Competing Paradigms: On “The Last Writings of Thomas S. Kuhn”

January 16, 2023   •   By Paul Dicken

The Last Writings of Thomas S. Kuhn: Incommensurability in Science

Bojana Mladenović

In a sense that I am unable to explicate further, the proponents of competing paradigms practice their trades in different worlds.
— Thomas S. Kuhn, The Structure of Scientific Revolutions

It is not unimaginable […] that a tribe should be discovered of which both language and behavior remained inscrutable despite sustained effort. But shared genetic endowment and overlap between environments makes it unlikely, and it is not clear that to speak of such a tribe as human would not involve a contradiction in terms.
— Thomas S. Kuhn, The Plurality of Worlds

DELIVERING THE SHEARMAN Memorial Lectures at University College, London, in November 1987, Thomas Kuhn reflected on the moment that launched his philosophical career. He had been a graduate student at the time, preparing the obligatory freshman survey course in the history of science, and trying to understand how anyone ever accepted Aristotelian physics. Kuhn vividly recalls gazing out of his window for some time before suddenly grasping that Aristotle had meant something very different by “motion” from his Newtonian successors. Aristotle’s notion was broader: it encompassed how an acorn grows into an oak, or how the healthy decline into sickness, rather than just how physical objects move from one location to another. Many of the apparent absurdities of his theory — the impossibility of a projectile continuing to move once released from its source — were actually the result of reading contemporary concepts of motion back into his text. It was essentially a problem of translation: of not being able to think oneself into an entirely different world, and not simply a problem of trying to capture the nuance of the original Greek. Put simply, two different eras could not be expected to carve nature along the same joints.

Kuhn’s insight would overturn how we think about scientific progress. The traditional model had assumed a steady process of conjecture and refutation, beginning from a body of self-evident observations (the stone drops, this iron rusts, that raven is black) from which we infer universal generalizations; ongoing tweaks then further refine the theory for the purposes of ever better predictive accuracy. The process was famously illustrated by Karl Popper, who envisaged the entire scientific method as revolving around attempts to falsify our scientific theories by testing their deductive consequences against self-evident observations. But as Kuhn would stress in his influential The Structure of Scientific Revolutions (1962), this model fails to capture the actual historical record, and the often radical discontinuities that exist between successive scientific theories (such as between Aristotelian and Newtonian mechanics, or between the theory of humors and modern pharmacology). And that is because there are, in fact, no self-evident observations against which our theories can be tested.

The comparison with translation is instructive: we begin with individual words and familiar phrases, and tentatively infer the overall meaning of the text, but the rub is that the overall meaning of the text guides the translation of individual words and phrases. There is no neutral starting point, and so the hermeneutic circle continues until interpretative equilibrium is established. It is the same with science. We begin with a well-defined problem, and an accepted set of experimental techniques. These “exemplars” become the fulcrum around which a scientific discipline crystallizes. What Kuhn calls “normal science” then proceeds by applying similar techniques to similar problems, which over time generates an overall scientific worldview. Yet, just as some words and phrases will resist translation, so will some scientific problems resist satisfactory resolution. Initially, such anomalies are dismissed, or shelved for a later date, as happened with Newtonian mechanics’s failure to precisely predict the perihelion of Mercury. But eventually the interpretive strain becomes too great, and a new set of exemplars is adopted as a guide to future investigation — as when Einstein’s success in plotting the erratic orbit of the inner planet suddenly became the critical test for any putative theory of mechanics, and thus the conclusive proof of relativity.

Every work of translation is a compromise. But while philosophers like Popper seemed willing to accept any degree of violence to the text or model or theory if it could yield contemporary insights, the perfect fidelity to the original source material demanded in Kuhn’s early work offered no possibility of translation at all. If the standards by which one judges good scientific practice — and even what counts as scientific practice in the first place — are themselves determined by our overall scientific worldview, then those standards cannot be what determines the adoption of one scientific theory over another. In challenging the assumption that the past can always be judged (usually unfavorably) by the standards of the present, Kuhn had managed to problematize the idea that competing scientific theories can ever be compared at all. Indeed, Kuhn often spoke of a “Gestalt switch” or even a “conversion experience” to describe the transition from one scientific theory to another.

But while such conclusions were enthusiastically endorsed by a new generation of historians and sociologists during the revolutionary 1960s, who were eager to deconstruct the politics of scientific authority, Kuhn’s work was largely rejected by philosophers as implausible, or even self-refuting. And so, the years following The Structure of Scientific Revolutions saw Kuhn variously attempting to distance himself from his critics’ interpretations, or indeed misinterpretations, of his previous work. Ironically, he had become victim to his own problem of translation. In a follow-up volume of essays entitled The Essential Tension (1977), he argued that, while the standards by which we assess our scientific theories change, certain overarching values — accuracy, simplicity, fruitfulness — nevertheless remain constant. But these essays didn’t really help his case, with the same critics quick to note that different scientists would still make different assessments about the relative importance of these supposedly “overarching values.” Fundamentally, Kuhn found himself faced with the translator’s dilemma: balancing fidelity to the past’s version of reality with the need for contemporary relevance. As his last writings show, it was a dilemma that would occupy him for the rest of his career.


The Last Writings of Thomas S. Kuhn: Incommensurability in Science (2022), edited by Bojana Mladenović, brings together unpublished lectures from the 1980s and 1990s (including the aforementioned Shearman Memorial Lectures) alongside an unfinished book manuscript provisionally entitled The Plurality of Worlds. It provides a fascinating sketch of Kuhn’s mature thought up until his death in 1996. Most significantly, he came to believe that the difficulty inherent in comparing different scientific theories in the absence of shared standards — the problem of translation — actually only applies to a highly specific part of our scientific vocabulary. These are what Kuhn calls “taxonomic kinds,” the putatively referring terms (mostly nouns) that are responsible for dividing the world up into, say, “planets” and “stars” and other distinct elements. These taxonomic kinds have to be learned together as part of a group, in terms of both their instances and their contrasts. For Kuhn, to adopt different scientific theories is to divide the world up in different ways; the principal hurdle for moving from a geocentric to a heliocentric model of the solar system, after all, had nothing to do with astronomical observations or religious scruples, but rather with the conceptual challenges of reclassifying the earth as a moving planet (and explaining why we weren’t all thrown off into space). But crucially, he said, these changes would always be localized and take place against a larger background of shared meaning and otherwise overlapping scientific concepts — Ptolemy and Copernicus were trying to explain the same astronomical cycles — against which they can (at least partly) be evaluated.

Moreover, our capacity to form these taxonomies in the first place is itself something we all share, and Kuhn goes on to draw from work in developmental psychology to suggest that our most basic taxonomic kinds are even innate:

In societies with language, they are fundamental categories of thought: space, time, and individuated physical body; perhaps also a root concept of cause and of such basic social categories as self and other. They appear, collectively, to be innate […] their likely evolutionary source is the neural processes developed to track moving objects and to match different situations to a repertoire of behavioral responses.

More specific taxonomies are then built up through ongoing social interaction, with different communities coming to produce different taxonomies according to their different needs and priorities. The ancient Greeks grouped the sun and the moon together with the other then-known planets because their principal interest lay merely in tracking the “wandering” bodies against the fixed background of the constellations. Similarly, there are apparently 68 different languages that fail to differentiate green from blue, arguably because these communities historically lacked any readily available pigments for producing the latter. Another example: Many western palates still struggle to differentiate umami from the other categories of tastes. The important point, however, is that given the shared starting point for these different communities, it is always possible to retrace their steps and learn these different taxonomies.

Kuhn’s reliance here upon the (often provisional) results in development psychology marks a striking departure from his earlier work. This reliance seems at first to present something of a puzzle. The central message of The Structure of Scientific Revolutions was that the history of science is subject to radical and discontinuous change, which is why we struggle to translate between the terms of different scientific theories. Yet Kuhn’s later writings seem content to rely upon the results of our current scientific theories to provide a general account of how exactly those problems of translation arise in the first place. We are being asked to accept the historical contingency of our scientific knowledge — except those results which establish the historical contingency in question.

For the later Kuhn, the fundamental contingency of his approach is precisely its virtue. We all have to begin somewhere, and the results of our current scientific theories (however provisional) are the best we have available. This is what it means to take our own historical contingency seriously. But by the same token, just as it makes no sense to look for a perspective outside ourselves from which we can judge the reliability of our knowledge, it makes no sense to reject that knowledge simply because of its historical contingency — since that too requires us to stand outside ourselves, passing judgement from the very timeless perspective we can never attain. The extremes of realism and relativism thus both arise from the same error, and Kuhn’s last writings were his attempt to move beyond either.

The fashions of postmodernism notwithstanding, translation is possible because texts do not operate within a closed circle but instead endlessly defer their meaning to other texts. They are also about something, pointing out to the world, carving out their own taxonomic kinds, and they are used as a tool for communication. Similarly, scientific theories do not exist in an intellectual vacuum. They are constructed by human beings responding to sets of problems arising within certain environments. As Kuhn notes, we would not even be able to recognize something as another scientific theory unless it were at least partly translatable into our own concepts. Ultimately then, it is those things we have in common — our neurological equipment and our biology, as well as our embodiment within a shared environment — that ensure that no scientific revolution ever descends into complete incommensurability.


The unfinished sections of The Plurality of Worlds were projected to extend this treatment of translation and incommensurability beyond the history of science and to include almost every aspect of human culture. In Kuhn’s view, our various communities are ultimately constituted by the ways their languages divide up the world, and the conflicts and divisions between them are driven (or at least exacerbated) by problems of translation. But at the same time, he came to believe that, by broadening the scope of the problem of commensurability, we can see how to overcome those conflicts and divisions. Where Kuhn’s early work highlights the differences between scientific communities, his later work emphasizes the similarities within the differences. Perhaps most importantly, the very fact that we can recognize the differences as conflicts in the first place proves how much we have in common. The proponents of competing paradigms may practice their trades in different worlds, but, as Kuhn was at pains to stress in his last writings, sometimes those worlds are closer than we think.


Paul Dicken is a writer and philosopher based in rural England, and the author of Getting Science Wrong: Why the Philosophy of Science Matters (2018).