1. Introduction
Traditionally, philosophers have thought of discovery as the discovery of ideas. A central question in classical debates about discovery has been whether there is a logic to how scientists discover their ideas, theories, hypotheses, and so forth (see Schickore [Reference Schickore and Edward2022] for an overview). In contrast, the discovery of things in the world, and more specifically the discovery of natural kinds and their properties, has received much less attention (Kuhn Reference Kuhn1962; Achinstein Reference Achinstein, Jed and Warwick2001; Arabatzis Reference Arabatzis2006; McArthur Reference McArthur2011; Schindler Reference Schindler2015; Copeland Reference Copeland2019; Arfini et al. Reference Arfini, Bertolotti and Magnani2020; Duerr and Holmes Mills Reference Duerr and Holmes Mills2025).
This widespread neglect of discovery of “things in the world” seems somewhat scandalous: What makes the news and what we all get excited about are not so much the ideas scientists produce but rather whether the ideas lead to something more tangible, such as when the electron was discovered in 1896, gravitational light bending in 1918, the structure of DNA in 1953, or the Higgs boson in 2013. The reason for philosophers’ neglect of discovery cannot be due to triviality: Historians have long known that discovery is not, as Kuhn put it, like “a single simple act assimilable to our usual (and also questionable) concept of seeing” (Kuhn [Reference Kuhn1962] Reference Kuhn1996, 55). However, historians tend to overemphasize the complexity and the contingency of discovery (Brannigan Reference Brannigan1980; Schaffer Reference Schaffer1986; Arabatzis Reference Arabatzis2006). As Hanson provocatively put it, the works by historians “never reflect monochromatically: only spectra of concepts and arguments result” (Hanson Reference Hanson1962, 582).
Luckily, Kuhn did provide some structure (no pun intended!) to our conception of natural kind discovery. In chapter IV of The Structure of Scientific Revolutions (SSR), Kuhn argued that discovery—and his examples suggest that he meant discovery of natural kinds—involves not only observing for the first time that something is the case but also what it is that has been observed (Kuhn [Reference Kuhn1962] Reference Kuhn1996, 55). In other words, a discovery also requires a conceptualization of the newly observed or detected phenomenon. Based on this basic premise, Kuhn introduced a distinction between two types of discovery: a type of discovery in which a novel phenomenon is first conceived of theoretically and then observed, and a type in which a novel, unexpected phenomenon is first observed and then conceptualized. Because the latter type of discovery cannot be accommodated in the reigning paradigm, Kuhn associated this type with scientific revolutions; he associated the former type, of expected discovery, with normal science.
Kuhn’s distinction between the two types of discovery has been largely ignored, possibly because the two types were simply subsumed under the two rubrics of normal science and revolutions, and possibly also because in SSR, the distinction is not as clear as might have been; it is much more visible in an article by Kuhn that appeared in the same year as SSR in Science magazine and that was dedicated specifically to discovery (Kuhn Reference Kuhn1962). Interestingly, in this article, Kuhn does not even mention the central notions of SSR of paradigm change, revolutions, or incommensurability. This goes to show that Kuhn’s distinction can be treated independently from the rest of the controversial Kuhnian conceptual apparatus (see also Schindler Reference Schindler2015). Whereas Kuhn did not give any names to the two types of discovery, I have suggested earlier that we refer to them as that-what discoveries and what-that discoveries (Schindler Reference Schindler2015). That-what discoveries are discoveries in which a surprising phenomenon is first observed and then conceptualized, whereas what-that discoveries are discoveries in which the discovered phenomenon is predicted on the basis of present theoretical resources. I will use this terminology in what follows.
Despite the apparent neglect of Kuhn’s distinction,Footnote 1 there has been some discussion of the premise on which the distinction is based—namely, that discovery of natural kinds requires some substantial and correct conceptualization in the first place. Especially Hudson (Reference Hudson2001) has challenged this premise (see also Achinstein Reference Achinstein, Jed and Warwick2001). Here, I will not engage with these challenges to the Kuhnian premise; I have done so elsewhere (Schindler Reference Schindler2015). My focus in this article is rather on empirically testing this premise and the Kuhnian distinction between two types of natural kind discovery. In order to do so, I analyzed the Nobel Prizes in Physics from 1972 to 2024. My findings support both Kuhn’s basic premise and the distinction.
Why is Kuhn’s distinction worth pondering for philosophers? Well, first and foremost, the question of what a scientific discovery of natural kinds is, at its core, is a question of concept explication on a par with any other prominent questions that philosophers spend their time discussing, such as the following: What is the scientific method? What is a scientific explanation? What is causation? The fact that very few philosophers have hitherto tackled the question, or that they take the question to be trivial (mistakenly, I would argue), does not mean that the question is not a worthy philosophical one. On the contrary, given the centrality of natural kind discovery for science, I believe it would continue to be a scandal if philosophers kept ignoring the question.
As mentioned, my focus here will be on empirically testing an account of scientific discovery of natural kinds. Why should concept clarification have any business in seeking empirical adequacy? According to the standard Carnapian account, concept explication involves several desiderata, one of which is similarity between explicandum and explicatum, which is meant to ensure that the analysis explicates the intended concept and not something else, or even nothing at all (Carnap Reference Carnap1950). This is one way of viewing this project; another would be to treat philosophical theories as akin to scientific ones in that the former are just as much in need of empirical justification as the latter (Giere Reference Giere1985; Donovan et al. Reference Donovan, Laudan and Laudan1988).
What’s in it for those more interested in the practical application of ideas rather than in conceptual clarification per se? Concept clarification is obviously meant to alleviate conceptual confusion, which can have very concrete practical implications. In the case of discovery, confusion about what it is that constitutes a discovery of natural kinds may lead to skewed and unfair credit-attribution practices. Those, in turn, may undermine the reward system of science, which has been argued to be critical for the progress of science (Strevens Reference Strevens2003; Zollman Reference Zollman2018).
The article is structured as follows. Section 2 introduces Kuhn’s distinction and presents the hypotheses for the empirical test. Section 3 expounds the method that was used in the empirical analysis of the Nobel Prizes. Section 4 reports and discusses the results obtained. Section 5 summarizes the results and draws normative conclusions regarding current credit-attribution practices.
2. Kuhn’s distinction and hypotheses
Kuhn’s basic premise that underlies his distinction is best motivated by the example that Kuhn himself used in SSR—namely, the discovery of oxygen. Joseph Priestley was arguably the first person to isolate and notice oxygen as a separate species of gas in 1774 (and to publish it). Until the end of his life, though, he believed that he had discovered dephlogisticated air. But that’s clearly not what Priestley discovered: Phlogiston does not exist. What he did discover, instead, was a gas that the phlogiston theory could not accommodate easily—namely, oxygen. Thus, the discovery of oxygen not only involved that there was a new gas but also a realization about what that gas actually was. Moreover, because in the discovery of oxygen the new entity was first detected and then correctly conceptualized, we can speak of a that-what discovery.
The other type of natural kind discovery—namely, what-that discovery—plays less of a role in both SSR and Kuhn’s Science article; it is still more visible in the latter than in the former. Again, in this type of discovery, the newly detected phenomenon is not at all surprising because it has been predicted by theory or anticipated by the ruling paradigm. Examples that Kuhn gives of what-that discoveries are the new chemical elements predicted by Mendeleev and the discoveries of neutrinos and radio waves (Kuhn 1962, 761; [Reference Kuhn1962] Reference Kuhn1996, 58). As Kuhn memorably put it, this kind of discovery is “occasion only for congratulation, not surprise” (Kuhn [Reference Kuhn1962] Reference Kuhn1996, 52). For example, the discovery of the Higgs particle in 2012 at the Large Hadron Collider (LHC) at CERN was no surprise at all to physicists: It had been predicted by the Higgs mechanism and the standard model of fundamental particle physics almost 50 years earlier.
Naturally, Kuhn associates that-what discoveries with scientific revolutions and what-that discoveries with normal science, the latter of which seeks to increase the scope and precision of a paradigm by way of puzzle-solving (Kuhn [Reference Kuhn1962] Reference Kuhn1996). But as I already mentioned, Kuhn’s distinction between the two types of discovery can be used without committing oneself to the entirety of the Kuhnian apparatus, which involves arguably irrational theory change, incommensurability, perceptual changes, world (!) changes, and more. In this article, I will use this truncated version of Kuhn’s account of discovery (see also Schindler Reference Schindler2015).
Readers who disagree that Kuhn’s account of discovery can be isolated from other elements of his view are invited to conceive of the account of discovery as self-standing. The purpose of this article is not exegetical: I would not want to claim that Kuhn himself would subscribe to every part of it or that the account captures all that Kuhn thought about discovery, but I certainly would still want to describe the account as Kuhnian (see, e.g., Okasha [Reference Okasha2011] for a similar approach to part of Kuhn’s oeuvre). Before articulating hypotheses for the empirical study that I conducted, I will outline the basic elements of this account of discovery in more detail in the next section: the requirement of correctness, the scope of the distinction, and the temporal profile of the two types of discovery.
2.1. The requirement of correctness
An exegetical detail I would indeed stand by is this: Kuhn believed that scientific discoveries of natural kinds require a correct conceptualization of the detected phenomenon. For anybody even only loosely familiar with Kuhn’s philosophy, this statement must appear highly suspect; it requires some explanation.
Let us consider again the discovery of oxygen. Why is Kuhn reluctant to say that Priestley discovered oxygen? After all, it was he who first detected it. Given Kuhn’s view of incommensurability, one might have thought that Kuhn would say that Priestley discovered dephlogisticated air in his paradigm and that this discovery was somehow replaced in the next paradigm. But Kuhn chose not to say that. On the contrary, Kuhn is pretty clear on Priestley not deserving full credit for the discovery because he held the wrong conception. Kuhn is even reluctant to call Lavoisier the discoverer of oxygen. Lavoisier successfully isolated oxygen (upon advice by Priestley) and understood that the new gas called for a radically new conception, but he still held a number of false beliefs about oxygen: For him, oxygen gas was a combination of the “principle” of oxygen (apparently modeled on the “principle” of phlogiston) and caloric, the nonexistent substance of heat, which would not be abandoned until 1860. Kuhn points out that it would be absurd to say that oxygen was not discovered until then. Still, he does think that the developed concepts of the new phenomenon must be somewhat correct. He therefore concludes that oxygen was discovered sometime between 1774 (when Priestley first isolated it) and 1777 (when Lavoisier concluded the new gas was oxygen gas) or, as he says, “shortly thereafter” (Kuhn [Reference Kuhn1962] Reference Kuhn1996, 55).
Hudson (Reference Hudson2001) has called out Kuhn on this rather unsatisfactory degree of vagueness concerning the requirement of correctness; as he put it, Kuhn has “left us with the quandary concerning how well one must conceptualise the discovered object” (78). Hudson’s own proposal is, roughly speaking, that one only has to have enough conceptual resources to successfully identify the new phenomenon in question (Hudson Reference Hudson2001, 77). And for that, the conceptual resources used need not be correct (Hudson Reference Hudson2001, 88). Hudson has no qualms attributing the discovery of oxygen entirely to Priestley. But this does not solve the issue raised by Kuhn: Priestley did not isolate dephlogisticated air but oxygen. This fact must surely be part of any cogent account of the discovery of oxygen.
My own proposal has been that the discovery of natural kind X requires not just an observation of X but also the correct identification of some of X’s essential properties—namely, those properties that suffice to correctly individuate X at a particular time (Schindler Reference Schindler2015). Both theoretical and observational contributions deserve credit. Priestley clearly was the first to observe oxygen. Still, it was Lavoisier, not Priestley, who correctly identified the essential property of oxygen as having a specific mass (in virtue of being a chemical substance). Lavoisier had other beliefs about oxygen that were clearly false, but what he got right about oxygen sufficed to correctly identify it as a new substance at the time. In other words, one need not have a complete description of all the essential properties of X.
Whatever one’s own preference for what a full account of scientific, natural kind discovery might be,Footnote 2 we can here assume that the Kuhnian account entails a correct, but not necessarily complete, conceptualization of the newly observed phenomenon as an instance of a new natural kind or new property. The hypothesis that suggests itself is therefore as follows:
Hypothesis 1: Scientific discoveries of a new natural kind (or a new property of a kind) consist of both an observation of an instance of that kind (or property) and an at least partially correct — but not necessarily complete — conceptualization of that instance as belonging to that kind.
A competing hypothesis would be that it is sufficient for a discovery of a natural kind that a new phenomenon is observed, even when the provided conceptualization is not at least partially correct or otherwise faulty (e.g., ad hoc, with little independent support).
2.2. The scope of Kuhn’s distinction
Kuhn made clear that he did not believe that all scientific discoveries of natural kinds fell into the two types of discovery that we have discussed (Kuhn Reference Kuhn1962, 761n3). Kuhn mentions the discovery of the positron as an example, where Anderson first observed the positron in 1932, without much theoretical guidance, and where Blackett and Occhialini at the same time obtained experimental evidence for the positron, using Dirac’s equation, which predicted antimatter (see Hanson Reference Hanson1963). It seems that in the hands of Anderson, the discovery was a that-what discovery, whereas it was a what-that discovery in the hands of Blackett and Occhialini. However, Blackett and Occhialini did not publish their results until after Anderson’s article came out. This is surely relevant to the question of who discovered the positron and what type of discovery this is: Only published results are available to the scientific community and thus verifiable. Consequently, it was Anderson, not Blackett and Occhialini, who received a Nobel Prize for the discovery (in 1936). Of course, the date of publication is contingent and could have been the other way around. Then we may have had to group the discovery as a what-that discovery. In either case, a discovery will fall into either of these two classes, unless both kinds of discoveries are published at exactly the same time. I know of no such case.
It is plausible that individual discoveries are connected in longer discovery chains. As we shall see later in this article, there are examples of that-what-that discoveries, where a new phenomenon is first observed, then the phenomenon is conceptualized, and then this new conceptualization leads to predictions of new phenomena that are then observed (see sec. 4 for examples). Still, we can analyze such a discovery “chain” as a that-what discovery that was followed by a what-that discovery. The two types of discovery are thus best conceived of as minimal units, which may very well be connected to other discoveries, as in the aforementioned way or otherwise.Footnote 3 I therefore set out to test the following hypothesis:
Hypothesis 2: Scientific discoveries of natural kinds fall into two basic classes: what-that discoveries and that-what discoveries.
From the way Kuhn describes the two kinds of discoveries of natural kinds (see earlier discussion), we can derive the following:
Hypothesis 3: What-that discoveries of natural kinds are expected, and that-what discoveries are surprising (both relative to the accepted conceptual resources at the time).
Note that these first three hypotheses are clearly empirically testable: Hypothesis 1 would be wrong if it would suffice for a natural kind discovery to detect a new phenomenon, or vice versa, if it would suffice for a natural kind discovery to have an incorrect conception of the detected phenomenon; hypothesis 2 would be false if there were only one type of discovery or if there were several types of discovery, none of which would fall under these two basic types; and hypothesis 3 would be false if what-that discoveries were described as surprising or that-what discoveries were described as expected.
There are further hypotheses that one can draw from Kuhn’s account. They concern that-what discoveries and revolutionary paradigm change, the temporal profile of the two types of discoveries, and the relative importance of observation and theory. I will reserve a section for each in what follows.
2.3. Are all that-what discoveries revolutionary?
Kuhn poses the question of whether all that-what discoveries involve paradigm change and answers it hesitantly: “To that question, no general answer can yet be given” (Kuhn [Reference Kuhn1962] Reference Kuhn1996, 56). Kuhn then discusses the discovery of X-rays in 1895 by Roentgen and argues that although the discovery did not require a change in paradigm theory, it did bring about a paradigm change with regard to the instrumental and experimental techniques used by scientists when conducting cathode-ray experiments (Kuhn [Reference Kuhn1962] Reference Kuhn1996, 58–59). Kuhn does not comment further on this question, so it would seem that Kuhn believed that at least most that-what discoveries require revolutionary paradigm change of some sort or another, be it theoretical, instrumental and experimental, or other. Furthermore, he did not associate what-that discoveries with revolutions at all: For him, they were just part of normal science. This gives rise to our fourth hypothesis:
Hypothesis 4: That-what discoveries require revolutionary paradigm change, whereas what-that discoveries do not.
Of course, Kuhn thought that revolutionary paradigm change entails the incommensurability of paradigms. However, I am not willing to make this commitment; it is also not necessary to recognize paradigm change as revolutionary, that is, as truly novel, far-reaching, disturbing with regard to previously held views, and so forth. In this article, we will use this more modest meaning of “revolutionary.” Again, the main purpose of this article is not exegetical; readers unhappy with this truncation are invited to think of the account used in this article as self-standing (see also my remarks at the beginning of sec. 2).
2.4. Temporal profiles and epistemic uncertainty
Kuhn attributes different temporal profiles to that-what and what-that discoveries. Relative to the observation of the new phenomenon, Kuhn thinks of that-what discoveries as “necessarily” extended in time, whereas what-that discoveries occur in an “instant” (Kuhn 1962, 761–62; [Reference Kuhn1962] Reference Kuhn1996, 52–56). That is so because in what-that discoveries, scientists are conceptually prepared for what is to come, and when they observe the new phenomenon, the discovery is completed. In contrast, in that-what discoveries, when the new phenomenon is first observed, time is required to conceptualize the new phenomenon. It is for this reason that Kuhn also thinks that it is hard to precisely date these kinds of discoveries, and accordingly, it is why these discoveries are more often associated with priority disputes (Kuhn does not provide any systematic evidence for the latter claim; Kuhn Reference Kuhn1962, 760–61).
It is questionable how important this temporal contrast really is to Kuhn’s distinction. After all, there is a sense in which what-that discoveries are also extended in time; Kuhn’s decision to measure the time extension of discoveries relative to the observation of the new phenomenon—rather than relative to the conceptualization—seems arbitrary to some extent. What seems more important to Kuhn’s distinction is that at the time of first observation of the new phenomenon, there is epistemic uncertainty involved in that-what discoveries regarding the identity of the discovered phenomenon that is absent in what-that discoveries. A relevant hypothesis is therefore as follows:
Hypothesis 5: After the first observation of a new natural kind, there is greater epistemic uncertainty about its identity in that-what discoveries than in what-that discoveries.
3. Analyzing the Nobel Prizes: Method
In order to test the Kuhnian account of natural kind discovery and the five hypotheses formulated in the previous section, I chose to analyze the Nobel Prizes in Physics over the past 53 years. The Nobel Prize is widely regarded as the highest and most important accolade in the natural sciences. An analysis like the present one should be based on the best and most important bits of science, and the Nobel Prizes almost guarantee to highlight such discoveries.
Because Kuhn’s preferred field was physics, the Nobel Prizes in Physics were a natural choice for my analysis. I restricted my analysis from — at the time of the current analysis — the most recently awarded Nobel Prize in 2024 to 1972 for a significant practical reason: The text analysis that I carried out was based on documents provided by the Nobel Committee, which do not reach back further than 1972 (NobelPrize.org 2025). Also, there are added complications with the earliest Nobel Prizes, such as a heavy bias against any theoretical work until WWI (Friedman Reference Friedman2001). No other systematic biases have been reported after WWI that would be relevant to the current study.Footnote 4
Previous analyses of the Nobel Prizes have focused on the individual scientists receiving the prizes, such as their age, nationality, and area of expertise (Karazija and Momkauskaitė Reference Karazija and Momkauskaitė2004; Nilesh and Pranav Reference Nilesh and Pranav2018; Bjørk Reference Bjørk2019). The current study chose a different unit of analysis—namely, the content of the prizes. Each Nobel Prize (NP, for short) is associated with a very short phrase indicating what the prize has been awarded for. The Nobel Committee refers to these phrases as “prize motivation” (they can also be found on the individually designed Nobel diplomas), but because of their brevity, I will refer to them here as “praise phrases.” By the statutes of the Nobel Foundation, each prize can be divided between a maximum of two contributions to science, “each of which is considered to merit a prize.” (NobelPrize.org 2025) That means that one prize has maximally two praise phrases, each of which I effectively treated as a separate prize (barring phrases summarizing two praise phrases under a common theme, which I ignored).Footnote 5 Furthermore, the Nobel Prize in Physics cannot be awarded posthumously, cannot be shared by more than three people, and cannot be awarded to institutions (in contrast to the Nobel Peace Prize).
In the period from 1972 to 2024, there were 53 Nobel Prizes in Physics. Of those, 18 came with 2 praise phrases each, amounting to 71 praise phrases overall. The current study focused on only praise phrases associated with the discovery of new natural kinds; there were 33 of them. There were three other categories of praise phrases, which I ignored: general theoretical contributions to the field (without specified natural kind), instrumental inventions, and modeling achievements (also without specified natural kind). Table 1 lists these categories and the respective counts of praise phrases.
Table 1. Number and Percentages (Rounded) of Praised Achievements in the Physics Nobel Prizes, 1972–2024, Distinguished by Category
| Natural kinds | General theoretical | Instrumental | Modeling | All | |
|---|---|---|---|---|---|
| Praise phrase (pp) | 33 | 10 | 27 | 1 | 71 |
| % of all pp | 46% | 14% | 38% | 1% | 100% |
Here is an example of a praise phrase that honors the discovery of a new property of a natural kind—namely, the structural duality of collective and individual nucleon motion in atomic nuclei (NP 1975):
-
Aage Niels Bohr, Ben Roy Mottelson, and Leo James Rainwater, “for the discovery of the connection between collective motion and particle motion in atomic nuclei and the development of the theory of the structure of the atomic nucleus based on this connection” (NobelPrize.org 2025)
Although in this example, both theoreticians (Rainwater and Bohr) and an experimenter (Mottelson) were honored, in most cases, the Nobel Committee bestowed the prize upon either experimenters or theoreticians: 21 versus 8 of 33, respectively. For example, the NP 2013 for the discovery of the Higgs boson went only to the theoreticians (although the experimental contributions are also clearly acknowledged in the praise phrase):
François Englert and Peter W. Higgs, “for the theoretical discovery of a mechanism that contributes to our understanding of the origin of mass of subatomic particles, and which recently was confirmed through the discovery of the predicted fundamental particle, by the ATLAS and CMS experiments at CERN’s Large Hadron Collider” (NobelPrize.org 2025)
Only in four cases did the Nobel Committee award the prize to both experimentalists and theoreticians for the same natural kind discovery (as in the case of NP 1975, mentioned earlier). Whether or not the Nobel Committee conferred the prize to experimenters, theoreticians, or both was not part of my analysis of natural kind discoveries (however, I will come back to the issue of credit distribution in the final section of this article).
Let us now consider examples of instrumental (26 instances), modeling (1 instance), or theoretical achievements with no particular natural kind discovery but rather more general theoretical contributions to the field (10 instances):
-
Instrumental: For example, Bertram N. Brockhouse, “for the development of neutron spectroscopy,” and Clifford G. Shull, “for the development of the neutron diffraction technique” (NP 1994; NobelPrize.org 2025)
-
Modeling: For example, Syukuro Manabe and Klaus Hasselmann, “for the physical modelling of Earth’s climate, quantifying variability and reliably predicting global warming” (NP 2021; NobelPrize.org 2025)
-
Theoretical; no particular observation: For example, James Peebles, “for theoretical discoveries in physical cosmology” (NP 2019; NobelPrize.org 2025)
For all discoveries whose Nobel Prize praise phrase mentioned the discovery of a new natural kind or property, I recorded the years in which the observations pertaining to a discovery were made (the “that”), as well as when the theoretical understanding of it was developed (the “what”), if at all. The Kuhnian account of discovery would lead us to expect a list of what-that and that-what discoveries to result from this (see hypotheses 1 and 2). In order to see how the discoveries were described (“surprising,” “expected,” or “revolutionary”) and whether hypotheses 3 and 4 were correct, I then conducted a text analysis of the press releases and explanatory documents that the Nobel Foundation provides on its website back to the NP 1972 (NobelPrize.org 2025). For this analysis, the text associated with each discovery was hand-coded. The results were cross-validated by an independent coder. The test of hypothesis 5 was a bit more involved and will be explained later in the results section.
It turned out that there were several complications in categorizing discoveries by simply recording the observation of a new phenomenon and its correct conceptualization. This became clear from a closer study of the texts accompanying the discoveries and from the way the discoveries were described. For the remainder of this section, I will explain these complications. Readers not interested in these complications may now skip to the results discussed in section 4.
Most importantly, perhaps, there were discoveries where a theory was developed before the observations, but where the theory so happened not to guide the experiments. This, for example, was the case for the first compelling evidence for the existence of quarks. James Bjorken predicted the scaling behavior of subatomic (hadronic) particles in 1967, but the MIT–Stanford Linear Accelerator Center (MIT-SLAC) deep inelastic scattering experiments of 1968 unexpectedly discovered scaling and then sought Bjorken’s help for interpreting the results (Riordan Reference Riordan1992). For our purposes, the discovery thus counts as a that-what discovery. The Nobel Prize of 1990 went to the experimentalists (Bjorken never received a Nobel Prize). Other examples in which the relevant theory was available to the experimenters in principle but not used concern Arno A. Penzias and Robert Woodrow’s discovery of the cosmic microwave background in 1965 (NP 1978)Footnote 6 and Michel Mayor and Didier Queloz’s 1995 discovery of the first exoplanet, called “51 Pegasi b,” for which they received the NP 2019.
There were discoveries that proved to be deceptive. Again, Mayor and Queloz’s discovery of “51 Peg” is such a case. Their discovery may at first seem to be a straightforward observation of planetary motion in just another solar system. Thus, one may think that this discovery should be grouped as a what-that discovery. However, it turns out that the discovery was in fact unexpected and required some real theoretical work: 51 Peg has the size of Jupiter but is surprisingly close to its sun. Many astronomers thus at first suspected that the observations were caused by other effects, such as stellar pulsation and star spots. It had to be argued on theoretical grounds that 51 Peg had actually formed at much larger distances away from its sun (around 5 AU instead of 0.5 AU) and then migrated toward the sun before it could be accepted that the first exoplanet had been discovered (NobelPrize.org 2025).Footnote 7 It is clear from the documents provided by the Nobel Foundation that Mayor and Queloz initially experienced some significant resistance in getting their discovery recognized as an exoplanet (see sect. 4.3). I therefore categorized this discovery as a that-what discovery.
There were also discoveries that seemed to combine two types of discovery in one award, as well as discoveries that spread over several awards. Let us begin with the first kind of case, which can be illustrated with the NP 1975, which, as already mentioned, was awarded to Bohr, Mottelson, and Rainwater for their work on models of the atomic nucleus. Rainwater’s work was prompted by observations of aberrations from spherical symmetry in the charge distribution of certain atomic nuclei—aberrations known as electric quadrupole deformations—which were at odds with the existing liquid drop models. Rainwater then proposed that this observed asymmetry arose from an asymmetry in the distribution of nucleons within the nucleus. Rainwater’s contribution may thus be thought of as a contribution to a that-what discovery. Rainwater’s ideas were then further developed by Bohr into a full-fledged theory of collective motion in deformed nuclei, which was then confirmed experimentally by Mottelson. This part of the discovery should clearly count as a what-that discovery. Because the Nobel Committee opted not to give the award to the observations prompting Rainwater’s theoretical work, and because the prize went to Bohr for his theory and Mottelson’s confirmation of the theory instead, I decided to categorize the discovery as a what-that discovery.
There are discoveries that have spread over several prizes. For example, charge conjugation–parity (CP) symmetry violation was first observed in kaon decay in 1964 by James W. Cronin and Val L. Fitch and explained in 1973 by Makoto Kobayashi and Toshihide Maskawa, who integrated symmetry breaking into the standard model by introducing a third quark generation (namely, bottom and top). Their model was supported by the discovery of this third generation of quarks (1977 and 1995) and fully confirmed by experiments measuring CP violation in B-meson decays in 2001, as predicted by the Kobayashi–Maskawa (KM) model (BaBar at Stanford and Belle at Tsukuba, Japan). Clearly, the episode could be described as a discovery “chain” (namely, a that-what-that discovery), but this chain is still constituted by the two basic types of discovery: a surprising that-what discovery (for which the experimentalists Cronin and Fitch received the NP 1980) and a what-that discovery resulting from a prediction of the KM model (for which the theorists Kobayashi and Maskawa received the NP 2008).
There was another example that followed a similar pattern—namely, the discovery of solar neutrinos, which resulted in two Nobel Prizes (2002 and 2015). Neutrinos were already postulated in the 1930s, but it was not until the 1950s and 1960s that models of the expected flux of solar neutrinos were developed by John Bahcall. The famous Homestake experiments led by Raymond Davis—and later also the Kamiokande experiments led by Masatoshi Koshiba—then detected a significant discrepancy between the observations and the predictions of these models (by not less than one-third). This discrepancy was later explained in terms of neutrino oscillations.Footnote 8 Intriguingly, Davis and Koshiba did not receive the NP 2002 before this explanation had been confirmed in experiments by Takaaki Kajita and Arthur B. McDonald in 2001, for which they, in turn, received the NP 2015. The episode can again be described as a discovery chain consisting of a that-what discovery and a what-that discovery.
4. Results and discussion
In this section, I will report the results I obtained from my analysis of the Nobel Prizes in Physics from 1972 to 2024, with an aim of testing Kuhn’s account of scientific discovery. I will first report the results relevant to hypotheses 1 and 2 (sec. 4.1), followed by hypotheses 3 and 4 (sec. 4.2), and then hypothesis 5 (sec. 4.3).
4.1. Hypotheses 1 and 2: Correct concepts and the two types of discovery
The results confirm hypothesis 1: For the period I looked at, there was only a single discovery that received the Nobel Prize before it was correctly understood what the phenomenon was that was discovered. There were no prizes that were made before an at least partially correct (and not necessarily complete) conception of the new phenomenon had been provided. Remarkably, none of the developed conceptions later turned out to be incorrect or incomplete, except one.Footnote 9
Hypothesis 2 was also fully confirmed: All of the natural kind discoveries that I identified fall into the classes of what-that discoveries and that-what discoveries. More specifically, the study found 19 that-what discoveries and 14 what-that discoveries. As what-that discoveries, I counted (parts of) the Nobel Prizes of the following years: 2022, 2020, 2017, 2015, 2013, 2008, 2006, 2001, 1995, 1993, 1984–85, 1979, and 1975. As that-what, I counted the years 2019, 2011, 2005, 2003(2×), 2002, 1998, 1995–96, 1990, 1988, 1987, 1982, 1980, 1978, 1976, and 1972–74. See the appendix at https://osf.io/wyf2q/ for a complete list of discoveries.
4.2. Hypotheses 3 and 4: Discovery predicates
The way that the discoveries were described on www.nobelprize.org—namely, as “expected,” “surprising,” or revolutionary—matched their categorization as that-what or what-that discovery almost perfectly (see table 2).Footnote 10 The cross-validation by an independent coder obtained an excellent match (Cohen’s kappa = 0.862).Footnote 11 I consider hypothesis 3 to be confirmed.
Table 2. Nobel Prize Descriptions. The table shows descriptions associated with the two types of discovery. Percentages are relative to the total number of discoveries of the respective kind
| Surprising | Expected | Revolutionary | |
|---|---|---|---|
| That-what (19) | 100% (19) | 0 | 74% (14) |
| What-that (14) | 7% (1) | 93% (13) | 50% (7) |
The analysis also revealed that that-what discoveries were more frequently described as revolutionary than what-that discoveries (74% vs. 50%). The intercoder cross-validity was again excellent (Cohen’s kappa = 0.832). This result is ambiguous with regard to hypothesis 4: Although the difference seems to accord with Kuhn’s view that that-what discoveries are revolutionary, still 25% of that-what discoveries were not described as revolutionary. While this may be blamed on poor description in the analyzed documents, it’s perhaps more significant that still half of all what-that discoveries are described as revolutionary, too. This is clearly not how Kuhn thought about these types of discoveries.
4.3. Hypothesis 5: Epistemic uncertainty
As mentioned in section 2, in the Kuhnian view of discovery, there is more epistemic uncertainty about the identity of the discovered phenomenon in that-what discoveries than there is in what-that discoveries (this is hypothesis 5). One way in which this epistemic uncertainty could express itself is in the temporal distance between the observation of the new phenomenon and its first correct conceptualization: One might expect that distance to be larger in that-what discoveries than in what-that discoveries. Table 3 shows that this is not at all the case: Whereas for that-what discoveries the median temporal distance is 1 year, it is 38 years for what-that discoveries (table 3).Footnote 12
Table 3. Median (and Mean) Temporal Distance in Years between Observation and Conceptualization. The table indicates the median (and mean) time passed (in years) between the observation and conceptualization (in that-what discoveries), and vice versa (in what-that discoveries)
| Time lag between that and what: median (and mean) | |
|---|---|
| That-what | 1 (13.1) |
| What-that | 38 (26.9) |
At first glance, one may think that this result contradicts Kuhn’s view that that-what discoveries “necessarily” take time, whereas what-that discoveries happen in “an instant” (see sec. 2.4): Here, it seems to be the reverse. Yet we must remind ourselves that Kuhn makes this statement relative to the first observation of the new phenomenon. Thus, the time that passes between the prediction and the first observation in what-that discoveries is irrelevant to this claim. I therefore conducted another test, which I will get to after commenting on this result in a bit more detail.
An obvious explanation for the difference recorded in table 3 is that the required observations can be hard to come by: It can be technologically challenging to build and conduct the experiments necessary for the confirmation of theoretical predictions (in what-that discoveries). The discovery of the Higgs boson, which required millions of euros and many years to build the LHC, is a case in point. Not all experiments are as difficult, but there could be a preselection of especially arduous experiments for the Nobel Prizes because it might be precisely those kinds of what-that discoveries that the Nobel Committee finds prize-worthy.
By comparison, developing the right concepts can be cheap, as is evidenced by the small temporal difference between the observation and the right concept. There were in fact six discoveries in which the right concept was developed in the same year as the observation. Then again, there are also cases in which it took a very long time to come up with the right ideas: As already mentioned, it took 46 years for superconductivity to receive the correct theoretical explanation (NP 1972). Thus, while the development of new concepts is not guaranteed to be relatively fast, compared with experimental confirmations, it is much faster on average.
The other way that I went about testing hypothesis 5 was by measuring the temporal distance between the time when a discovery was completed and the time when the Nobel Prize was awarded. The underlying idea was that more epistemic uncertainty would result in the Nobel Committee taking more time to award a prize. Because Kuhn attributes greater epistemic uncertainty about the new phenomenon to that-what discoveries, it should be these discoveries for which the Nobel Committee should take relatively more time to make an award. This prediction was borne out by the results (see table 4): It took the Nobel Committee more than twice as long to award the prize to that-what discoveries as it did to award it to what-that discoveries. I take this to confirm hypothesis 5.
Table 4. The Two Discoveries and the Median (and Average) Temporal Distance from the Completion of the Discovery to the Award of the Prize (in Years)
| That until prize | What until prize | |
|---|---|---|
| That-what | 20 (30.7) | 15.5 (17.6) |
| What-that | 7 (8.2) | 52 (47.6) |
The reason that suggests itself for this difference is that it takes the community more time to assess whether the conceptualization is appropriate in that-what discoveries, whereas in what-that discoveries, the community has had ample time to do so before the observation was finally made. This explanation is also supported by the descriptions that can be found in the documents provided by the Nobel Foundation (NobelPrize.org 2025). Consider, for example, an excerpt from the Nobel lecture by Didier Queloz for his discovery of the first exoplanet, for which he (and Michael Mayor) received the Nobel Prize in 2019:
In the following years we would be confronted with a wave of skepticism. It would take years for the community at large to accept the reality of 51Peg hot Jupiter and to modify the paradigm about the universality of solar system planetary architecture […] The main issue was that it didn’t fit in the planetary formation paradigm without seriously tweaking this paradigm. Changing a well-established theory is rarely the first idea a physicist is considering out of an unusual experimental result. And yet the foundation of planet formation theory needed to be revised. (Queloz Reference Queloz2019, 121–22)
Clearly, there was epistemic uncertainty about the identity of the newly observed phenomenon. Comments like these are not at all untypical for that-what discoveries. Also consider, for example, Martin Perl’s Nobel Lecture for the discovery of tau muons, for which he received the NP 1995. In a section entitled “Is It a Lepton?,” Perl writes that “our first publication was followed by several years of confusion and uncertainty about the validity of our data and its interpretation,” among other things, because theory did not require a third charged lepton, and there was also otherwise no reason to expect a third lepton (Perl Reference Perl1995, 186–87). Establishing the identity of the new phenomenon also requires ruling out all sorts of competing interpretations of experimental results (Perl Reference Perl1995, 187–88).
5. Conclusion and normative implications
The last 53 years of Nobel Prizes suggest that Kuhn’s distinction between two types of natural kind discovery is indeed an insightful one to make. According to Kuhn, any scientific discovery of natural kind X requires not only an observation of X but also a correct conceptualization of X. This is clearly evidenced by the Nobel Committee refraining from awarding prizes for the discovery of phenomena for which no adequate understanding had yet been developed; except one, none of them turned out to be incorrect.
The current analysis also found that all natural kind discoveries awarded a Nobel Prize fell into the category of either a what-that discovery or a that-what discovery. Most significantly, perhaps, in the material provided by the Nobel Foundation, that-what discoveries are consistently described as “surprising,” whereas what-that discoveries are described as “expected.” The current study also found evidence for the view that that-what discoveries are associated with much greater epistemic uncertainty. While that-what discoveries were more frequently described as “revolutionary” than what-that discoveries, still half of the latter were described as revolutionary, too. This result is clearly contrary to what Kuhn thought, but in my view, it does not fundamentally undermine the usefulness of the distinction, especially if one takes some distance from several elements in the Kuhnian package of ideas (as this article does).
The Kuhnian distinction between the two types of natural kind discoveries allows us to bring order to an otherwise unstructured set of discoveries. But it does more than that. It also has normative implications. In particular, if the Kuhnian picture of scientific discovery is true and a discovery of a new natural kind or its properties requires both an observation and a theoretical understanding, then recognizing only one of these elements would clearly not do justice to the discovery, nor would it to the discoverers. As we saw, the way that the Nobel Prizes have been awarded suggests that the Nobel Committee at least implicitly embraces the Kuhnian view of discovery: Again, the prizes are not awarded for observations of new phenomena whose identity is not yet understood; all natural kind discoveries fall into Kuhn’s two classes with their distinct characteristics; and that-what discoveries must await their prizes longer than what-that discoveries, indicating greater epistemic uncertainty on part of the Nobel Committee and the wider community. And yet, despite the Nobel Committee’s apparent espousal of the Kuhnian view, its decisions concerning which contributions to highlight in the awards and whether to award the prize to theoreticians or experimentalists are rather inconsistent and exhibit a strong bias for experimental contributions. Both of these aspects are problematic. Let us consider these two issues briefly in turn.
Sometimes the Nobel Committee rewards only theoretical contributions, sometimes only experimental contributions, and only occasionally both, without apparent reason. For example, the NP 1984 for the discovery of the W and Z particles went only to the experimentalists (Rubbia and van der Meer), and the NP 2013 for the discovery of the Higgs boson went only to the theoreticians (Higgs and Englert), even though they are both what-that discoveries in high-energy physics.Footnote 13 It would have been open to the Nobel Committee to reward the other element of the discovery (theoretical or experimental) as well by recognizing a third scientist in each of these cases, and maybe this is what the Nobel Committee should have done. Another salient case is the discovery of nuclear fission, for which only Otto Hahn received the Nobel Prize, and not Lise Meitner, whose theoretical understanding of the detected physical processes was absolutely crucial to understanding them as fission in the first place. She, too—it is widely agreed—should have received a Nobel Prize.Footnote 14
These are not exceptional cases. As table 5 shows, only a small fraction of all that-what and what-that discoveries were discoveries in which both the experimenters and theoreticians were awarded a Nobel Prize (11% and 14%, respectively). Table 5 also shows that the Nobel Committee clearly sports a preference for experimental contributions for both types of discovery.Footnote 15
Table 5. Awarded Contributions with Respect to the Two Types of Natural Kind Discoveries. The table indicates the number and percentages of the types of contributions to the Nobel Prizes that were recognized by the Nobel Committee. Percentages are relative to the overall number of discovery types (that-what: 19; what-that: 14). Percentages are rounded
| That-what | What-that | |||
|---|---|---|---|---|
| Both experimenters and theoreticians | 2 | 11% | 2 | 14% |
| Experimenters only | 12 | 63% | 9 | 64% |
| Theoreticians only | 5 | 26% | 3 | 21% |
Even when taking into consideration the fact that the Nobel Committee does not award Nobel Prizes posthumously, and assuming that it might have recognized each of the deceased discoverers, this would bring us only up to 32% of that-what discoveries and 36% of what-that discoveries for which both contributions could have been awarded.Footnote 16 The practice would thus still be rather far off the ideal.
One might object: Why think that the Nobel Committee’s credit-distribution practice is not ideal, instead of concluding that it is the Kuhnian account that is, after all, wrong?Footnote 17 Note, though, that facts about award emphasis—namely, whether the prize should go to theoretical or experimental contributions—are not facts that could possibly undermine the facts obtained in support of the Kuhnian account concerning natural kind discovery. Those latter facts, recall, are facts about whether a correct conception of the newly observed phenomena had been found at the time of the award (independently of whether that contribution received a prize), whether the discoveries fall into the two classes identified by Kuhn, and whether they have the characteristics assigned to them by Kuhn. In other words, facts about award emphasis and facts about natural kind discovery are evidentially orthogonal.
While the current study presented evidence that both experimental and theoretical contributions are required for a natural kind discovery (in accordance with the Kuhnian view), the current study in no way excludes the possibility of self-standing theoretical, experimental, or other kinds of discoveries. As far as I’m concerned, one may even speak of the contributions to a natural kind discovery of X as a theoretical and experimental discovery of X, so long as one appreciates that both are required for natural kind discovery.
The current study has limitations. First, one may argue that the discoveries highlighted by the Nobel Prize must satisfy the highest standards for discovery and that these standards for discovery may be much weaker for science that is getting less limelight. Whether or not this is the case cannot be determined here. The second limitation is fairly obvious: The current study focused on physics and only on a subset of all Nobel Prizes awarded in this area. It would be interesting to see to what extent Kuhn’s distinction also applies in the other scientific disciplines in which the Nobel Prize is awarded for natural kinds.
I should emphasize, at last, that the primary goal of this article is not to criticize any specific award practices. Simply by virtue of the very small number of awards and their limitation to three awardees, it is quite obvious that the Nobel Prizes will never be able to do full justice to the complexity of discovery. Rather, the explication of the concept of natural kind discovery offered here—beyond providing insight—is meant to motivate broader reflection on the question of how, in general, credit ought to be distributed in science, which I think is thoroughly needed.
Acknowledgments
I wish to thank two anonymous reviewers for their very helpful feedback. Thanks also to the audiences at the Bergen Workshop for the Philosophy of Science (2024), the Hong Kong University of Science and Technology, the Epistemology of Experimental Discovery Workshop at the University of Bristol (2025), and the Centre for Science Studies at Aarhus University.
Funding Statement
None to declare.
Declarations
None to declare.
Open access
