The limits of science

In New Atheism and the Scientistic Turn in the Atheism Movement, philosopher Massimo Pigliucci raises (but fails to dispose of, or even really argue) an interesting question: How far can we stretch the semantic boundaries of the word "science" without doing violence to the plain meaning of the word or rendering the word so broad as to be effectively meaningless? Pigliucci defines "scientism" as

a totalizing attitude that regards science as the ultimate standard and arbiter of all interesting questions; or alternatively that seeks to expand the very definition and scope of science to encompass all aspects of human knowledge and understanding. (144)

The first part of the definition, "totalizing attitude," seems vacuous and disconnected from reality (and Pigliucci spends zero efforts, but the second part, expanding the definition "all aspects of human knowledge and understanding," is a bit more interesting. Pigliucci's makes his strongest argument when he cites Sam Harris's book, The Moral Landscape: How Science Can Determine Human Values. According to Pigliucci, Harris wants to apply the word "science" to any rational, empirical inquiry into facts. Pigliucci cites Harris's own words, which are actually somewhat milder: Harris does not want to draw a "hard distinction between 'science' and other intellectual contexts in which we discuss "facts.'" Pigliucci argues that "facts" are too varied to even, as Harris does, place science on a continuum with other inquiries into facts (149). According to Pigliucci, the best definition of science is what professional scientists do: a collection of activities having common threads, including systematic observations and experiments, hypothesis testing, general theories about the world, peer review, and public and private funding (151). Pigliucci argues that if we "expand the definition of science to pretty much encompassing anything that deals with 'facts,' loosely conceived . . . the concept of science loses meaning" (151). Pigliucci argues throughout that using science to address philosophical inquiry — e.g. inquiry into religion and moral inquiry — is wholly inappropriate; philosophical inquiry is itself a rational inquiry into the facts, but it is so dissimilar to science that they are completely distinct categories. To conflate the two categories harms both philosophy and science. Notwithstanding Pigliucci's failure to cite anyone actually using his extreme broadening of "science" to mean anything having to do with facts (making this assertion a trivial straw man), his argument fails because he equivocates the word "fact," he has a too-narrow definition of science, and he commits the fallacy of the excluded middle. Repairing these errors allows a legitimate broadening of the word science, one which can address the issues that Pigliucci believes are beyond the boundaries of science, even broadly conceived.

(I do not address the thesis that Pigliacci actually argues in his paper, which seems to catalog a series of philosophical and scientific errors made by prominent New Atheists. I am instead trying to address a thesis that Pigliucci argues tangentially and perhaps only implicitly.

Although Pigliucci takes a proper philosophical attitude towards the word "science," in that he argues for a definite connotation regardless of usage, he takes a lexicographer's attitude toward the word "fact." Pigliucci argues that the word "fact" connotes "too heterogeneous a category" for science to encompass. Pigliucci asserts a broad definition of "facts," which includes all statements that one cannot successfully deny; Pigliucci asserts, for example, that one cannot, for example, deny that the sum of the interior angles of a triangle (on a plane) add up to 180° (150). But this argument can be read as simply the tendency of speakers of natural languages to apply the same word to different categories. Pigliucci's example is telling: Euclidean geometry is not a fact even in the loosest empirical sense of a fact as a true statement about the world. Instead, Euclidean geometry is a mathematical formalism; to determine whether or not Euclidean geometry accurately describes the real world, we need to actually observe and measure angles. And we find that often, Euclidean geometry does not accurate describe the world, as when we draw triangles on a sphere or the Reimann surfaces near a large mass. We can take the amorphous mass of meanings that constitute the lexicographical content of "fact" and easily divide them into distinct* categories: common observation, deductive certainty, settled scientific theories, social totems, and confident assertions. There is no need to hold that broadening the definition of "science" requires that the broader definition include every lexicographical denotation of "fact."

*more-or-less distinct, letting ordinary people use the settled cases and letting philosophers investigate the edge cases.

Unlike "fact," Pigliucci tries to construct a philosophical definition of science, but his definition which he introduces without argument) is simultaneously too broad and too narrow. It is too broad in the sense that by noting primarily the sociological and institutional character of science, Pigliucci fails to distinguish between institutional science and pseudoscience. Whether or not he was successful, Karl Popper tackles this distinction in The Logic of Scientific Discovery and Conjectures and Refutations. Popper tries to differentiate endeavors that look legitimately scientific, e.g. physics, astronomy, evolutionary biology, from endeavors that that look pseudoscientific, e.g. psychoanalysis, astrology, Marxist history, even though all of these endeavors share (or could easily share) not only all of the institutional characteristics that Pigliucci lists, but also reliance on observation and experiment. In Popper's view, one indispensable distinction is that real science is falsifiable, pseudoscience is not; this distinction is not institutional but philosophical, not a matter of somehow testing hypotheses, but a specific method of testing hypotheses.

Pigliucci's definition is too narrow in that we can easily conceive of science being done without many of the institutional characteristics he lists. How general must a theory be to be "scientific"? Is, for example, forensic science really a science? Forensic science seeks to discover what actually happened at a particular point in time, almost the exact opposite of the construction of a general theory about the world. If forensic science is not a science, what is it? Do we need systematic peer review — in something other than the trivial, over-broad sense that all communication is received and modified by listeners — for an endeavor to be scientific? Must we have public or private funding, again in other than the trivial sense that everything is in some sense economic? For decades, science was self-financed, pursued by people with their own income from other sources. Pigliucci's definition of "science" is as absurd as defining "dining" as something being done in a restaurant using food, which would include eating at McDonalds and exclude my friend, who is an excellent amateur cook, preparing dinner at home.

Finally, Pigliucci's entire objection to scientism is an exercise in the fallacy of the excluded middle. In Pigliucci's view, the only alternative to his rigid, narrow (and absurd) definition of "science" is an anything-goes descent into linguistic anarchy. It is astonishing, and should be incomprehensible, that a professional philosopher would commit such an elementary fallacy. When it suits him, Pigliucci perfectly comfortable with words having broad, heterogeneous meanings; he does not, for example, see his broad definition of "fact" to render the word meaningless. Even the over-broad, uncited definition of "science" he objects to — anything having to do with "facts" — is still meaningful, so long as "facts" is meaningful. We can have words with broad meanings that are still meaningful; Pigliucci, however, sees no middle ground — at least regarding "science" if not "facts" — where there manifestly can be, and probably is.

Pigliucci's article does point us towards a more useful definition of "science," rejecting the idea that there is an "objectively" correct definition of science about which everyone in the world might be mistaken; there are only more or less useful definitions. To do so, we reject the institutional components of Pigliucci's definition as accidental, required for performing only certain kinds of scientific investigations in a particular social and cultural context. Instead, we retain the philosophical components. First, we can talk scientifically only about the real world*. Second, we construct theories**, collections of logically connected statements, about the real world. Third, these theories must be falsifiable: it must be logically possible that there are determinable facts that would disprove a theory. Fourth, we accept common observation as the factual basis for attempting to falsify a theory; for a theory to be falsifiable by observation, the theory must logically entail that some observation is impossible. Finally, we invoke parsimony: we discard any part of a theory that does not change the entailed observations as unnecessary. Like any other definition in natural language, this definition still leaves edge cases (is history a science?) but it seems to carve out a core of unambiguously scientific discourse and unambiguously unscientific and pseudo-scientific discourse. The question is: is it useful?

*thus eliminating talk about Middle Earth or Erehwon as scientific. We can, of course, talk scientifically about the texts and authors of The Lord of the Rings and Erehwon, which are real.

**both general and specific theories, which allows us to talk scientifically both about the law of gravity, a general theory, as well as specific theories such as whether O.J. Simpson murdered Nicole Simpson and Ron Goldman, and why Supernova 1987A has a visible ring.

This definition seems to exclude a lot of religious thought as either unscientific or scientifically false. In The God Delusion, Richard Dawkins proposes the "God Hypothesis." Dawkins asks: what happens when we try to construct religious thought as science, broadly conceived?

We can apply these criteria to God. First, we hypothesize that God is real, with real properties. Second, we create a theory about what a real God and His properties means. A God doesn’t just sit there; what does He do? Third, we make this theory testable: we must be able to determine whether it is true or false. Fourth, we must test the theory by observation or experiment. Finally, we ensure the theory is parsimonious: that is, if we took out God, the theory wouldn’t explain as much. Once we have followed all these steps, we have a scientific theory that includes God, which we can test against what we actually observe.

But constructing this kind of theory of God puts believers on the horns of a dilemma. Centuries of scientific investigation show that the best scientific theories, testable by observation, include nothing like a personal God. We find only a universe of blind, mechanical laws, including natural selection, with no foresight or ultimate purpose.

Alternatively, a believer could reject one or more of the criteria for a God theory, but doing that has profound implications. If she admits that God is not real, she’s already an atheist. If she says God doesn’t do anything, who cares? If her theory cannot be tested at all, then there's no way of telling if it's true or false. If her theory can be tested only by private revelation, not by observations available to everyone, she unjustifiably claims private knowledge. And if her theory is observationally identical to a theory that does not include God, then she’s again an atheist, for a God who makes no difference is no God at all.

The only remaining question is whether some people would find this analysis useful, and I know many people who, applying this analysis, have abandoned their religion.

We also have any number of authors, notably Josh McDowell, author of Evidence that Demands a Verdict, who at least attempt to make a scientific case, in the broad sense noted here, for the existence of God.

Does this definition include or exclude anything obviously objectionable? We seem to admit lawyering, but lawyers are not obviously unscientific. This definition excludes pure mathematics (even if a lot of mathematicians are Platonists), but I suspect most mathematicians would not object to being placed outside the boundaries of science. This definition definitely excludes philosophy; I do not know, however, whether Pigliucci would be encouraged or enraged by such exclusion.

Finally, the question remains: does this definition of science "encompass all aspects of human knowledge and understanding"? It certainly does not encompass all aspects of human understanding (even if the definition of "understanding" is so broad as to render the term meaningless). As noted above, it does not include mathematics, literature, or even philosophy, which are uncontroversially parts of human understanding. Perhaps, however, it does encompass all knowledge; it is perhaps the case that anything that legitimately deserves the name "knowledge" really must be scientific, in the sense described above. But I need not answer this question to dispose of Pigliucci's case; it is enough to find that this broad definition of science is useful and largely unproblematic.

[edited 8/22/14: with the assistance of Jerry Coyne, I have rewritten the passage beginning, "We can apply these criteria to God," to make it clearer and less technical. I also fixed the formatting of the footnotes.]