Book review: Theory and Reality: An Introduction to Philosophy of Science by Peter Godfrey-Smith (UPDATED)

In a recent post on philosophy of science for ecologists, Brian identified Harvard philosopher Peter Godfrey-Smith’s Theory and reality: an introduction to the philosophy of science as promising-looking. I thought it looked promising too, so I read it (Kindle edition). Here’s my review. (UPDATE: another review here)

The book is based on introductory lectures in philosophy of science that Godfrey-Smith used to give at Stanford. It assumes no background knowledge of philosophy, and so is perfectly accessible to anyone reading this blog. But it’s aimed at people interested in philosophy, and takes that interest for granted. It doesn’t spend much time trying to argue you into an interest in philosophy you don’t already have. And it’s not aimed at teaching you just the bits of philosophy of science that you need to know in order to be a good scientist, or a better scientist than you are already. For instance, at various points it links philosophy of science to topics in epistemology and metaphysics that scientists have no particular reason to care about. So it’s not the philosophical equivalent of, say, an introductory biostats course or “math for ecologists” or whatever. Whether you find philosophy of science useful in your day-to-day scientific work is up to you.* But if you want to have a better sense of where philosophers of science are coming from, and be able to identify and understand those bits of philosophy of science that are relevant to you as a scientist, then I think you’ll find this book very helpful.

The first 2/3 of the book is a chronological survey of the most important work in philosophy of science from the early 20th century up until almost the present day, with a few nods to important earlier figures. It switches from chronological to topical organization to cover recent work. I think the chronological organization is effective. Features of later work that might otherwise seem puzzling make more sense when you know about the earlier work that later work was either trying to build on or improve upon. The book also includes a couple of chapters on fields on the boundary of philosophy of science (sociology of science, and feminist philosophy of science and “science studies”).

It’s a short and easy read—I knocked it off in a couple of days. It’s the philosophy of science equivalent of one of those bus tours that takes you to, say, Westminster Abbey, the Palace of Westminster, Tower Bridge, the Tower of London, St. Paul’s Cathedral, Stonehenge, and Bath all in one day. Yes, those bus tours only hit the obvious highlights, and yes they only give you a quick superficial glance at the sights you’re seeing, and yes they leave you wanting to go back for more. But they fill a real need. So for “Westminster Abbey, the Palace of Westminster…” read “the logical postivists, Hempel, Quine, Popper, Kuhn, Lakatos, Feyerabend, Laudan, Goodman,” plus various other figures discussed more briefly.

It’s an opinionated survey. Godfrey-Smith always tells you what he thinks of the ideas he discusses, and he uses the final chapters in part to lay out and stump for his own views. I welcomed this. I wouldn’t want a he-said, she-said survey that just describes what philosophers have written without any attempt at evaluation. Reading someone else’s evaluation helps me form my own evaluations, rather than getting in the way of me forming my own evaluations. Especially since Godfrey-Smith always gives a fair (sometimes generous) description before he launches into (often critical) evaluation. Nor does he skip any major ideas he disagrees with. At least, I didn’t notice any obvious gaps in the coverage, and I know enough philosophy that I’m fairly sure I would have. And Godfrey-Smith tells you when his own views are unorthodox, as opposed to when he’s voicing widely-shared opinions, so he never comes off as trying to railroad you towards his own views.

I broadly agree with Godfrey-Smith’s views, and I think most scientists will too. He’s a “naturalist”, which in this context means a philosopher of science who takes as his starting point how actual (rather than hypothetical or idealized) scientists go about their business. (Not that he thinks science as its currently practiced is above criticism, including philosophical criticism.) The only point where I seriously disagreed with him was his explication of subjective Bayesianism, which is surprisingly light on criticism. Godfrey-Smith strongly criticizes many other views, his overall judgement seems quite good, and he’s familiar with current everyday scientific practice. So I don’t understand how he could fail to strongly criticize a view that has been strongly criticized by various recent philosophers of science (e.g., Deborah Mayo), and that has never gotten any traction among practicing scientists or statisticians.** Especially because the subjective Bayesian view grew in part out of Carnap’s work in logical positivism, and Godfrey-Smith follows the rest of philosophy of science in writing off Carnap’s work as a dead end.

For those of you who worry that philosophy of science is remote from the actual practice of science, well, if your read this book you’ll discover that many philosophers of science worry about that too. As the book discusses at length, perhaps the biggest issue in philosophy of science ever since Thomas Kuhn’s Structure of Scientific Revolutions in 1962 has been figuring out the philosophical implications, if any, of how science was and is actually done. Those implications aren’t obvious. There’s an old philosophical dictum, dating back at least to Hume, that “is does not imply ought”. That is, descriptive and normative issues are two different things. How scientists do science, or how they’ve done it in the past, doesn’t on its own imply anything about how science should be done (especially since scientists themselves often disagree with one another on how to do science). But on the other hand, the descriptive and normative aren’t totally independent of one another either. As another old philosophical dictum goes, “ought implies can”. That is, any claim about how things should be presupposes that they could be that way. I have to say that I sometimes found Godfrey-Smith a bit unclear on the relationship between descriptive and normative claims in philosophy of science. Or maybe just a bit unclear on what claims are being made in the first place. In particular, some (not all) recent work at the interface of philosophy of science and other disciplines—sociology of science, “science studies”, feminist philosophy—arguably suffers from muddying descriptive and normative claims, and from lack of clarity about exactly what’s being claimed in the first place. Godfrey-Smith notes this, but his summaries of this work sometimes suffer a little from the same flaw, I think. In his admirable urge to take seriously recent work at the boundary of philosophy of science and other disciplines, I think he’s a bit less critical and demanding of that work than he should be. For instance, there’s a serious discussion of Bruno Latour, whom Andrew Gelman for one frequently mocks, and not without reason. I don’t think Godfrey-Smith should’ve mocked or omitted Latour and related figures. But I do think the chapters on sociology of science and feminism/science studies are a bit too uncritical and drift a bit too far away from philosophy of science sensu stricto. And Godfrey-Smith’s explication of his own naturalism wasn’t totally satisfying to me. I left the book with more of a sense of what his naturalism isn’t than what it is. Bill Wimsatt is one naturalist philosopher of science who’s good at linking how science is actually done to a normative account of how it should be done. For instance, Wimsatt emphasizes how human beings are cognitively limited in all sorts of ways. Many scientific practices involve heuristics and “rules of thumb” that would be suboptimal or even undesirable for a cognitively-unlimited being, but that are optimal given humans’ cognitive limitations. The scientific preference for “simple” or “parsimonious” models is a good example. As Godfrey-Smith notes, philosophers of science have tried mightily—and failed abysmally—to find a universal justification for preferring simple models. After all, the truth might be complicated. And there doesn’t seem to be any other desirable property (testability or whatever) that invariably increases with the simplicity of one’s model. But as Wimsatt (but not Godfrey-Smith) notes, real scientists’ preference for simple models doesn’t have the sort of universal justification philosophers traditionally have sought. Rather, a preference for simple models is justified in many contexts (not all!) for heuristic reasons, such as that real human beings just can’t wrap their heads around complex models, and that simple models often (not always!) provide a good-enough approximation to more complex models.

At the end of every chapter are suggestions for further reading, along with brief comments (e.g., identifying which of the readings are accessible and which are advanced and technical). This is very helpful. There’s also a glossary, which I didn’t really need since I’ve read some philosophy of science already, but which I imagine would be a godsend for someone totally new to the subject. And while we’re on the subject of vocabulary,Godfrey-Smith is good about alerting the reader to (and avoiding) loaded terms that get used in different ways by different philosophers.

The style is clear and readable. There are occasional jokes, often rueful apologies for using well-worn examples. And the book shows its origins as introductory undergraduate lectures in a good way. Godfrey-Smith is good at picking clear examples to illustrate broader points. And he’s good at picking examples that undermine your pre-philosophical intuitions, and so motivate you to stop and think about something that might’ve otherwise seemed obvious.

I took away from the book a better understanding of some philosophical topics that I previously hadn’t understood. For instance, going into the book I’d found Nelson Goodman’s notion of “grue” to be weirdly pointless. I stand corrected on that, at least in part. I now see the point of “grue”, and it was interesting to find that it’s actually scientifically relevant (although I still think Goodman did himself no favors by making his point with such a weird hypothetical***). And I really like Godfrey-Smith’s own resolution of the “grue problem”. I now have a better understanding of Kuhn, including the various tensions and ambiguities in Kuhn’s thought. And I have a better sense of the current lay of the land in philosophy of science.

I’d recommend the book to anyone who wants a quick accessible overview of philosophy of science, including scientists who want an overview so that they can then hone in on the bits of philosophy of science most relevant to their own work.

*At least in part. It might also depend on the state of your field—you might have no choice but to learn and do some philosophy of science. Godfrey-Smith discusses the possibility that philosophical issues loom large for practicing scientists only under certain circumstances, such as during Kuhnian “paradigm shifts”. I think this is right. Part of why I’m interested in philosophy of science is that I think philosophical issues loom larger in a young field like ecology than they do in, say, chemistry.

**“Bayesian” scientists and statisticians come in various stripes, but hardly any are subjective Bayesians in the sense Godfrey-Smith explicates, and most would find that sort of subjective Bayesianism totally bizarre. Andrew Gelman, for instance, is a self-described Bayesian, but is emphatically against the sort of subjective Bayesianism philosophers of science apparently have paid the most attention to.

***And I say that as someone who very much sees a place for ridiculous hypotheticals.

5 thoughts on “Book review: Theory and Reality: An Introduction to Philosophy of Science by Peter Godfrey-Smith (UPDATED)

  1. Admittedly, the philosophy of science is not something I read all that much about, so I appreciate this post, Jeremy. Godfrey-Smith’s book sounds like a good resource for investigators like myself who are more rooted in applying empirical approaches to theoretical development. So I will be downloading it in the coming days.

    Even though I’m a bit of a neophyte on these things, I had some immediate thoughts when I read your comment: “As Godfrey-Smith notes, philosophers of science have tried mightily—and failed abysmally—to find a universal justification for preferring simple models.” I think the tie-in you mention to limited cognition provides a potential justification. Simply enough, we have to start somewhere.

    I’ve done a ton of mechanistic work related to investigations of molecular pathways. For any one of the known pathways, we can look at the history of the literature, and witness a typical S-shaped curve in terms of its discovery. Usually, we begin with a simple two or three step process. Such as, until very recently, we believed one of just a few mutations caused any given cancer. Now we are in the log phase of discovery concerning molecular pathways of cancer, and have realized for most cancers, hundreds of possible combinations of mutations give rise to any specific cancer.

    So the model becomes more complex, and more representative of reality, after and only after the simplistic model has been presented, digested & investigated by many more people. However, the approaches remain simplistic because each new investigation is adding two or three building blocks to what will become an ornate structure.

  2. Pingback: Whewell’s Gazette: Year 2, Vol: #03 | Whewell's Ghost

  3. A bit late here, but I’ll put in a good word for Latour, contra Gelman (whom I almost always agree with otherwise). Adam Calhoun of Neuroecology has a great “interview” with Latour and Steve Woolgar’s book Laboratory Life here:

    The quotations presented from the book may be a bit hyperbolic – or, conversely, much of what he describes may be obvious to practicing scientists – but it at least has the virtue of being born out of hundreds of hours spent observing a real scientific lab, down to its minutia. Later (post-1990) Latour books are, for the most part, less useful to scientists and less engaged with scientific practice, though still very interesting. Anyway, I’m not going to claim that every scientist (or any scientist) should go out and read Latour, only that he shouldn’t be lumped together with the vast party of other French intellectuals who make all kinds of grand, obscure pronouncements about science, many of whom he criticizes in his wonderful essay “Why Has Critique Run Out of Steam?”

  4. Pingback: Book review: The Knowledge Machine by Michael Strevens | Dynamic Ecology

Leave a Comment

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.