth noting that the epistemological impact of anomalies for scientific theories is parallel to the impact of anomalies in areas outside science. For example, just exactly when is it no longer reasonable to believe in biblical inerrancy in light of anomalous data outside Scripture or problem passages? Answering this question is no easier in theology than in science.
Scientists also try to solve conceptual problems. In general, these are difficulties with the conceptual aspects of a theory, and they come in two types. First, there are internal conceptual problems. These arise when the concepts within a theory appear to be vague, ambiguous, circularly defined, or contradictory. For example, some have argued that the wave-particle nature of electromagnetic radiation is contradictory, that the evolutionary pathway from reptile scales to bird feathers through a series of slightly changed intermediaries is unclear and vague, that the use of imaginary time by cosmologists such as Stephen Hawking is unintelligible, and that “survival of the fittest” is circularly defined. The point is not that these objections have been decisive, but rather that they are examples of internal conceptual problems.
Second, there are external conceptual problems. An external conceptual problem arises for some scientific theory T, when T conflicts with some proposition of another theory T’, when T ‘ is rationally well founded, regardless of what discipline T’ is associated with. T may be logically inconsistent with T’ or they may conflict in a lesser way by being jointly implausible and not mutually reinforcing, though still logically consistent. For example, if T is an oscillating model of the universe that implies an infinite past for the universe and T’ is a philosophical argument implying that the past must be finite, then T’ is an external conceptual problem for T that advocates of T must solve. Again, if naturalistic evolutionary theory implies that living organisms are merely physical (because they are wholly the result of a physical process operating on physical materials), then if philosophical arguments exist for some form of mind-body dualism, these constitute external conceptual problems for naturalistic evolutionary theory. If theologians have rationally justified theological arguments against, say, views of human origins that claim that humans originated in China, then these arguments would be an external conceptual problem for that scientific theory.
Science has never exhausted the rational, and science has directly interacted with other disciplines, in epistemologically positive or negative ways, throughout the history of science. External conceptual problems provide examples of the fact that science is not merely complementary to fields like philosophy and theology, because in such problems the former directly interacts with the latter. More will be said in chapter nineteen about what it means to say that science is merely complementary to philosophy and theology. But for now, we may understand such a claim to mean that a scientific description of some phenomenon (say, the formation of water from hydrogen and oxygen) and a philosophical or theological description of that phenomenon (God creates water from hydrogen and oxygen) are two different, noninteraction approaches to the same phenomenon. Such a claim may be quite proper in certain cases (the water example), but the presence of external conceptual problems is one of several factors about scientific methodology that show the inadequacy of a complementarian approach to the integration of science with theology and philosophy if that approach is understood as the final word on the subject.
3. The use of scientific ideas and scientific explanation. Scientists use laws and theories to explain things. Some philosophers like Thomas Kuhn have argued that there is no basic type of scientific explanation, but rather as many different models of scientific explanation as there are branches of science. So, very little in general can be said about scientific explanation. Others disagree and have argued that the vast majority, perhaps all of what we would count as scientific explanation, fits into one or both of two different options. First, there is the covering-law or inferential model of explanation made popular by Carl Hempel and Ernest Nagel. According to this view, two factors make an explanation a scientific one: the logical form of the explanation and the nature of the explanation’s premises. The terms explanans and explanandum mean, respectively, “that which does the explaining” and “that which is to be explained.” Regarding logical form, scientific explanations explain some phenomenon by giving a correct deductive or inductive argument for that phenomenon in one of the following three ways:
In each case, the thing to be explained (the explanandum) is explained or “covered” by inferring it from premises (the explanans), the first of which is a general law and the second of which is a statement of initial conditions. In the deductive-nomological version, the explanans contains only universal generalizations and the argument is deductive. In the deductive-statistical version, the explanation is a deductive argument that contains at least one statistical generalization in the explanans. In the inductive-statistical version, the explanation is an inductive argument (signified by the double line below C1) that includes at least one statistical generalization in the explanans. In each case, a good scientific explanation of some explanandum E will embody one of the three logical forms above.
A second feature of the covering-law model of scientific explanation focuses, not on logical form, but on the nature of the premises. The explanans of scientific explanation must be scientifically testable in some way—for example, by observation and experiment—and some would add, they must be true, lawlike generalizations.
Most agree that the covering-law model captures at least part of the nature of scientific explanation. Nevertheless, it has had its detractors. Some have claimed that the model has certain conceptual problems (e.g., What does it mean to test something by observation and experiment? Don’t fields outside science do this as well? Do scientific generalizations have to be true or only thought to be true and well-confirmed? Are false scientific explanations still scientific explanations? What is a lawlike generalization, and how does one differ from an accidental generalization?).
Others have pointed out that the model leaves out important aspects of scientific explanation. For example, it is hard to see how the covering-law model is explanatory in itself. The statement “This x has F because all x’s have F and this is an x” merely postpones explanation by inviting the question “Why do any x’s have F in the first place?” An explanation of this latter question will use new concepts embedded in models of what x is like (e.g., metals conduct electricity because little entities called electrons are free in metals). So the covering-law model is not sufficient for an explanation. Covering-law explanations are really just generalized descriptions of what happens, but they are not actually explanations of why things happen.
These problems have led some to embrace a second model: the realist, causal model of explanation. Advocates of this view point out that, according to the covering-law model, the essence of scientific explanation is describing general correlations among observational phenomena that allow us to deduce some result. An example of this would be the ideal gas equation, PV = nRT (where P is pressure, V is volume, T is temperature, n is the number of moles of gas present, and R is the ideal gas constant). The ideal gas equation describes the regular correlations of observational phenomena (P, V, T) in gases and allows us to deduce, say, the temperature of a gas from knowledge of its pressure and volume. But this so-called explanation is just a description of how phenomena are related, not an explanation of why they behave as they do. An explanation of why the temperature of the gas is what it is, however, will use a model that pictures unobserved entities and processes postulated as being causally responsible for observational correlations. On this view, the behavior of a gas is explained by picturing gases as groups of molecules analogous to tiny billiard balls that engage in elastic collisions, travel through space, bounce off container walls, and travel in a velocity proportional to the temperature of the gas.
On the causal, realist model, scientific explanation goes beyond what can be observed by building models of theoretical entities and processes that are causally responsible for observational phenomena. It is interesting to note that philosophers like Bas C. van Fraassen and Stanley Jaki have shown that this type of scientific explanation represents a very close parallel to arguments for God’s existence used in natural theology. For example, one form of the design argument starts with various types of design as observational phenomena to be explained and postulates the existence of a personal designer who is causally responsible for these phenomena.
Critics of the realist, causal model are often empiricists of one sort or another who claim that commitment to a scientific theory only involves commitment to the fact that a theory will make observationally accurate predictions. Such a commitment does not include believing in unobservable, theoretical entities that, they claim, are merely excess, metaphysical baggage for a scientific theory. This criticism will be part of our focus in the next chapter.
We have seen that there are different views about how scientific explanations should work. There are also different kinds of explanations used in science. There are compositional or structural explanations: the properties of an object are explained in terms of the properties or structural relations of its parts. There are historical explanations, which explain the properties and existence of an object in terms of the temporal development and history of the object and its ancestors. With functional explanations the capacities of an object are explained in terms of the function they play in some system—the function of x (the heart) is to do y (pump blood). Transitional explanations explain a change of state in some object in terms of some disturbance in the object and the state of the object at the time of the disturbance. Finally, there are intentional explanations, which explain the behavior of an organism or the existence of some state of affairs in terms of the beliefs, desires, fears, and intentions of that organism.
Before we move to the next area of scientific methodology, one more issue should be mentioned. It is sometimes claimed that the notion of God as a creator or designer who “intervenes” in nature cannot serve as part of a scientific explanation. This claim is usually defended by some sort of covering-law model of explanation, coupled with the idea that science must explain things by using natural laws and, obviously, a miraculous act of God is not part of a natural law.
But this claim is simply false. Scientists do not merely explain by using natural laws. They also explain by citing causal entities, processes, events, and actions, as we have seen. For example, cosmologists explain certain aspects of the universe, say its rate of expansion, not only by using natural laws, but also by citing the big bang as a single causal event. Now scientists in some branches of science, for example, the search for extraterrestrial intelligence, archaeology, forensic science, psychology, and sociology, use personal agency and various internal states of agents (desires, willings, beliefs, intentions) as part of their description of the causal entities and processes cited in their explanations. There is no reason that can be derived from the nature of scientific explanation for why the same type of argument could not be used to explain some sort of phenomenon in biology or a related field, say, in explaining the origin of life. Whether the explanation would be a good one would, of course, largely be a scientific question.
In this context, many scientists and philosophers make a distinction between empirical science and historical science. These broad categories of science differ in the kinds of questions asked and the explanations and methods used to answer them. In general, empirical science (also called “inductive,” “nomological,” or “operation” science) focuses on how the natural world operates in a repeatable and regular way. The study of acid-base chemical reactions is an example. Historical science (sometimes called “origin” science) focuses on single, past events (e.g., the death of the dinosaurs, the origin of first life) and attempts to explain how things came to be or why some event happened. This distinction relates to our present discussion about the notion of an act of God in science as follows: it can be argued that an appeal to an intelligent intervention of God is not appropriate in empirical science (unless God does, in fact, regularly “intervene” to do some repeatable action, e.g., as he does in regeneration), but it can be appropriate in historical science to explain the origin of life and so forth.
4. The nature of scientific experiments. Scientists make observations and perform experiments. An observation occurs when an observer stands outside the course of events being observed and does not manipulate them. An experiment occurs when the experimenter intervenes in the course of events by altering nature in some way to allow the experiment to take place. When scientists make observations and perform experiments, they often use instruments of at least three different types. First, there are instruments used in making measurements. Some of these are self-measurers, that is, they are examples of the thing being measured (e.g., a meter stick). Some are non-self-measurers that use the effects of some phenomenon to measure the cause (e.g., thermometers directly measure expansion and infer temperature). Second, there are instruments that extend our senses such as microscopes, telescopes, amplifiers. Third, there are instruments that isolate phenomena to allow them to be studied independently from their environment.
In using these instruments, scientists make various assumptions. Many of these assumptions are quite reasonable, and some may not be. But in any case, the use of an instrument to “observe” or measure something illustrates that when a scientist claims to observe something, he or she is often interpreting data in light of a large number of theoretical background assumptions about the thing being observed and the instruments used to observe it.
5. The testing of scientific ideas: scientific confirmation. The fifth aspect of scientific methodology includes issues of testing and justifying scientific laws and theories. Two questions are of central importance in understanding this aspect of scientific methodology: (1) How do positive test results from observation and experiment lend support to a scientific law or theory? (2) What factors are involved in the claim that a scientific law or theory is rational? Let us look at these questions in order.
If a scientific theory predicts or in some other way implies that certain things should be observed, and if those things are observed, how does this help to confirm the theory in question? There is no universal agreement among philosophers about how to answer this question. In fact, many philosophers believe that assessing a scientific theory is so multifaceted and complicated that no conclusions whatever about the truth or falsity of a theory follow from positive test results. Others disagree and claim that positive test results do, in fact, lend support to a theory or at least show that the theory has not yet been falsified. For those who hold to the value of positive test results, two major schools of thought exist: falsificationism and justificationism.
The main advocate of falsificationism has been Karl Popper (1902–1994), for whom positive test results only show a theory to be possibly true. Those results do not make the truth of a theory any more probable. Consider the generalization “All ravens are black.” No amount of confirming instances of black ravens shows that the generalization is true because we could always find a white raven in the future. The discovery of just one white raven would show the generalization to be false. Therefore, Popper concludes, scientific theories should be bold, risky conjectures that we try to falsify. Positive test results show that the theory has not been falsified yet, and if we test a theory frequently and if it passes all our tests, then the theory is corroborated. This does not mean that we have positive evidence for the theory, but only that we have repeatedly tried to falsify the theory, we have so far failed, and thus the theory is possibly true.
Justificationists like Rudolf Carnap (1891–1970) and Carl Hempel (1905–1997) disagree. There are different forms of justificationism, but advocates agree that positive test results increase the degree of probability that a theory is true. The more positive test results and the less negative test results we have, the greater the degree of confirmation our theory possesses.
In sum, falsificationism is the view that positive test results only show that the theory has so far not been falsified and that it is possibly true. Justificationism is the view that in one way or another, positive test results increase the probability that the theory is true and give it positive support.
What about our second question: What factors are involved in the claim that a scientific theory is rational? To begin with, we need to distinguish between the rationality of pursuit and the rationality of acceptance. The rationality of pursuit refers to a case where a theory is relatively new, undeveloped, and untested, but nevertheless it appears to hold promise. In this case, one may not have enough justification to think the theory is true, but it could still be reasonable to do research in light of the theory. The rationality of acceptance, on the other hand, refers to a theory that has been around long enough to be accepted or rejected, and for some reason or other, accepting the theory is rationally justified.
Regarding the rationality of acceptance, what is it that makes a theory rationally justified? There is no simple answer to this question, and a number of factors are relevant to answering it. One factor is whether there is a viable rival theory. If not, then even if our theory has serious problems, it is all we have. If there are one or more rivals, then we must assess our theory by comparing its adequacy vis-à-vis those rivals.
Another factor is whether and to what extent a given theory possesses various epistemic virtues. An epistemic virtue is a normative property that, if possessed by a theory, confers some degree of rational justification on that theory. An epistemic virtue increases the justification for believing a theory. Examples of epistemic virtues are these: theories should be simple, empirically accurate, predictively successful, fruitful for guiding new research, capable of solving internal and external conceptual problems, and should use certain types of explanations and not others (e.g., avoid action at a distance, appeal to efficient and not final causes).
The importance of epistemic virtues complicates the assessment of a theory’s rationality for at least two reasons. First, each of the epistemic virtues can be interpreted in more than one way. For example, what does it mean for a theory to be simple? Does it mean that a theory should have the lowest number of different kinds of entities, the lowest power of ten for the variables in its equations, or something else? One rival theory may utilize one interpretation of a specific epistemic virtue like simplicity, and another rival may use a different interpretation.
Second, two rivals may have different rankings of the relative importance of the virtues on the list above. For example, one rival may value simplicity over empirical accuracy, and the other rival may reverse this. Moreover, one rival may be successful regarding two or three virtues, and its rival may be successful regarding a set of different virtues. How are we to compare rivals in these cases? Only a case by case study can, at least in principle, answer this question.
In any case, the role of epistemic virtues in assessing the rationality of a scientific theory shows that a crucial-experiment model of theory assessment is naive. On this model, one rival predicts some phenomenon P and the other rival predicts not-P. You do a crucial experiment to see if P or not-P obtains, and you verify one rival and falsify the other. Unfortunately, things are not that simple when it comes to theory assessment in science—or in theology and other disciplines for that matter. This does not mean that we can never tell when a scientific theory is rational to believe; it just means that making such a judgment is not as straightforward as we are sometimes led to believe.
6. The nature of scientific ideas: laws and theories. So far we have been using the terms law and theory without defining them. Unfortunately, there is no universally accepted definition of either term. However, two remarks about scientific laws and theories can shed light on important distinctions used in this area of investigation. For one thing, there are three basic ways to distinguish a law from a theory. One way is to hold that a theory is roughly a hypothesis and, if it becomes well confirmed, can graduate to the status of a law. On this view, the only difference between the two is that a theory should be held tentatively and a law should be held firmly; that is, the difference lies in their relative degree of epistemological strength. While this way of speaking is fairly popular, it is the least helpful for understanding the nature of scientific methodology and thus is not widely used among philosophers of science.
A second way to distinguish a law from a theory focuses on their relative degrees of generality—a theory is broader in scope than a law. For example Kepler’s laws of planetary motion or Galileo’s law of free fall (s = 16t2 where s is distance and t is time) only hold for planetary motion (Kepler) or motion near the earth (Galileo). However, Newton’s laws of motion hold not only in these cases but more broadly in that they apply to the motion of other heavenly bodies and other forms of motion as well. Thus, on this way of distinguishing laws and theories, Kepler’s and Galileo’s statements about motion would be laws and Newton’s would be theories.
A third way to distinguish theories from laws is embraced especially by those who hold to some form of scientific realism (see chap. 18) and who hold to the realist, causal model of scientific explanation discussed earlier in this chapter. On this view, laws merely describe the lawlike regularities that are observed in nature, and theories explain those regularities by offering a model for the theoretical entities, structures, and processes thought to be causally responsible for those regularities. The ideal gas equation, PV = nRT, is a law, and the kinetic gas theory (gases are swarms of tiny particles that engage in elastic collisions and so forth) is a theory.
A second remark about scientific laws and theories involves the different ways that scientific laws can be classified. First, law can refer to a linguistic or conceptual entity. In this sense a law is something that a scientist can discover at a point in time; he can write it on a sheet of paper or have it in his mind. On the other hand, law can refer to a real disposition (e.g., hydrogen atoms will become positively charged if they lose an electron) or real relationship (e.g., an increase in the temperature of a gas will increase its pressure at constant volume) that obtains in the theory-independent world. In the former sense of law, it is appropriate to say that a law does not cause anything to happen, since it is a linguistic or conceptual entity. In the latter sense, however, laws do cause things to happen in the world by being among the factors that direct the way nature behaves. So whether a law can be said to cause something depends on how one is using the word law.
In addition, laws can be statistical (“The probability of an atom of U238 decaying during this time period is 50 percent”) or nonstatistical (“All copper expands when heated”). Further, good scientific laws are not always true. For example, the ideal gas law doesn’t hold for any gas in the real world, but it is still a helpful approximation, even though, strictly speaking, it is false.
7. The aims and goals of scientific ideas. Scientists and philosophers have surfaced several different aims or goals that are explicit or implicit ends scientists intend to reach when they formulate theories. Two different types exist: extrinsic and intrinsic goals of science. Extrinsic goals are the motives or reasons that scientists do science—for example, to glorify God, to exert power over nature, to protect the environment. More important for assessing the truthfulness or epistemic strength of a scientific theory are intrinsic goals. These goals are the epistemic virtues that scientists seek: simple theories, empirically accurate theories, and so forth. Further, part of understanding intrinsic scientific goals is how we should interpret a scientific law or theory that embodies various epistemic virtues and is, therefore, a “good” theory. Do scientists seek virtuous theories because they seek true theories that accurately describe the real world, or do they seek virtuous theories because such theories work and are useful fictions? Scientific realists adopt the former view of intrinsic scientific goals, and scientific antirealists adopt the latter. This debate will be the focus of the next chapter.
Chapter Summary
Philosophy of science is a second-order discipline of philosophy about science. Science is a first-order discipline about a specific realm of scientific study. There are two competing approaches to the philosophy of science. An external philosophy of science is a normative approach that applies general philosophical issues to the study and justification of science. An internal philosophy of science starts with good examples of science itself, is descriptive in nature, and sees philosophy as a branch of science. The three main areas of philosophy of science are the epistemology of science, the ontology of science, and the philosophy of nature. The remainder of the chapter focused on the epistemology of science.
There is no such thing as the scientific method. Rather, there is a cluster of practices and issues that are used as part of scientific methodology and in disciplines outside of science as well. Moreover, many who claim that there is something called “the scientific method” identify that method by the name inductivism. But inductivism is an inadequate approach to scientific methodology. A more eclectic model of scientific methodology includes various issues and debates in seven areas: the formation of scientific ideas, scientific questions and problems, the use of scientific ideas and scientific explanation, scientific experiments, testing scientific ideas and scientific confirmation, the nature of scientific laws and theories, the aims and goals of scientific ideas. Finally, our investigation of these seven areas of scientific methodology revealed several ways that theological and philosophical issues can enter into the very fabric of scientific methodology.
Checklist of Basic Terms and Concepts
adduction (or abduction)
anomalous problems
compositional or structural explanations
covering-law or inferential model of explanation
deductive-nomological
deductive-statistical
eclectic model of scientific methodology
empirical problems (solved, unsolved, anomalous)
empirical science
enumerative induction
epistemic virtue
epistemology of justification
epistemology of science
external conceptual problem
external philosophy of science (EPS)
extrinsic and intrinsic goals for science
falsificationism
first-order and second-order questions
historical explanations
historical science
hypothetico-deductive method
induction
inductive-statistical
inductivism
intentional explanations
internal conceptual problem
internal philosophy of science (IPS)
justificationism
laws versus theories
ontology of science
philosophy of nature
psychology of discovery
rationality of acceptance
rationality of pursuit
realist, causal model of explanation
transitional explanations
underdetermination of theories by data
“what,” “why,” and “how” questions