An Interview with Jean-Marc Lévy-Leblond

1 PHILIPPE LACOUR: You’re an emeritus professor of physics and a former program director at the Collège international de philosophie. You’ve published many works, founded the journal Alliage (covering culture, science, and technique) and edited the “Science ouverte” collection at the Editions du Seuil publishing house. You define yourself not as a “popularizer” but as a “critic of science.”[1] Could you say more about what you mean by that?

2 JEAN-MARC LÉVY-LEBLOND: Ever since the 19th century, modernity assumes that any cultural work is accompanied – must be accompanied! – by a work of criticism. Of course, the word “criticism” is to be taken in its most positive sense: not as a rejection or a questioning of the work, but as a consideration and an appraisal of its meaning and originality, whether it’s a work of literature, music or art. The contemporary cultural world, by its very nature – the search for the new that characterizes it – does not immediately lend itself to universal acceptance. It requires mediation allowing it to be grasped, or at least taken into consideration, by a large audience beyond a minority of enlightened amateurs. Scientific research, although it has internal collegial procedures at its disposal that are suited for evaluating the disciplinary validity of its results, is singularly lacking in means when it comes to gauging their epistemic interest, their philosophical impact and their social effects. As a result, laypeople – in whose name and supposed interest this work is carried out – are incapable of making a value judgment on it and giving their opinion of its relevance to society as a whole.

3 Thus, for the most part, scientific activity lacks a satisfactory level of cooperation with the cultural domain, which is a necessary prerequisite for its democratic control. Moreover, this deficiency (a “defisciency”) [2] inhibits the critical reflexivity of scientists themselves, often maintaining conceptual archaisms and obsolete forms of language that end up becoming epistemological obstacles, within the very disciplines that are the most active. As a result, I argue for a “critique of science” (critique de science) – to be carefully distinguished from a “criticism of science” (critique de la science), although they are bound up with each other – going beyond the mere mediation of knowledge. [3] Furthermore, the critique of science must first be an autocritique, insofar as the concepts created by an innovative theoretical development only come into existence bearing the very “timestamp” of the field of knowledge (a theoretical structure, a specific terminology) that they will revitalize. [4] A permanent critical process of rearrangement, of reorganization (refonte) in Gaston Bachelard’s sense of the term, is therefore necessary. [5] Fortunately, there have been pioneers in this area, such as Stephen Jay Gould, author of an extensive body of critical work in the life sciences. And many works in the past few decades concerning the history of the sciences have contributed decisively to this perspective. [6]

4 P. LACOUR: To start our discussion on the epistemological relevance of the notion of the clinic, I’d like to allude to an author who made it one of the themes of his thinking. Gilles-Gaston Granger said that the unity of science is not based on method (and even less on its subject), nor on “language,” but on intention (knowing reality objectively).[7] Do you agree with that definition, which emphasizes the principle (the scientific mind, a legacy of Bachelard) more than the rules (the actual procedures), which are too varied to lay claim to any sort of similarity?

5 J.-M. LÉVY-LEBLOND: It seems to me that if we expand the field of scientificity in that way, we completely lose sight of the specific question of its nature. Aren’t many human activities also based on objective knowledge of reality, in particular technical practices?

6 P. LACOUR: Of course, but science pursues a goal of knowledge, while the end purpose of technique is to take action. What’s more, the fact that technique is based on science is fairly recent: doesn’t that confirm the definition of science through the “intention” of objectivity?

7 J.-M. LÉVY-LEBLOND: I have difficulty accepting a definition of science that remains universal in space and time. Of course it’s true that Greek science, for example, had a “goal of knowledge.” And even then we’d probably have to add that although the atomists, to cite only them, probably tried to understand the world, it was with a metaphysical and even ethical objective that went well beyond the mere search for knowledge for its own sake. But more than anything, science in the modern sense – since the 16th century – has very little in common with the sciences of Antiquity, either concerning its (experimental!) methods, or its social organization. And, as Descartes explains, it strives to make us “as it were the masters and possessors of nature,” therefore playing its role of taking action. The “intention of objectivity” then appears more as a means allowing for effective action upon the world than as an end in itself. To be sure, this ambition remained unfulfilled until the end of the 18th century, but in any case modern science, from the outset, found itself imbued with a practical perspective, although it was not limited to that. The 19th and 20th centuries saw an increasingly close relationship between fundamental scientific knowledge and technical activity to such a point that the distinction has become essentially meaningless. Contemporary technoscience – where technical development, captured by the market, has largely taken control of scientific development to the detriment of speculative, disinterested research – clearly demonstrates the difficulty with a purely epistemological definition of science.

8 All this then makes it necessary to abandon the myth of a “unity of science” that would enable the formulation of methodological and epistemological standards to which any investigation that wishes to earn the label “quality science” must conform, at least in principle. The historical evolution of the criteria of scientificity in the most established sciences already challenges the idea that the intrinsic, absolute formulation of such criteria is possible.

9 For example, as far as demonstrability is concerned, mathematics itself has many examples where the classic proofs of some fundamental theorems, by mathematicians as revered as d’Alembert, Euler or Lagrange, are now seen as lacking demonstrative rigor on a formal level. Another conventional criterion of scientificity, the reproducibility of experiments, has essentially become obsolete in our era through the sophistication and the cost of experimental processes. This makes it unfeasible to duplicate many observations and measurements, whose validation is ultimately based more on the coherence of their results with other work than on their repetition. So we must recognize that the very nature of the notion of the “scientific proof” is not only provisional but also relative. In the end, the idea that there’s a scientific form of knowledge that’s distinct and separate from all other forms of knowledge is undermined.

10 This is all the more true if we consider the real diversity of the various fields, and still more if we take into account some of the practices of non-Western societies that must now be seen as scientific. In fact, I don’t really see how we can avoid giving the word “science” an essentially functional and institutional definition like: “any activity that takes place within the CNRS [8] (or some similar organization).” That way we could avoid the temptation of according some hypostatic value to the notion of science, which is the basis of both scientism and irrationalism.

11 P. LACOUR: According to Granger, science maintains a complicated relationship with the singular. For a long time (since Aristotle), it has been defined by the universal in contrast with the individual – which is why, for the Stagirite, history cannot claim the rank of science: its value is even less than that of poetry, which generalizes. But with the Galilean revolution in the 17th century, science became more closely linked to measurement, establishing an opposition between the quantitative and the qualitative. Bergson, for example, inherited that division when he associated space with the quantitative domain in order to better separate it from duration, which he reserves for qualities. And yet, as Granger notes, contemporary science is defined more by the structurable (that which can be formalized, in particular through modeling), in such a way that it becomes possible to formalize qualities.[9] Moreover, Granger observes (and encourages) the arrival of mathematics in all of the social sciences, notably (but not exclusively) those mathematical fields dealing with chance (statistics and probability studies). In this transformation of scientific rationality from the universal to the quantitative and then to the structurable, he detects something of a nostalgic desire for the particular, [10] which science, in this view, tries to save from its initial ostracism. In order to explore this tendency, Granger completes the formal pole of knowledge with the identification of a clinical (historical) pole, whose importance for the social sciences is fundamental. Jean-Claude Passeron maintains that formal science, as a result of the limits of formal symbolisms, cannot capture the singularity of the particular, confining itself to specific aspects. Only the natural languages, by virtue of their pragmatic properties, are able to focus on the singularity of individuals in the full detail of their context.[11] The result of this is that the social sciences, which use them extensively (but not exclusively, since those sciences also accommodate formal procedures), establish a special relationship with the singular. This is what makes all the difference between, for example, a statistical approach (in which one examines the changes in a chosen set of variables over time) and a historical one, which, while making a certain selection, deals with the totality of a context in all its complexity and interactions, not just some of its elements isolated as variables. On the level of an individual, then, what’s interesting is not their embodiment of certain particular characteristics, but their singular life history, singular in the sense of its exclusive originality. In other words, the empirico-formal sciences like physics experience universal regularities, all other things remaining equal (ceteris paribus), whereas the social sciences focus on singularities, in different contexts, via certain generalities (ceteris imparibus) – hence their affinities with the clinical method. What are your views on these analyses by Granger and Passeron on the role of the singular in science?

12 J.-M. LÉVY-LEBLOND: I’m not sure that the use of natural languages is particularly useful for clarifying the matter of knowing the singular. These languages are of course essential for enabling a “focus on the singularity of individuals in the full detail of their context,” whether it’s a question of living individuals or material objects, by the way. But their necessity is manifest when it comes to understanding – by that I mean thinking, not just calculating – even in the exact sciences, starting with mathematics, as the experience of the 20th century has shown. [12]

13 The idea that science should speak a “well-made” language, i.e. one that is perfectly logical and unambiguous, obviously goes back to the conception, which we inherited from the Enlightenment, of a transparent and coherent Reason. This quite naturally led to the formalist project of the early 20th century: a “pure” language stripped of all semantic ambiguity, all cultural attachments, which can only be an interplay of signs, an abstract ideography. This was the plan of modern mathematics and logic (Hilbert, Russell, etc.) during those years, a plan that was justified by the “foundational crisis” and the conceptual difficulties that the mathematicians of the late 19th century had come up against. This meant nothing less than the constitution of mathematics as a self-contained discipline, secure in its foundations and methods, immunized against the uncertainties and confusions of common language and thought. This undertaking, while making considerable progress possible in understanding the difficulties that had been encountered, and overcoming some of them, ultimately failed when the formal techniques that it regarded as all-powerful were themselves used by Gödel to demonstrate the inevitable incompleteness of a formalized and axiomatized arithmetic. [13] Consequently, the compromise could no longer be refused, and the project of being able to do without common language was revealed as the fantasy that it is. The same goes for physics, where the use of the most elaborate mathematical formalisms, in articles where one many find hundreds of equations studded with the most abstruse symbols, must ultimately give way to analyses and exegeses in plain language. After all, any concept needs to be named and any idea needs to be expressed, in a sense that reconnects with the etymological origin of the word “express”: to bring out the meaning behind formalism.

14 The contrast between the social sciences and the natural sciences, therefore, does not reside in a reliance on natural languages that would be specific to the first group, but in the fact that this reliance is obviously necessary for them, whereas the second group take it as an embarrassment or, at the very least, neglect to work on it and underestimate the effects of meaning from the inevitable metaphors. Their casual use of language then becomes the source of serious misunderstandings, for example when we see physics unadvisedly using terms like “uncertainties,” “black hole,” “superstrings,” “big bang” and many others. A keener sense of these terms from a linguistic point of view would forestall the mistakes they lead to, not just among laypeople, but also among the specialists themselves.

15 As for distinguishing the sciences of nature by the fact that they search for universal regularities, I think that that pays them an undeserved tribute. [14] As just one example, it’s true that one of the fundamental aspects of physics is its attempt to discover universal theories – but which are only provisionally considered universal! These currently include quantum theory or chronogeometry (which is widely, but unfortunately, referred to as the theory of relativity). [15] But the overwhelming majority of research in physics deals with specific areas of the material world, where the conceptual tools that are used have a limited applicability, and only a distant relationship to general theories. Fluid mechanics, which is so important for meteorology and aerodynamics, has no direct connection with the quantum rules that govern the behavior of individual molecules. And an immense part of the disciplines examining phenomena at speeds below that of light, for example celestial mechanics or atomic physics, never make use of the Einsteinian conception of space-time (except in borderline cases) and simply apply its classical (Galileo-Newtonian) approximation. In other words, while most work in physics may not focus on singularities, to borrow your expression, it does focus on specificities “in different contexts, via certain generalities (ceteris imparibus).”

16 P. LACOUR: Actually, the point is not to deny that language plays a role in the formal or empirico-formal sciences, but rather to emphasize that it does not take on the same importance in those sciences at all. You’re right to stress that some formal systems, like mathematics (but not propositional calculus and first-order logic) are incomplete, but Granger shows above all that their properties differentiate them from natural languages, which are characterized by “protological” aspects (what he calls “pure” pragmatism).[16] This means that the formal and formalized sciences consider the “non-actual”[17] dimension of reality (for example, in classical dynamics, the concepts of speed, acceleration, and mass), whereas the social sciences (and even biology, although Granger does not discuss it) focus on actual, indexically indicated reality (hence the recourse to natural language): Balinese cockfighting (Clifford Geertz), Native American potlatches (Marcel Mauss), the unemployment of the 1930s (John Maynard Keynes), the rationality of Western capitalism (Max Weber), that Sunday in Bouvines (Georges Duby), [18] etc.

17 J.-M. LÉVY-LEBLOND: Again, I’m afraid that the distinction between social and “anti-social” [19] sciences is somewhat exaggerated. First of all, the category of “empirico-formal” sciences hardly seems relevant to me. The natural sciences, while they obviously have an empirical side (but isn’t this true of the social sciences as well, indeed of any science?) are a long way from having the same relationship to formalism. Theoretical physics is entirely formalized because – as I’ve already indicated – it is mathematically constituted. But although chemistry, for example, makes use of empirical and structural formulas, as well as chemical equations, it’s certainly not formalized as deeply or as widely. The earth sciences, like the life sciences, are even less so, except in special cases (the use of geometry in plate tectonics, statistics in population genetics, etc.).

18 In any case, whatever degree of formalization they have, I don’t think that the natural sciences can think independently of the natural language in which formalism embeds itself. Let’s focus on the case of physics. The quantities of classical mechanics that you mention only attain their status as a concept through their denomination, which requires their specification during their implementation. Take the symbols F, m, and a that appear in an equation like Newton’s law, F = ma. In order for them to have a signification, they have to be defined beforehand: “m represents the mass of an object on which a force F is exerted, which gives it an acceleration a.” Once these definitions have been adopted, we can let the formalism proceed according to the computational rules of the mathematical theories invoked, which precisely obviates the need to think. And once these calculations are complete, we will have to return to the concrete situation under analysis and apply the results to the movement of the object in question, thus rediscovering the specific “indexical indication” you mention: the orbit of the Moon around the Earth (Newton) or the discovery of a new planet (Neptune, by Galle), etc.

19 If I may make a somewhat risky comparison, formalism is a mechanized means of transportation for ideas, like the train, the car, etc., are mechanized means of transportation for things and bodies. To go very far, for example to go from Paris to New York, you take an airplane, inside of which you don’t move. But it must be emphasized that this delegation to the machine can only be partial: the beginning of your trip (leaving your home) and the end (arriving at your friend’s house), necessarily take place on foot, by your own corporeal means. The same goes for dealing with a physics problem on a formal level: the wording of the problem and the application of its solution have nothing to do with formalism and require a natural language, precisely because it concerns a particular situation. The airplane can go anywhere (maybe not…), but your trip is specific.

20 P. LACOUR: Some of the disciplines in the social sciences now claim a “clinical” orientation: psychology, but also sociology and anthropology in particular. As a result, the clinical approach loses its exclusively medical sense (of care), and its privileged connection with the pathological, refocusing on diagnosis, which is understood as knowledge of an individual in a given context (the therapeutic and predictive dimensions of the clinic become secondary). Does the term “clinical knowledge,” affecting the individual, or the singular, seem relevant to you? Isn’t it a tendency that we also encounter in some fundamental sciences when it comes to considering complex and unique phenomena: a volcanic eruption, a hurricane, a supernova, not to mention the unique global phenomenon constituted by the current climate crisis?

21 J.-M. LÉVY-LEBLOND: I well understand the concern that these disciplines have for highlighting the research, practices, training, etc. that are carried out among those involved in and interested by the object of study. In so doing they confirm their characterization as social sciences – “social” in the fullest sense of the term. They also distance themselves from the objectivist (or better still, desubjectivized) models proposed by the natural sciences. As for the term “clinic” itself, hopefully its newly expanded meaning will allow it to move away from its medical origins and meet with some response among the natural sciences.

22 P. LACOUR: So how do the exact sciences, especially physics, relate to singularities in your opinion?

23 J.-M. LÉVY-LEBLOND: Physics and the natural sciences in general, like any undertaking of knowledge, can only start with phenomena or single, indeed unique, objects. The geologist will start by becoming interested in a given stone, the entomologist in a given insect, the astronomer in a given meteorite, etc. But these scientists won’t stop until they manage to make these objects specific instances of general categories, thereby subjugating the singular to the collective: this stone is flint, this insect is a green rose chafer, this meteorite is a shergottite, etc. They don’t deny the singularity of a given object: instead, they see it as contingent and therefore defying their explanatory criteria. The particular form of this flint, the slight asymmetry of this insect’s antennae, the indentations on this meteorite: these traits that make these objects unmistakably unique are not handled by the explanations that the sciences mentioned have at their disposal. In fact, understanding these singularities – if not explaining them – would require an almost impossibly long and large study, for a result that would hardly be interesting from a disciplinary perspective, although it could be quite relevant in another context. To take the flint as an example: saying that it’s an artifact, that it has been shaped, immediately classifies it as a prehistoric object, which then leads to an attempt to place it in one of the prehistoric categories: Acheulean biface, Mousterian scraper, etc. So it seems to me that any hypothetical interest the social sciences may have for the singular – the desire to understand it – is not a matter of their specific methods or use of language. It’s just a consequence of their orientation: their human orientation, since – with few exceptions (household pets?) – our personal singularities interest us insofar as we are human individuals.

24 P. LACOUR: And yet the formalized sciences often give the impression that they’re only interested in universal relations (“laws”) between virtual or abstract objects (for example, in classical dynamics, force corresponds to the product of mass and acceleration, while speed is the ratio of distance over time), in particular through the introduction of invariants corresponding to certain parameters of reality (mass in this case). They seem to neglect the specific individual aspects of these objects, except – through the introduction of a frame of reference and the use of the probable – in the reduced form of a local application (finding a solution to a problem in ballistics, for example), or a particular embodiment of a formal property.[20] Does that reputation seem justified to you? Why?

25 J.-M. LÉVY-LEBLOND: I’m not sure that that reputation is deserved. Of course, physics has been in search of abstract universal laws for a long time. But one should note that the very term “law” has essentially disappeared from the terminology of contemporary physical sciences. [21] They’ve become more careful, and are probably more sensitive, if only implicitly, to the provisional and approximate character of their knowledge. On the level of their most general research, in fundamental physics for example, they speak of “theories” (with an “s”!) or “models” or even just “rules.” They thus avoid claiming that their representations of reality are absolutely and definitively adequate. Even when the term “law” is still used to characterize a given relation between physical quantities, this is usually a historical residue that does not mask or no longer masks the approximate, phenomenal validity of that relation. Ohm’s law, a basic law of electricity (V = IR) is a good example of this, because the simple notion of resistance (R) only gives a cursory and approximate description of the behavior of electrons in an electrical conductor, a complex property that more exhaustive investigations put into perspective and explore in more depth.

26 Similarly, and to an even greater extent, the discovery of the ever-increasing richness of phenomena in chemistry and the life sciences means abandoning any hope of formulating “universal relations between abstract objects” that would suffice to underpin and explain the set of observed properties – even if the introduction of some general notions endows vast expanses of knowledge with meaning and structure, as is the case with chemical reactivity or evolutionary genetics. And although the construction of such generic notions remains one of the objectives of research, it doesn’t hinder or stand in the way of the interest in “tangible aspects” of reality. Researchers in chemistry are constantly focused on the multiplicity of molecular structures and the specific nature of their properties, in the same way that those in biology study the extraordinary variety of living species and their behavioral patterns. Modern astrophysics provides a particularly striking example of this active significance of the singular. Observations through the use of large telescopes and explorations by space probes have revealed the unexpected diversity of planets. Those in our Solar System, as well as the countless exoplanets orbiting the many other stars in our galaxy that reveal themselves to us, present properties that are so unusual that they require specific kinds of analyses. So we are driven to understand why our two closest neighbors differ so much from Earth. While the average temperature of our planet is around 15°C, allowing water to remain liquid and enabling the development of life, that of Mars is -60°C (again on average) while that of Venus is 460°C!

27 P. LACOUR: The epistemologist Jean Gayon[22] recalled that there is no “universal law” in biology: there are just essentially local generalizations (except for natural selection). But does that mean – sticking with the case of physics – that there is no meaning to the opposition between the abstraction of the exact sciences and the singularity of the perceived individualities?

28 J.-M. LÉVY-LEBLOND: It’s true that I don’t see any opposition between, on the one hand, the construction by the natural sciences of abstract concepts expressed within general theories and, on the other, these same sciences’ study of particular phenomena and singular objects, whose specificity does not prevent them in any way from being grasped and analyzed, at least in part, by these theories. Of course, they can’t take into account all of the characteristics of the object in question, as I’ve already pointed out concerning the cases of the flint and the green rose chafer. No geologist can explain the specific shape and the exact dimensions of some pebble found on the beach, even if they understand the mechanisms of erosion that have rounded and flattened it, just as no zoologist can account for the precise number of hairs in your cat’s coat, although it knows the origin and the function of that fur. In this sense, certainly, the natural sciences fail to grasp the singularity of the perceived individualities in its entirety. But we have other ways of becoming interested in such specificities: the arts, poetry, contemplation. However, before we criticize sciences for their inability to grasp these singular features, we must recognize that the aesthetic approaches cannot grasp them completely either, given that many of these features elude our perceptions, whether we have sharpened those perceptions or not.

29 The fact remains that the natural sciences would do well to adopt a more definite “clinical orientation,” on an ethical level at least, if not on an epistemic one. On this point, Bakunin delivered a salutary warning by emphasizing that science is only interested in abstract individuals, and not in “James” or a particular rabbit. [23] While the insistence on paying more attention to the question of the individual and the singular does not justify a negative value judgment of the methods and results of the natural sciences, it does make it possible to call on these sciences to exercise more critical self-reflexivity in the choice of subjects for their inquiries, and in the evaluation of the potential consequences of their results. It would only be too easy to concretize these remarks by examining the development of nuclear physics or biogenetics.

30 P. LACOUR: At the beginning of their book Objectivity, [24] Lorraine Daston and Peter Galison give the example of a physicist with a passion for the study of splashes, discovering their ever-changing forms by means of a transition to “mechanical” objectivity, through the use of new (photographic) instruments. Does this mean that the singularity of physical phenomena is always to be discovered at the “business ends” of our technical inventions, as Bachelard’s “phenomeno-technique” suggests?[25]

31 J.-M. LÉVY-LEBLOND: It’s undeniable that the improvements in the techniques for investigating reality, the refinements in observation tools, and the wider range of experimentation lead to a better categorization of the phenomena being studied, whether in physics, chemistry, or biology. The microscope makes it possible to distinguish between certain species of insects that look alike to the naked eye. Astronomical spectroscopy leads to the discovery of visually imperceptible differences between classes of stars. The mass spectrometer enables the separation of different isotopes of a chemical element. But I’m not sure that “singularization” is the appropriate term here, since a reciprocal movement of synthesis accompanies this analytical process, a movement that groups the individual differentiated objects into collective categories. This is clearly seen in zoology where distinct individuals are placed into species, while the species are collected into genera, and so on. It’s also true in chemistry where a deeper knowledge of the differences between the properties of certain molecules (e.g. between sodium carbonate and potash) is accompanied by an understanding of their potential shared behavior, leading to their categorization into functional groups (the bases in this case). It’s true again in astronomy where the very diversity of the composition of stellar spectra and their temperatures allows for their classification into a set of identifiable categories.

32 As a result, instead of using the term “singularization,” I prefer to speak of “discrimination” while linking it indissolubly to the inverse process of conglomeration. There is nothing original in what I’m saying, ultimately. After all, in the clinical approach – in the strict medical sense, first of all – doesn’t the interest in the concrete singularities of a given patient result in considering them as a particular case of a specific class, and not as a unique individual with no known precedent?

33 P. LACOUR: Not necessarily, to be precise. When dealing with an individual, two approaches do seem possible: the first consists in trying to get a better sense of them through a (sometimes evolving) set of characteristics, seen as specific cases of abstract properties. From this perspective, you could be seen as: a teacher, a physicist, French, living in the south of the country, conversant in English… In this case, the comparison consists in placing the case under consideration into the series of similar cases (within the characteristic relationship under consideration): for example, you illustrate the fact of teaching, of knowing physics, etc. What matters then is the abstract property and the fact that it is embodied in a given set of cases (that may be incorporated into a statistical or probabilistic calculation), and not the fact that you are the one who possesses this property instead of someone else (one of your physicist colleagues, for example).

34 The second approach consists in highlighting the uniqueness of the individual under consideration, through a singularizing gaze, and therefore in giving a contrastive, exclusive meaning to the comparison with other cases. This is precisely what Max Weber attempted to do: on the one hand, he attempted to establish a (single) causal link between the Protestant ethic and Western capitalism (characterized by a rationalization of social life). On the other, he tried to show that this link does not exist between other religious ethics (Buddhism, Judaism, Hinduism, Taoism, Confucianism) and the economies of their respective cultures. In his view, this explains the fact that capitalism came into being in the West, and not elsewhere, at a particular time, and not before. In such a perspective, one compares in order to singularize (in reaction to the surprising uniqueness of Western capitalism), not to serialize, which would detract from the specific originality of the case under consideration (for Weber, a universal rule like “the greed for gold” explains nothing).[26] A sort of inversion of the vector of knowledge then becomes apparent: the abstract concept (the ideal type) is not the goal of knowledge, but the means it uses to refine the singularity of its object. By describing it (or recounting it) better, one gets a better sense of the particular causality that produced it.

35 J.-M. LÉVY-LEBLOND: But I don’t see how these two approaches would be mutually exclusive. On the contrary, the “singularizing gaze” will have to establish which characteristics don’t go far enough in specifying the object under study. So it’s therefore necessary to draw on the general categories in order to demonstrate their shortcomings. That said, we should recognize that “the uniqueness of the individual” can be asserted in two ways – which do not contradict each other, incidentally.

36 On the one hand, it can be asserted negatively, by establishing that its known properties cannot exhaust its singularity. So in the case of a scientific object, we would try to discover a new abstract property whose only appeal was its generality, one sufficient to specify a category where the object was merely the first member to be discovered. In this way the Sun, a singular object if there ever was one, is seen from an astronomical perspective as a member of the class of stars of spectral type G2V – and an unremarkable one at that. This does not prevent its uniqueness from being recognized by the life sciences, climatology, etc. But is this so different from the clinical approach in medicine? Recognizing the unprecedented singularity of a patient’s symptoms is only interesting if they become the “patient zero” of a new illness, [27] and if their symptoms are then observed in other individuals.

37 On the other hand, there is a more ordinary way to assert “the uniqueness of the individual”: by noting that no one else shares their particular set of categories. Now I’m probably not the only person who is “a teacher, a physicist, French, living in the south of the country, and conversant in English,” but I am definitely the only one with these characteristics who has my name. We then see that the search for the singular runs a great risk of leading to a tautology. Nevertheless, the issue is not necessarily devoid of interest, since it can be used to call into question the relevance of an abstract category that is invoked to characterize an individual’s membership in a collective class. Here I think of the notion of “genius,” which is so hard, perhaps impossible, to define. What makes it possible to assert that Leonardo, Mozart, Victor Hugo, etc., are geniuses, and what do they have in common other than the fact that they belong to the list of those we recognize as geniuses? As regards the natural sciences, there are obviously situations that must be recognized as exceptional, where a given phenomenon, or a given object, is so unusual or rare within its category that a detailed study is called for. I think, perhaps paradoxically, of the properties of water, which are still poorly understood: for example, the fact that it becomes lighter as it solidifies – ice cubes float! – unlike the overwhelming majority of known liquids. In such cases, without a doubt, “the abstract concept (the ideal type) is not the goal of knowledge, but the means it uses to refine the singularity of its object.” We may therefore recognize that two approaches are possible, depending on whether one is interested in generalities or specificities. But this duality, in my view, does not seem to discriminate at all between the natural sciences and the others.

38 P. LACOUR: As a matter of fact, the ambition of the social sciences does not seem to me to undermine this “ordinary” assertion that would not distinguish them from common sense, but rather to emphasize and clarify the singularity of historical phenomena, through a method of contrastive (non-serial) comparison and to explain them (notably through the use of a particular causality). Incidentally, as a result of this vocation for historical singularities, the social sciences are, according to Weber, “eternally youthful.”

39 J.-M. LÉVY-LEBLOND: I freely admit that the historical dimension, which is a constituent element of the social sciences, plays an essential role concerning the need to recognize the singular and the impossibility of rejecting it. But hasn’t historicity suddenly emerged as one of the new characteristics of the natural sciences as well? Ever since the 19th century, many fields have thus found it necessary to take their temporal dimension into account. Geology has allowed us to discover to what extent the current state of the world is dependent on a history that goes back billions of years. The theory of evolution has shown how life on Earth can only be understood through its history, one that is certainly contingent and therefore singular. What would have happened if a meteorite had not fortuitously struck the Earth 65 million years ago, leading to the death of the dinosaurs and the advent of the mammals? And modern cosmology is an intrinsically historical science, ever since the discovery of the universe’s expansion a century ago.

40 Let’s focus on the physical sciences for a moment. Clearly, one of the many goals of the research in these fields is their unification – on the level of principles if not concrete forms – within increasingly general theories, and eventually within one single theory. However, nothing seems to indicate that this goal can be reached. It seems more likely to me that new discoveries will constantly undermine the success of this project. I would be happy if that perspective, which would not disappoint me at all, lets the natural sciences remain “eternally youthful” as well.

41 P. LACOUR: According to you, one of the issues in the scientific understanding of the singular is tied to the development of technoscience, which plays an increasingly large role in modern society. You’re an experienced critic of this trend.[28] How would you define it? What are the advances that it makes possible? What are its potential risks?

42 J.-M. LÉVY-LEBLOND: The term “technoscience,” which has gained wide acceptance – and rightly so in my view – designates the current mode through which knowledge is both produced and applied. It takes note of the disappearance, in nearly all areas of scientific activity, of an essential distinction between fundamental and applied research. “Pure” science (although the word “pure” is highly problematic – let’s just say speculative, unfinished research) has been largely marginalized in the natural sciences today. The influence, not to say the control, of political and economic forces plays out “upstream” – in the very choices of subjects for research and in the social organization of scientific institutions – and not just “downstream,” in the practical applications of the acquired knowledge.

43 An organizational form of this type undoubtedly allows for productivity gains in the development of technological items intended for consumers and encourages the updating of the products on offer. But it’s reasonable to see this as much more advantageous to the producers, whose profits are thereby replenished through the programmed obsolescence of their inventory, than to the purchasers. As for scientific knowledge as such, which on the face of it is independent of its applications, it can only suffer under such pressure. Moreover, the paradox is that in the context of this imposed shortsightedness, some research that could potentially be rich in promising applications is probably consigned to an uncertain future.

44 P. LACOUR: Today, technoscience too often takes command of thought [arraisonne la réflexion] concerning the norms of action, on the pretext of feasibility (since we know how to do it and we can, let’s do it). And yet this approach can be criticized for normative reasons (of an ethical, legal, or political nature), but also for purely scientific ones. I’m thinking in particular about the current development of artificial intelligence (big data, deep learning), [29] based on statistical technologies, which achieves highly fruitful results, notably in profiling individuals or in targeting personal characteristics. However, there is no clear way of explaining how these results are obtained (the “black box” effect). Does this mean that, in this case, technoscience becomes a blind heuristic (to the detriment of epistemology and its concern for explanation), a technology without science? Don’t we then risk endowing artificial intelligence with the role of standing in for human intelligence, whereas AI’s initial, much humbler goal was to understand human intelligence better?[30]

45 J.-M. LÉVY-LEBLOND: The advent of technoscience is a real step backwards, because it ultimately results in a progressive erasure of theoria to the benefit (financial or otherwise) of praxis. Ever since the 17th century, the ambition of modern science has been to understand the world in order to transform it – let’s set aside any value judgment of those transformations. Technique, which is a much older aspect of the history of humanity than science, had been independent from science for a long time. The “trades and crafts” [arts et métiers] knew how to act upon living or inert matter, without really understanding the nature of their operations. To give just one example, we can mention metallurgy, which, after millennia of empirical progress, led to the superb Damascus steels of the Middle East and Japan, even though the brilliant blacksmiths who developed them were obviously unaware of the nature of metals: their crystalline forms, their phase diagrams, etc.

46 Many recent technoscientific advances, however, remain misunderstood, or not understood at all, on a theoretical level. This includes high-temperature superconductivity, cloning and its low effectiveness, etc., not to mention fundamental negative discoveries like dark matter and dark energy, which constitute most of the mass of the universe. And yet according to the promoters of artificial intelligence, big data [31] will allow us to produce new hypothetical knowledge solely from correlation calculations based on the exploration of the mass of available data, without requiring the identification of any potential causal mechanisms, or the proposal of innovative theoretical arguments. This process has even emerged as a new standard of scientificity: one of the boosters of big data goes as far as to say that the flood of data will make the scientific method obsolete. [32]

47 It may well be the case, therefore, that after four centuries where knowledge of the world went hand in hand with its transformation, we have returned to an era of purely empirical short-term technical development, in complete harmony (if one may say) with political and social changes that are out of control.


  • [1]
    Jean-Marc Lévy-Leblond, “Pour une critique de science,” in La Pierre de touche (Paris: Gallimard, 1996), 149-164, and “La culture scientifique, pourquoi faire?,” in Le Tube à essais: Effervesciences (Paris: Seuil, 2020), 175-198. Enlightened “amateurs” get a “second opinion” from science instead of using it as an authoritative argument. As conceptual “flâneurs,” they interpret knowledge and modestly reorganize what they have learned while clearing up some of their own misconceptions. See the introduction (“Avant/”) and the conclusion (“/Après”) to Aux contraires: L’exercice de la science et la pratique de la pensée (Paris: Gallimard, 1996). On the difference of this approach vis-à-vis epistemological tradition, see “La chauve-souris et la chouette: Petite spectroscopie de la philosophie des sciences,” in La Pierre de touche, 269-285.
  • [2]
    Lévy-Leblond, Impasciences (Paris: Seuil, 2003).
  • [3]
    Maurice Goldsmith, The Science Critic (London / New York: Routledge & Kegan Paul, 1986).
  • [4]
    Against the effects of conceptual “petrification,” see Lévy-Leblond, “La nouvelle Méduse, ou la science en son miroir,” in La Vitesse de l’ombre (Paris: Seuil, 2006), 217-231.
  • [5]
    For the case of Einsteinian relativity, see also Lévy-Leblond, “La relativite aujourd’hui,” La Recherche 96 (January 1979): 23, “La relativite aujourd’hui,” La Recherche 316 (janvier 1999): 83, “La relativité un siècle après,” Les dossiers de la Recherche 18 (March 2005), his preface to Henri Bergson’s Durée et simultanéité (Paris: Garnier-Flammarion, 2021), and “Eu égale emme-cé-deux,” in La Pierre de touche, 309-322.
  • [6]
    See for example Histoire des sciences et des savoirs [3 volumes], ed. Dominique Pestre (Paris: Seuil, 2015).
  • [7]
    Gilles-Gaston Granger, La Science et les sciences (Paris: PUF, 1993).
  • [8]
    The Centre national de la recherche scientifique, which despite its name is an organization with research units throughout France and abroad as well [translator’s note].
  • [9]
    Granger, Pensée formelle et sciences de l’homme (Paris: Aubier, 1960) [Formal Thought and the Sciences of Man, trans. Alexander Rosenberg (Dordrecht NL: Reidel, 1983)].
  • [10]
    Philippe Lacour, La Nostalgie de l’individuel: Essai sur le rationalisme pratique de Gilles-Gaston Granger (Paris: Vrin, 2012).
  • [11]
    Jean-Claude Passeron, Le Raisonnement sociologique: Un espace non-poppérien de l’argumentation (Paris: Albin Michel, 2006) [Sociological Reasoning: A Non-Popperian Space of Argumentation, trans. Derek Robbins (Oxford: Bardwell Press, 2013)]; Lacour, La Raison au singulier: Réflexions sur l’épistémologie de Jean-Claude Passeron (Paris: Presses universitaires de Nanterre, 2020).
  • [12]
    Lévy-Leblond, “La langue tire la science” and “Parler science,” in La Pierre de touche, 228-265; “La méprise et le mépris,” in La Vitesse de l’ombre: Aux limites de la science (Paris: Seuil, 2006), 139-154.
  • [13]
    For a long time, Gödel’s theorems on the incompleteness of arithmetic seemed to be applicable only to highly complex cases. Recently, however, mathematicians have highlighted undecidable expressions that can be formulated in simple arithmetic terms. See Jean-Paul Delahaye, “Mesurer le temps en allumant des mèches,” Pour la science 527 (September 2021): 80-85.
  • [14]
    Lévy-Leblond, “La science est-elle universelle?,” in La Vitesse de l’ombre: Aux limites de la science, 197-216.
  • [15]
    Lévy-Leblond, “Vrai/Faux” and “Absolu/Relatif,” in Aux contraires: L’exercice de la pensée et la pratique de la science (Paris: Gallimard, 1996).
  • [16]
    Granger, “Les conditions protologiques des langues naturelles,” in Formes, opérations, objets (Paris: PUF, 1994).
  • [17]
    See Granger, Le Probable, le possible et le virtuel: Essai sur le rôle du non-actuel dans la pensée objective (Paris: Odile Jacob, 1995)
  • [18]
    This is a reference to Duby’s book Le Dimanche de Bouvines (translated in English as The Legend of Bouvines), on the Battle of Bouvines that took place on Sunday, July 27, 1214. During this battle, the French army under the command of King Philip II defeated an allied army of German, English and Flemish troops led by the Holy Roman Emperor Otto IV [translator’s note].
  • [19]
    Lévy-Leblond, “Des sciences sociales et inhumaines?,” in La Pierre de touche, 128-137; “De l’utilité des sciences sociales et humaines pour celles qui ne le sont pas,” in Le Tube à essais: Effervesciences, 33-41.
  • [20]
    Granger, Le Probable, le possible et le virtuel, chapter 4, 99-128; Sciences et réalité (Paris: Odile Jacob, 2001), chapter 4, 139-172.
  • [21]
    Lévy-Leblond, “La Nature obéit-elle à des ‘lois’?,” in Le Tube à essais: Effervesciences, 199-207. On this topic, see also Bas Van Fraasen, Laws and Symmetry (Oxford: Clarendon Press, 1989).
  • [22]
    Jean Gayon with Victor Petit, La Connaissance de la vie aujourd’hui (London: ISTE Edition, 2018) [Knowledge of Life Today: Conversations on Biology (London / Hoboken NJ: ISTE/Wiley, 2019)].
  • [23]
    See Mikhail Bakunin, God and the State, trans. Benjamin Tucker (New York: Dover, 1970).
  • [24]
    Lorraine Daston and Peter Louis Galison, Objectivity (New York: Zone Books, 2007).
  • [25]
    See also Ian Hacking, Representing and Intervening: Introductory Topics in the Philosophy of Natural Science (Cambridge UK / New York: Cambridge University Press, 1983).
  • [26]
    Max Weber, The Sociology of Religion, trans. Ephraim Fischoff (Boston: Beacon Press, 1993); Weber, The Protestant Ethic and the Spirit of Capitalism, trans. Stephen Kalberg (Oxford: Blackwell, 2002).
  • [27]
    See Luc Perino, Patients zéro (Paris: La Découverte, 2020).
  • [28]
    Lévy-Leblond, “La science à l’épreuve… de la société,” in La Pierre de touche, 25-89.
  • [29]
    “Big data, deep learning”: in English in the text [translator’s note].
  • [30]
    Aurélien Bénel, “Modéliser ce qui résiste à la modélisation,” Revue Ouverte d’Intelligence Artificielle 1.1,, 2020, Web, 17 October 2022, 71-88.
  • [31]
    “Big data”: in English in the text [translator’s note].
  • [32]
    Chris Anderson, “The End of Theory,” Wired (June 2008). There is a sharp and penetrating epistemological critique of this idea in the article by Cristian S. Calude and Giuseppe Longo, “The Deluge of Spurious Correlations in Big Data,” Foundations of Science 22.3 (2017): 595-612.