Last time I looked to find a middle ground in philosophy of science, between Francis Bacon’s historically untenable inductivism and Paul Feyerabend’s irrationalism. I noted then that I think Karl Popper, Thomas Kuhn and Imre Lakatos all attempt to stake out a position somewhere in this ground, with varying degrees of sucess. I turn to them now.

Karl Popper rightly acknowledges the scientific importance of fallibilism and uncertainty: science is powerful not because its conclusions can be proved right, but because it can acknowledge when they are proved wrong. Popper notes that science in practice advances more by falsification than by induction: the role of empirical data is not to ground generalizations, but rather to disconfirm them. One can legitimately formulate a theory in abstraction that says all swans must be white; the important thing is that one reject it when one observes a black swan.

But Popper’s critique of inductivism does not go far enough. The history of science also indicates that theories are not disproved in isolation. W.V.O. Quine, in his famous “Two dogmas of empiricism”, astutely noted that “Any statement can be held true come what may, if we make drastic enough adjustments elsewhere in the system.” We see this happening in the history of Tibetan medicine – where dissection revealed no tantric anatomy, but various thinkers adjusted that anatomical theory to suit the empirical evidence – but just as much in the history of modern science in the West. When one finds an organism has biological features that are maladaptive to its environment, one does not throw out the theory of evolution, but adjusts it to explain those features more adequately. Quine’s claim in this respect was similar to an earlier and more limited claim made by Pierre Duhem, so it has come to be referred to as the “Duhem-Quine thesis”, and it it is a damning criticism of the idea that science works merely by the falsification (let alone confirmation) of individual theories.

It is in terms of something like the Duhem-Quine thesis that Thomas Kuhn made his famous distinction between “normal science” and “revolutionary science”, with the former dependent on a set of unquestioned assumptions or first principles (“human beings evolved from primates through natural selection”, “matter is made of atoms”) – such that any empirical anomalies one encounters are always to be explained in ways that do not falsify those assumptions. Revolutionary science is much rarer and happens when those assumptions change. But how do they change? The difficulty for Kuhn is to identify how his approach doesn’t merely collapse back into Feyerabend’s – a consequence Kuhn is determined to avoid. In the epilogue to The Structure of Scientific Revolutions he affirms that “I am a firm believer in scientific progress.” What is not so clear on Kuhn’s account is how that progress can happen, given that he claims paradigms are incommensurable – there is no common standard by which they can be evaluated against each other. (There are some thin or formal standards – prefer the simple to the complex, and so on – that people working in different paradigms can accept, but they are not rich enough to resolve any actual disagreement.)

The most helpful approach that I have found is that of the Hungarian philosopher Imre Lakatos (the Hungarian pronunciation is EEM-ray LAHK-uh-tohsh). Lakatos explicitly tries to stake out a position between Popper and Kuhn, preserving an approach which is non-inductive but rational. He distinguishes between a weak version of the Duhem-Quine thesis, which (against Popper) “only denies the possibility of a disproof of any separate component of a theoretical system,” but rejects a strong version that “excludes any rational selection rule among the alternatives.” And so he refuses the idea that paradigms or traditions (he calls them “research programmes”) are incommensurable. Each tradition has anomalies, observations or discoveries that do not quite make sense with the tradition’s first principles, that still need to be explained. (Aristotle would have called them aporias, a word usually translated as “puzzle”.) It is a requirement of good science – a thin standard perhaps – that one acknowledge the anomalies that exist in one’s own research program, as well as working to resolve them. And success or failure in resolving them can mark a research programme as either “progressive” or “degenerating” – a progressive program changing from its own inner logic, while a degenerating program changes in response to external criticism.

When one takes the approach of Kuhn or Lakatos to natural science, though, an interesting thing happens. For those of us studying disciplines like ethics that are not primarily empirical, the messiness of science turns out to be helpful. Bacon’s inductivism and Popper’s falsificationism put empirical evidence at the heart of their theories; there would be no way to confirm or disconfirm an ethical theory on their view. But Kuhnian paradigms or Lakatosian research programs can exist in any field of inquiry. Anomalies do not have to be empirical; they are conflicts between two elements of a system. They can be present within mathematics, and so they can also be present in ethics. And as Kuhn himself rightly notes (in his “Notes on Lakatos”), “Scientific behavior, taken as a whole, is the best example we have of rationality.” Philosophy should not simply imitate science, but I think it does well to draw significant lessons from it.

And the strength of Alasdair MacIntyre’s methodology, in my view, is the way it applies the approach of Kuhn and Lakatos to comparative ethics. Different traditions of ethical inquiry are effectively different paradigms or research programs (Kuhn himself uses the term “tradition” synonymously with “paradigm”). MacIntyre himself takes a position between Kuhn and Lakatos, preserving the concept of incommensurability in a way that he thinks answers Lakatos’s objections. I am not sure that his approach succeeds, and have become more sympathetic to Lakatos in this regard against Kuhn and MacIntyre. But the general approach of modelling one’s view of ethical inquiry on a view of scientific inquiry that takes account of the Duhem-Quine thesis – this approach has struck me so far as very fruitful, a progressive research program.