The Idiographic Lock-In of Science
Friday 03 May 2024
Wilhelm Windelband is known for his distinction between the idiographic and the nomothetic. This was given its first exposition (so far as my knowledge extends) in Windelband’s 1894 rectorial address, and here is how he formulated it at that time:
“The nomological sciences are concerned with what is invariably the case. The sciences of process are concerned with what was once the case. If I may be permitted to introduce some new technical terms, scientific thought is nomothetic in the former case and idiographic in the latter case. Should we retain the customary expressions, then it can be said that the dichotomy at stake here concerns the distinction between the natural and the historical disciplines.”
Less well known is Windelband’s Theories in Logic, which was originally written as “Die Prinzipien der Logik” and appeared in the Enzyklopadie der Philosophischen Wissenschaften, 1912, but the English translation was published as a short book. (I previously mentioned this book in a PS to newsletter 240.) In this short book Windelband offers a further elaboration of his distinction, and this elaboration was of particular interest to me. Windelband’s distinction, if strictly observed, cleaves science in two, with the natural sciences on one side of the division, and the historical sciences (or, if you prefer, the humanities) on the other side. But in Theories in Logic Windelband suggests that sciences might be more or less nomothetic, more or less idiographic, and the nomothetic or idiographic proportion of a science may change over time:
“…we have to distinguish between those sciences which are governed by laws and those which deal with events, between nomothetic and ideographic inquiry. It is this which really makes the difference in intellectual interest between Natural Science and the Humanities. But we cannot repeat too often that we are here only speaking of ultimate aims and hence of those sciences which appear as polar opposites, between which the real work of Science moves in manifold gradations, so that in any particular case we can only speak of a preponderance of one or the other moment — as Rickert, in his penetrating analysis of this relation, has pointed out. The ultimate goal of all investigation of Nature is to attain timeless generic concepts of being and happening, but that does not exclude the fact that the way thereto leads over stages of simpler interconnexions in which it rests and provisionally halts. For it is precisely in the real that the nomothetic rationalisation of Reality must find its limits. On the other hand, the specific object of all historical inquiry is a construction which is significant chiefly because it can never recur, and which has to be lifted out of its entanglement in the non-significant elements lying all around it. To understand such a construction, however, History requires general concepts and axioms, which she is certainly more likely to borrow successfully from general experience than from the natural sciences… and it creates for itself the possibility of characterising this unique object by a peculiar kind of generic concept and by a comparative study of the conformity of events to law.” (p. 57)
Windelband is saying many things in this passage, but the lesson I have plucked out of this is that Windelband’s distinction implies that natural science and the humanities are polar opposites in terms of their methodology, but in fact the development of science involves “manifold gradations” in which the nomothetic and the idiographic preponderate by turns. (NB: the English translation of the rectorial address uses “idiographic” while Theories in Logic uses “ideographic”; I have not consulted the German text of either of these to see how Windelband himself rendered his coinage.) This means that natural science (and the humanities as well) might swing like a pendulum between a nomothetic pole and an idiographic pole, being now more nomothetic, and now more idiographic.
I’ve been thinking about this since I read Theories of Logic last summer, and this has influenced my way of thinking about the nomothetic/idiographic distinction, and this influence has probably already expressed itself in these newsletters over the past year, since I frequently talk about the taxonomy of the sciences. Just as there is no science of civilization, there is no science of science, and that means that our understanding of science (and of civilization) is pre-paradigmatic (to use a Kuhnian term, the significance of which we will see in a moment). It seems paradoxical that the individual sciences can be as well defined as any part of human knowledge, while the whole enterprise of science remains essentially unknown to us — if we could christen this paradigm with a memorable name maybe the realization of not knowing what we are doing when we do science would get some traction and some theoretical attention.
Whether or not this is a paradox, it should not surprise us, as this is baked into the very substance of science. Science progresses when it manages to hit upon a productive set of abstractions that we can use to leverage a very narrow way of looking at the world. The “manages to hit upon” is key here, because, given the lack of a science of science, there is no method of converging upon optimally productive abstractions; we can only cast about for them. But finding productive abstractions is also idiosyncratic: some individuals have a much better feel for converging on productive abstractions than others (indeed, better than most of the rest of humanity), and this idiosyncratic efficacy is again a function of science remaining an art rather than being a science. And we should keep this in mind when we consider the ancient question of whether history is an art or a science, because science itself is an art, and will remain an art unless or until we have a rigorous science of science. In the absence of a rigorous science of science, it would be nice at least to have a proof that a rigorous science of science is possible, or that it is impossible, but we don’t even have this much.
The narrower our science, the narrower the scientific abstractions we employ, the more likely we are to make progress. The wider and more comprehensive our science, the more comprehensive the scientific abstractions we employ, the less likely we are to make quantifiable progress. There are a couple of things going on here. The most comprehensive concepts are employed in philosophy, and at some point in the narrowing of the meaning of concepts we cross the threshold from doing philosophy to doing science. Passing in the opposite direction, when we start from the narrow concepts employed in the special sciences and pass on to more comprehensive concepts, we pass a threshold from doing science to doing philosophy. Also, the more comprehensive the concept, the more foundational it is, and simple, foundational concepts are related to other simple, foundational concepts, and this makes them difficult grasp, difficult to elaborate, and difficult to systematize.
Also in Theories of Logic (pp. 79–81), Windelband offers an account of scientific abstraction that in essentials agrees with what I have come to on my own, and finding this in Windelband has encouraged to me to continue to develop this way of thinking about scientific abstraction. This influence has also been expressed in these newsletters, as, for example, in the discussion of the utility of abstractions in newsletter 284.
Given that science can become more nomothetic or more idiographic as it develops, and that the whole of the scientific enterprise may pass between nomothetic extremes and idiographic extremes, it occurred to me today that a natural science that tends to the idiographic will have a very different relationship to anomalies than a natural science that tends to the nomothetic. In a preponderantly nomothetic natural science, anomalies will be regarded dismissively, as little more than noise in the data, whereas in a preponderantly idiographic natural science, anomalies will be viewed as significant and worth particular attention. Thus in natural science tending to the idiographic, a prominent anomaly will not necessarily result in a model crisis and a paradigm shift, while in natural science tending to the nomothetic, once an anomaly becomes sufficiently obvious, it poses a serious problem for that science and may spur a model crisis.
It can be the case that we already know the anomalies that will dog a new theory even as that theory appears, but when the pendulum swings to the nomothetic extreme, it may be easy to overlook (apparently trivial) anomalies in the general enthusiasm for the explanatory power of the new theory. But when the pendulum swings to the idiographic extreme, the anomaly stands out as being no less important than the fully explained phenomena.
At the moment when minds are most ripe for a new theory, it is almost impossible to introduce a new theory because all of the alternatives to the established theory also involve known anomalies. The scientific community fractures; research programs proliferate, and efforts are divided. Out of this division there are different pathways, including:
- A truly new theory with minimal relationship to the divided scientific milieu appears that scatters all else before it;
- Science stagnates for lack of a unified vision;
- A long, slow process of selection eventually allows a much revised theory from the earlier divided scientific milieu to gradually consolidate its influence and thus to become the new scientific paradigm.
This is not intended as an exhaustive list, but merely what I have taken off the top of my head. As we all know, there are interpretations of the Kuhn cycle that make science and theory change an essentially irrational process (and because of the lack of a science of science we have no definitive argument against this position), and other interpretations allowing that, while there is ongoing theory change, there are good reasons for theory change that are not arbitrary or merely human, all-too-human. The irrational interpretation gives us science that oscillates among paradigms but does not demonstrate actual progress in knowledge. This is essentially the Spenglerian view of science. The rational interpretation gives us science with directionality, which is incrementally converging upon actual knowledge, despite the two-steps-forward-and-one-step-back rhythm of scientific progress. This latter could be called the Hegelian view of science.
The history of modern science to date (and I have argued in several newsletters that modern science is distinctive, so I believe it justified to begin with modern science) starts in the late fifteenth century, a pre-history of modern science that goes to the middle of the sixteenth century when Copernicus’ De revolutionibus appears. After Copernicus we see the rapid expansion of mathematical methods by Galileo and Kepler and their followers. In this earliest period of modern science, ideas about the nature of science are very mixed, and we can’t really identify a preponderance, but we can identify a growing core of mathematization. With the Enlightenment we have a strong swing to a nomothetic paradigm in natural science. This endured up through the beginning of the twentieth century, and began to change sometime mid-twentieth century. Now science is swinging toward an idiographic paradigm, as scientists show a great interest in exceptions and anomalies, and less interest in universally valid laws.
In a cyclical history of science, we would expect this idiographic paradigm to continue to develop, until, in a few centuries, the pendulum begins to swing back to the nomothetic. But if science ultimately embodies a kind of Hegelian rationality, and progresses even as it passes through dialectical extremes, it might pass into another period of nomothetic preponderance, but the next time around science will have learned its lessons, and its nomothetic extreme would incorporate idiographic insights, incorporating them while moving beyond them. Here is a pessimistic interpretation that occurred to me: after passing through the Enlightenment nomothetic paradigm, natural science, in swinging toward an idiographic paradigm, gets lost in the weeds of detail, and never really fully extricates itself from this idiographic preponderance of method, due to the mechanism of idiographic science celebrating anomalies. We could call this idiographic lock-in. In this interpretation, science in its mature form cannot resist the Siren song of paralysis through analysis, insisting on reducing all general laws to particular observations, and, in so doing, science dissipates itself in a profusion of detail and cannot find its way back to productive abstractions.
