Inductivism

Fortunately, several philosophers noticed that Aristotle’s methodology was inadequate to describe the kind of research Galileo, Boyle, and other experimentalists were conducting. French philosophers René Descartes (1596 - 1650) and Blaise Pascal (1623 - 1662) were among the first to propose novel methods of scientific inquiry. However, the earliest attempt to outright replace Aristotle’s Organon with a wholly new theory of scientific methodology came from the English philosopher Francis Bacon (1561 - 1626). Bacon made no effort to conceal his purpose: he titled his work Novum Organum—the New Organon, first published in Latin in 1620.The full title of the book, in English, is The New Organon: or True Directions Concerning the Interpretation of Nature. A modern English translation is available here.

Bacon’s vision was not merely to describe what scientists were already doing, however. He recognized that a new understanding of empirical methodology was needed to guide scientific inquiry and unify the rapidly proliferating observational and experimental methods of the emerging natural sciences. He also believed—with more than a touch of egotism—that he could supply a radically new methodology that would transform the sciences and give humanity unprecedented power over nature:

Man’s control over things depends wholly on the arts and sciences, for we can’t command nature except by obeying her. A further point: it sometimes happens that one particular discovery is so useful to mankind that the person who made it and thus put the whole human race into his debt is regarded as superhuman; so how much higher a thing it is to discover something through which everything else can easily be discovered!Francis Bacon, The New Organon: or True Directions Concerning the Interpretation of Nature, translated by Jonathan Bennett (Book 1, section 129), available online here.

In that sense, Bacon’s proposal was not descriptive but prescriptive, outlining a new method by which scientists ought to conduct their reasoning and experimentation.For further discussion of Bacon’s project and its impact, see Davis C. Innes, Francis Bacon (Phillipsburg: P&R Publishing, 2019).

Bacon argued that science, done properly, involves inductive reasoning rather than deductive reasoning. Inductive reasoning is the kind of reasoning we employ when generalizing from patterns we observe, or when making predictions on the basis of our past experiences. Unlike deductive inferences, an inductive inference doesn’t guarantee the truth of the conclusion, even if the premises (starting assumptions) of the argument are true. Nevertheless, inductive reasoning can provide strong evidence supporting a conclusion. For example, when you conclude on the basis of your experiences that a bolt of lightning will be accompanied by a clap of thunder, the premises (that you just saw a lightning bolt, and that in your prior experience every such flash was accompanied by a subsequent thunderclap) do not guarantee the conclusion that you will soon hear thunder. It is logically possible for the lightning to occur without you hearing the sound. Nonetheless, given your prior experiences, the fact that you just saw lightning provides you with good evidence (i.e., a reason to believe) that you will soon hear thunder.

In his Novum Organum, Bacon suggested that a new kind of logic, distinct from Aristotle’s deductive logic, would enable scientists to derive scientific theories directly from the empirical data (measurements and observations) gathered through experimentation. Although inductive reasoning could not yield absolute certainty about the truth of any theory, it nonetheless makes science more rational and logical than other ways of learning about the world, Bacon argued, because scientist employ strict inductive rules of reasoning to reach their conclusions.

On Bacon’s view of scientific methodology, a good scientist begins by conducting many experiments, making careful observations of the results, and then follows strict rules of inductive reasoning to derive theories from those observations. This portrayal of science, which would later be called inductivism, supplies the following demarcation criterion:

• Genuine science uses inductive logic to derive theories directly from empirical data.

Because scientists infer theories via strictly logical rules of induction, moreover, inductivism suggests that science is superior to other ways of investigating the world. Unlike many non-scientific modes of inquiry, science is purely objective in the sense that it doesn’t depend on the biases, presuppositions, or assumptions of the scientist—or so inductivism seems to imply.

Scholars disagree on the extent to which Bacon’s methodological recommendations shaped scientific practice in subsequent years of the scientific revolution. My own impression is that inductivism had relatively little impact on the development of experimental methods in the natural sciences,As Brian Hepburn and Hanne Andersen observe in their Stanford Encyclopedia of Philosophy entry on Scientific Method, “It is hard to find convincing examples of Bacon’s method being put in to practice in the history of science.” but it was tremendously influential in shaping—or, rather, distorting—philosophical opinions about the essential characteristics of science. Variations of Baconian inductivism continued to be endorsed and developed well into the 19^th century,In addition to John Stuart Mill, another prominent 19th-century advocate of inductivism was English polymath John F. W. Herschel, whose book A Preliminary Discourse on the Study of Natural Philosophy (1830) helped to popularize the inductivist view of scientific methodology. most famously by English philosopher John Stuart Mill. In 1843, Mill published a seminal book detailing five inductive methods for identifying causal relationships in experimental data. These methods of inductive reasoning, known today as “Mill’s Methods,” purported to encapsulate the defining characteristics of scientific investigation. The full title of Mill’s book made its aim explicit: A System of Logic, Ratiocinative and Inductive, Being a Connected View of the Principles of Evidence, and the Methods of Scientific Investigation.

Eventually, however, philosophers of science began to recognize serious problems with inductivism. Here are several of the most devastating objections:

1. First and foremost, inductive logic isn’t a thing. Contrary to Bacon’s and Mill’s optimistic suggestions, there simply are no rules of logic for inductive reasoning as there are for deductive reasoning. The problem isn’t merely that philosophers haven’t yet developed a good system of inductive logic. It’s much worse than that. As it turns out, there can’t be a purely formal and objective logic of induction, as philosopher Nelson Goodman conclusively demonstrated in his 1955 book Fact, Fiction, and Forecast. (For a brief explanation, see this page of my e-book Skillful Reasoning.)
2. Moreover, neither deductive logic nor inductive reasoning (nor any combination of the two) can take empirical data as input and yield a theory as output. In other words, there are no strictly logical methods or procedures by which observational evidence can generate theories. Scientists have to use their own creative imagination to concoct new scientific hypotheses.
3. A third problem for Bacon’s view is closely related to the previous two. No matter how many observations and measurements we make, there are always infinitely many possible hypotheses compatible with the data we have collected. Most of these are hypotheses that no one has thought of. For example, in order to explain the astronomical data recorded by his predecessor Tycho Brahe, Johannes Kepler proposed the hypothesis that a planet’s orbit is shaped like an ellipse. (See Chapter 1.) However, infinitely many other shapes are also compatible with Brahe’s data. These alternative shapes are much more complicated than Kepler’s simple and mathematically elegant ellipse hypothesis, but the data did not rule them out. (This is illustrated with an example in the figure below.) For this reason, scientists cannot use straightforward inductive inferences to decide which hypothesis to accept. Scientists must employ non-empirical criteria like simplicity and mathematical elegance to decide between alternative theories.

The third objection to Bacon’s view highlights an important fact: data underdetermines theory. That is, empirical data cannot—by itself—determine which of many possible theories is correct. It can’t even narrow the possibilities down to a finite set of theories! This simple fact is called the underdetermination problem. (It is closely related to the notorious curve-fitting problem in mathematics: no matter how many points you draw on a graph, there are infinitely many possible curves that fit those points.) The underdetermination problem poses difficulties not only for inductivism but for other philosophical views about science, as we’ll see in what follows.

In addition, philosophers of science have come to recognize that science cannot be purely objective in the way Bacon, Mill, and other inductivists had hoped. There are at least two unavoidable ways in which subjective elements are involved in scientific inquiry. First, empirical data doesn’t collect itself. The scientist must use her own preconceived ideas about what kinds of observations are relevant to a given phenomenon before she can even begin to collect data about it. Scientists’ personal biases, presuppositions, and agendas can—and do—play a role in this aspect of scientific inquiry. Second, scientists must use their creative imagination when developing new hypotheses, as noted in the second objection above. Scientific theories don’t come directly from the data, but from us—from our own creative and imaginative attempts to make sense of the data. This creative process of hypothesis construction is inherently subjective, fraught with all sorts of biases and presuppositions.

So, the process of formulating scientific theories is, in fact, a thoroughly subjective one. The process of testing a scientific theory, on the other hand, might be more objective. Can there be a purely objective method for testing hypotheses once they have been formulated? That is an important question to which we’ll return shortly.