There are two basic views in philosophy about the nature of the groupings and classifications that we use to deal with reality. For many philosophers, both today and in the past, reality is like a cake: there are various legitimate ways to cut it into pieces, and which way we should adopt depends not on a fundamental structure of reality itself, but on other factors—e.g., our practical interests and needs, or contingent facts about our linguistic practices. For others, reality is rather like an animal in the butchery: there is a privileged way to divide it into pieces, which cuts reality at its joints. Those who favour the second view face serious metaphysical tasks: not only do they have to provide an argument for the view that reality is structured by joints at all, they also have to develop and defend an account of what those joints exactly consist in—otherwise, the assumption of such joints risks being a mere appeal to mysterious entities.
A particularly influential way of accounting for the nature of reality’s joints goes back to Aristotle. Reality is ordered, for Aristotle, by kinds that have essences, i.e., necessary features.Footnote 1 These necessary features are, for Aristotle, not just a matter of the conventional meaning of our kind terms; rather, they are something that we need to discover. Discovering kinds and their necessary features, developing definitions that incorporate them, and thus grasping the objective structure of reality, is a chief goal of science, as Aristotle understands it.Footnote 2
Aristotle’s view that natural kinds have essences—a view that is often called natural kind essentialism (NKE)—has been enormously influential, not only in metaphysics in the Aristotelian tradition, but also in the history of science: it was a dominating view in botany and zoology until far into the eighteenth century, and key figures in the development of modern biology such as Linnaeus were firmly committed to it (Mayr Reference Mayr1982: 165, 175–77).Footnote 3 Moreover, NKE still has many advocates: it has become popular again since Putnam’s and Kripke’s seminal work in the 1970s (Putnam Reference Putnam1975; Kripke Reference Kripke1980), and there has been a recent wave of work on ‘new essentialism’ that defends different variants of the position against critics (e.g., Ellis Reference Ellis2001, Reference Ellis2002; LaPorte Reference LaPorte2004; Devitt Reference Devitt2008).
But while NKE offers a popular way of spelling out the idea of reality as being structured by ‘joints’, the question of how this view can be argued for positively has received relatively little attention. In this article, I will argue that the resources for an interesting, positive argument in favour of NKE can be found in Hegel’s writings. Hegel arguably believes in the possibility of carving reality at its joints, and it has been proposed (although not always in these terms) that a version of NKE is crucial to Hegel’s metaphysics, and indeed to his entire philosophical project.Footnote 4 The argument that I will ascribe to Hegel in the following is therefore concerned with a central tenet of Hegel’s philosophy; moreover, it connects several key topics in Hegel’s thought, including the category of the ‘Concept’, the cognitive achievement of ‘comprehension’ (Begreifen), and the nature of subjectivity.Footnote 5 But Hegel’s defence of NKE is also of considerable philosophical interest independently of Hegel’s system, as it uses premises that can be made plausible outside of that system.
The paper falls into two main parts. My aim in the first part (sections I to III) is to show that prominent recent advocates of NKE have failed to provide compelling arguments for their view. After expanding on natural kinds and NKE in section I, sections II and III critically review some arguments for NKE in the recent literature: section II will discuss whether Kripke’s and Putnam’s work on the semantics of natural kind terms supports NKE, while section III will assess Brian Ellis’s argument for NKE. By arguing that those defences of NKE, influential though they are, fail to be compelling, I hope to provide reason to look elsewhere for inspiration for an argument in favour of NKE. The second part of the paper (sections IV to IX) tackles this desideratum by turning to Hegel. Section IV will briefly explain how Hegel can be read as subscribing to NKE, while sections V to IX will reconstruct and defend an argument for NKE that Hegel offers.Footnote 6
I. Natural kinds and Natural Kind Essentialism
Natural kinds (if there are any) are kinds that are ‘out there’ in reality—they are not merely imposed by our own cognitive activity, as ‘artificial kinds’ are. Three important, widely accepted conditions for the naturalness of a kind are the following. First, natural kinds support induction: the detection of shared stable features in a representative sample of kind members gives good (although defeasible) reason to believe that all other members of the kind have those features, too. Second, they are linked to natural laws: if two entities belong to the same natural kind, they are governed by the same specific laws, which hold for them in virtue of their kind membership. Third, natural kinds are normally thought to form hierarchies: an individual can belong to two natural kinds only if one of these kinds is a genus that contains the other as a species. This requirement guarantees that there is one way in which things are objectively divided up, and hence one privileged way of classifying them.
NKE proposes a specific, essentialist interpretation of such kinds. It can be formulated as a conjunction of two claims:
NKE (1) The objects of at least some sciences fall into natural kinds, which these sciences aim to capture in their classifications; and
(2) Natural kinds have essences: For each natural kind N, there is a set of properties P={P1, … } such that with metaphysical necessity, an object is an instance of N iff it possesses each property from P.Footnote 7
The above formulation of (1) leaves it deliberately open as to which sciences exactly have objects that fall into natural kinds. For example, I will count a view on which there are natural kinds in physics and chemistry, but not in biology, as a version of NKE. Moreover, by ‘objects’ of sciences I understand individuals (in a sufficiently broad sense) that are studied by those sciences, and over which their theories quantify—such as elementary particles in physics, samples of chemical substances, or biological organisms.
Initial motivation for NKE, thus understood, comes from the fact that its essentialist interpretation of natural kinds fits well with the other features of natural kinds that I have sketched before. The essences assumed by NKE could explain why natural kinds support induction (properties of single instances can be generalized if they are part of the essence of the kind), are linked to natural laws (if obedience to a particular natural law is one of the necessary conditions for kind-membership), and form hierarchies (if membership in higher-level kinds is part of the essence of a lower-level kind). Moreover, if they have essences, natural kinds are governed by metaphysical necessities. Again, this fits well with the idea of natural kinds as forming the ‘joints’ of reality: a classification that captures a metaphysically necessary structure in reality is intuitively seen as non-arbitrary, if compared to alternative, ‘artificial’ orderings that merely pick out some contingent relations of similarity between things.Footnote 8
II. Putnam and Kripke on Natural Kind Essentialism
As I have mentioned earlier, Putnam and Kripke have been pivotal in the revival of NKE over the last decades, and both adherents and critics of NKE often refer to Putnam’s and Kripke’s semantic framework as lending support to or even entailing NKE.Footnote 9 Putnam and Kripke argue for a view on which a given sample is part of the extension of a natural kind term if and only if it is of the same kind as a relevant standard: according to Kripke’s causal theory of reference, this standard consists in the sample with reference to which the term was initially introduced (Kripke Reference Kripke1980: 135); according to Putnam’s semantic externalism, it consists in the samples to which members of the speaker’s linguistic community normally apply the natural kind term (Putnam Reference Putnam1975: 141f., cf. 149–52). In turn, the relevant notion of ‘sameness’ of kind is understood by both authors in terms of sameness of microstructure (in Putnam’s case: if the samples in the actual environment of the speaker’s community have a common microstructure) (Kripke Reference Kripke1980: 138; Putnam Reference Putnam1975: 160). Assuming that the relevant standard that fixes the use of ‘water’ in our actual world has the microstructure H2O, it follows that with metaphysical necessity, water is H2O: across all possible worlds, those and only those samples which share the microstructure of H2O are samples of water. So having the microstructure H2O is (part of) the essence of the kind water.
It is therefore natural to understand Kripke’s and Putnam’s account of the semantics of natural kind terms as containing or entailing a form of NKEFootnote 10—more precisely, a form which assumes that a determinate hidden structure or microstructure is a necessary and sufficient condition for kind membership. By contrast, it is less clear how exactly Kripke and Putnam can be seen as arguing for this claim. Nathan Salmon (Reference Salmon1982) discusses a reconstruction of Putnam’s argument that Keith Donnellan has proposed in unpublished work, and that draws on Putnam’s account of how we can specify the meanings of natural kind terms through ostensive definitions (Putnam Reference Putnam1975: 141f.):
P1: It is necessarily the case that: something is a sample of water if and only if it is a sample of the same liquid that this↗ is a sample of.
P2: This↗ has the chemical structure H2O.
P3: Being a sample of the same substance as something consists in having the same hidden structure.
Therefore
C: It is necessarily the case that: something is a sample of water if and only if it has the hidden structure H2O. (Salmon Reference Salmon1982: 166, slightly modified)
As Salmon points out, this argument is valid as it stands, but it will not convince an anti-essentialist, because P3 already presupposes a version of NKE. Someone who denies NKE might give another definition for sameness of substance (or kind)—at least for the case of cross-world sameness with respect to nomologically different possible worlds—and would therefore reject both P3 and the conclusion.
On an alternative reading, Putnam’s and Kripke’s commitment to NKE is justified by the intuitions that are prompted by their famous thought experiments. When the visitor from Earth finds that what people on Twin Earth call ‘water’ actually has the microstructure XYZ, we expect him to conclude that the term has a different meaning, for XYZ is not water (Putnam Reference Putnam1975: 140, cf. 150). When imagining a possible world in which every area that contains gold in the actual world contains superficially identical fool’s gold, we would not say that gold has a different microstructure in the counterfactual world; rather, we would say that the substance in question in the counterfactual world is not gold (Kripke Reference Kripke1980: 124f., cf. 119). That people often find thought experiments like these convincing suggests that we intuitively assume a metaphysically necessary connection between kind membership and the possession of determinate microstructural features. And—we may add—unless one subscribes to a general scepticism about intuitions, there is no reason not to treat the essentialist intuitions in question as bona fide cases of an exercise of our capacities for rational insight, as they are standardly appealed to by metaphysicians.
Unlike Donnellan/Salmon’s reconstruction, this reading ascribes to Kripke and Putnam a reasoning in favour of NKE which is not simply question-begging. Nevertheless, the appeal to intuitions in favour of NKE can be challenged on the basis of Psychological Essentialism, a well-established hypothesis in cognitive science according to which our cognitive systems are biased towards essentialist interpretations of our environment (Leslie Reference Leslie2013).Footnote 11 As an ample body of behavioural data from children, including pre-schoolers, and apes suggests (Cacchione et al. Reference Cacchione2016), we and other animals have a disposition to think and act as if the stable properties of things had their causal root in essences. Our bias shows that we relate those essences to, or identify them with, the inside of things, and order them in hierarchies. Moreover, there is evidence suggesting that the essentialist bias affects the reasoning of adults, too, and even can lure them into patently irrational attitudes. For instance, the widespread suspicion that transplanted organs affect the personality of the patient who receives the organ has been explained as manifestation of the essentialist bias postulated by Psychological Essentialism (Meyer et al. Reference Meyer2013).
It is hard to avoid the conclusion that our intuitions in favour of microstructuralist NKE that are prompted by Putnam-/Kripke-style thought experiments are a manifestation of the bias postulated by Psychological Essentialism (with ‘microstructure’ being a more sophisticated way of conceiving of the ‘inside’ of things) (Leslie Reference Leslie2013). May we assume that this bias constitutes an ability for reasoning and belief-formation that is sufficiently reliable or truth-tracking to provide justification for beliefs that are based on this ability? There are two possible scenarios in which the bias can, in principle, have such a justification-providing status. First, it could be reliable by coincidence. In this scenario, NKE is true, and we have evolved to have an essentialist bias, but there is no causal connection between them. On some versions of externalism about justification at least, our essentialist bias could then produce justified beliefs, as long as we possess no other reliable process such that, if we exercised that process, we would not have the essentialist beliefs (Goldman Reference Goldman1979). Second, a bias that produces essentialist intuitions could be evolutionarily valuable because NKE is true: the truth of NKE could have an impact on biological and cultural evolution (just as truths about objects in our environment presumably have favoured the evolution of perceptual abilities that detect exactly such truths).
The probability that the first scenario—reliability by coincidence—obtains is very low. If someone assumed such a scenario to defend a positive justificatory role for our essentialist bias, this would be a mere ad hoc manoeuvre. The second scenario—reliability through an impact of NKE on evolution—is more likely. For one way to understand essences and natural kinds is as theoretical entities that are postulated by a science like chemistry or biology. As such, they ought to have explanatory power regarding observable phenomena, and hence an impact on those phenomena. On this view, it should be assumed that if NKE is true, its truth can affect evolution: if NKE is true, the environment in which organisms live is governed by natural kinds with essences, and in this case an essentialist bias can make it easier for an organism to cope with its environment.
However, there is still a problem here regarding the status of essentialist intuitions. For the question of whether the second scenario, as we have now described it, actually obtains or not, falls into the scope of science. In the case in which this scenario is actualized, we should expect that NKE provides the best explanatory framework for science. If science does not favour NKE (either because it speaks against it, or because it is neutral with regard to it), we should assume that the scenario in question does not obtain (in which case our essentialist intuitions could still be reliable, but only by coincidence).
We will come back later to whether science actually favours NKE or not. What matters for the moment is the following: if science does not favour NKE, someone who uses essentialist intuitions to argue for NKE has to assume that the first scenario obtains, and this would be merely an ad hoc assumption. If, instead, science does favour NKE, this will be a reason to hold NKE that is independent of the essentialist intuitions. While in this latter case, we may assume that our essentialist intuitions are produced by a reliable mechanism, we can do so only because science favours NKE in the first place. The intuitions do not themselves add any further reason in favour of NKE in this case.
We can conclude therefore that essentialist intuitions of the kind prompted by Putnam-/Kripke-style thought experiments should not, after all, be treated as good reasons in favour of NKE. Instead, I next turn to an argument for NKE that directly builds upon considerations about science.
III. Ellis on Natural Kind Essentialism
Brian Ellis, who has developed the currently best worked-out version of NKE (Ellis Reference Ellis2001, Reference Ellis2002), motivates his view in a way that avoids appeal to intuitions. Rather, Ellis has proposed that NKE should be seen as part of the best way to spell out scientific realism—that is, broadly speaking, the view that the entities which our best scientific theories postulate actually exist. More precisely, Ellis argues that we should accept a version of the miracle argument for scientific realism which is restricted to entities that are ‘essentially involved in causal process explanations of [broadly] physical phenomena’ (Ellis Reference Ellis2001: 157): the actual existence of the entities that science refers to in our best causal theories is itself the best explanation for the success of those theories. On the resulting version of scientific realism:
if a physical causal mechanism is postulated to explain something, and it provides a good explanation—one that is better than any other currently available—then it is reasonable to believe in the reality of the proposed mechanism. Moreover, if the proposed mechanism assumes the existence of entities having various properties or structures, and these properties or structures are appealed to in the explanation that is given, we should believe that these properties or structures exist too. A causal process realist must therefore be a realist about certain kinds of properties and relations, as well as about the things that are supposed to possess them. (Ellis Reference Ellis2001: 157)
Thus, for Ellis, NKE is a metaphysical theory that accounts for those aspects of reality that make possible successful explanations of phenomena in terms of causal processes. More specifically, Ellis thinks that scientific explanation itself aims de facto at uncovering essences: ‘Not only are most scientific explanations concerned with the underlying causes of phenomena, and are thus realist in character’, he writes, ‘many are … also essentialist. That is, they are concerned with the essential natures of the kinds of causal processes they seek to explain’ (Ellis Reference Ellis2001: 160). Ellis claims that a great deal of inquiry in physics and chemistry aims to uncover the nature of such processes as alpha decay, or the reaction by which metallic copper is turned into cupric sulphate. He thinks that such processes themselves are grouped into natural process kinds with essences, and that science aims at discovering such essences—at understanding what makes those processes the processes they are. In turn, explanations of observed phenomena in terms of process kinds with essences depend on assumptions about the essences of natural kinds of objects: ‘If, for example, you ask what makes the cupric sulphate chemical reaction one of the kind it is, you will learn that it is the fact of its being a reaction involving just these kinds of molecules in just this kind of way’ (Ellis Reference Ellis2001: 161). It is the essences of kinds of objects that settle how instances of these kinds behave in given circumstances; for Ellis, the laws of nature are based on natural kinds.
The form of Ellis’s argument for NKE can be summarized as follows:
P1 Causal-process explanations in science S typically use as explanantia essences (consisting of individually necessary and jointly sufficient conditions) of kinds of processes and objects (and properties).
P2 The entities postulated by our best causal theories exist. (from the miracle argument for causal process realism)
C NKE is true for the objects of S;
and Ellis endorses this argument for physics and chemistry.
Ellis’s argument is flawed by a very similar shortcoming to Putnam’s argument on the Donnellan/Salmon reading: P1 presupposes an essentialist interpretation of scientific explanation (and even a very specific type of essentialist interpretation) which an opponent of NKE does not have to accept. For there are many alternatives to a view of scientific explanation as uncovering essences that Ellis does not rule out and that his opponent can resort to. Prominent non-essentialist interpretations of scientific explanation include views in which scientific explanation subsumes observed phenomena under natural laws that are not based on essences (Hempel Reference Hempel1965; Armstrong Reference Armstrong1983: 40f.; Maudlin Reference Maudlin2007);Footnote 12 views in which scientific explanation uncovers causal mechanisms that are not based on essences (Salmon Reference Salmon1984); and views in which scientific explanation unifies theories (Kitcher Reference Kitcher1989).
Now it might be possible to argue that while different ways of interpreting science are in principle possible, an essentialist interpretation has the best fit with current theories and the scientific practice behind them—and this may well be how Ellis himself would motivate the above premise P1 (applied to physics and chemistry). So does science support NKE in this sense? Critics have cited a number of examples from science to show that, far from supporting NKE, science actually speaks against it. For one thing, it has been pointed out that the stock example of NKE since Putnam and Kripke is mistaken:Footnote 13 being H2O is neither a necessary nor a sufficient condition for membership in the kind water, since the liquid and ice phases of water have a much more complex and variable composition.Footnote 14 Admittedly, water is a particularly complicated chemical compound (Hendry Reference Hendry2006: 873); but other problem cases for NKE are more pervasive. Thus, it has been argued that many chemical (and biological) kinds partially overlap each other: they are ‘cross-cutting kinds’. Such kinds do not fulfil the requirement of hierarchical order, which means that they are not natural kinds, and a fortiori not natural kinds with essences. For example, most, but not all, enzymes are proteins; and some, but not all, proteins are enzymes (Tobin Reference Tobin2010).Footnote 15 And the isotopes of the chemical elements can be classified not only according to their atomic number (such that the different isotopes of an element are grouped together), but also according to their mass number, the number of protons and neutrons (such that isotopes of different elements are grouped together) (Khalidi Reference Khalidi2013: 70). This mass number-based ordering has predictive power with regard to radioactive decay (Khalidi Reference Khalidi2013: 70); it therefore would seem to qualify as a non-artificial system of kinds, which cuts across the system of the chemical elements.
In response to the problem of such cross-cutting kinds, an advocate of NKE could abandon the requirement that natural kinds should be ordered hierarchically (cf. Tobin Reference Tobin2010). But this comes at a high cost, as it threatens the idea that there is one objective order of reality that we can capture in our classifications. Cross-cutting kinds support more than one scheme of classification. Which scheme we adopt will presumably depend on our interests and the context of our inquiry (e.g., whether we are concerned with the nuclear or chemical properties of atoms), and this seems to undermine the idea that reality can be carved at its joints.
There are two other ways in which the advocate of NKE can react to the problem of cross-cutting kinds. First, an advocate of NKE can restrict NKE to kinds that are less problematic. In particular, the elementary particles of physics are often seen as unproblematically amenable to an interpretation in terms of NKE. Given the fundamental scientific importance of the kinds of elementary particles (quarks, leptons, and bosons, with their respective subkinds), a form of NKE that preserves the status of a natural kind with an essence for these kinds would still be a substantive metaphysical position.
The problem with this position is that even if it is granted that kinds of elementary particles are bona fide cases of natural kinds, it becomes unclear how an essentialist interpretation of them can be motivated if they are the only true natural kinds. It was a key strength of NKE, as traditionally understood, that it promised a unified account of scientific explanation, laws of nature and taxonomy for whole sciences like physics and chemistry (or even biology). If the kinds of elementary particles are the only real natural kinds, the explanations, laws, and kinds in chemistry and in branches of physics like thermodynamics and optics must be accounted for without reference to natural kinds and essences.Footnote 16 And whatever metaphysical account is adopted in this regard, it is reasonable to expect that this account can also be extended to elementary physics. This puts advocates of the restricted version of NKE under pressure: they need to provide an independent motivation for why they defend NKE in the case of elementary particles, instead of embracing the unified non-essentialist account.
The other remaining option for the advocate of NKE is to accommodate cross-cutting cases by assuming that among the competing divisions of things (e.g., the division of atoms according to atomic number and according to mass number), only one really captures the natural kind-structure of reality. The other divisions can still be non-artificial and do important scientific work, but they are not divisions into natural kinds. Whether this option is viable or not depends, too, on whether NKE can be motivated independently. Without such motivation, this option will be a mere ad hoc move.
This leaves the advocate of NKE in an uncomfortable situation. As our discussion suggests, science does not itself speak in favour of NKE: science can be accommodated by NKE (through either of the two moves that I discusse in the last three paragraphs), but the motivation for NKE has to come from elsewhere. In other cases of particular metaphysical interpretations of science, such independent motivation typically comes from metaphysical intuitions. But as it turns out, essentialist intuitions should be taken at face value only if there is independent support for NKE from science—which there does not seem to be.
IV. Hegel on Natural Kind Essentialism
In this and the following sections, I hope to show that Hegel offers the resources for an alternative argument for NKE that is not question-begging and does not appeal to essentialist intuitions.Footnote 17 More precisely, Hegel can be seen as defending an essentialist interpretation of science in general—and hence, as offering an argument in support of Ellis’s premise P1.Footnote 18 When combined with the further assumption of scientific realism, this argument is apt to support the conclusion that NKE is true. Before beginning to reconstruct Hegel’s argument in the next section, I will in this section say a bit about his understanding of NKE. For the present purpose, I will restrict myself to a very brief sketch of the outlines of such a reading (for more detail, see Knappik Reference Knappik2016a).
In the third part of the Science of Logic, Hegel identifies the category of ‘the Concept’ as a central logical and ontological principle, claiming that all reality is a realization of the structure of the Concept—a structure that he defines as a unity of universality, particularity, and individuality.Footnote 19 This can be seen as the basis for a metaphysical theory on which finite reality is articulated by individuals that instantiate, in virtue of their particular features, universal objective kinds; Hegel’s preferred term for such kinds is ‘concepts’ (e.g., WL: 5:26).Footnote 20 So while Hegel describes his philosophy as a form of idealism, his metaphysics is actually congenial to the claims of scientific realists about the status of theoretical entities: as Robert Stern puts it in a related context, this ‘will only seem oxymoronic if by “idealism” one thinks only of Berkeley and not of Plato’ (Stern Reference Stern2016: 2; see also Stern Reference Stern2009: 59).
The way in which Hegel understands objective ‘concepts’, or natural kinds with essences, differs significantly from the views of many essentialists today. Unlike authors like Ellis who restrict natural kinds to physics and chemistry, Hegel also sees biological kinds as natural kinds with essences, and even treats them as his paradigm of natural kinds (at least in nature).Footnote 21 In the case of such biological kinds, the properties that make up the essence of a natural kind are not categorical properties for Hegel (such as microstructural properties, e.g., DNA, on some forms of contemporary NKE in biology: e.g., Devitt Reference Devitt2008), but dispositions or tendencies of the organism to live up to a kind-dependent normative standard. This standard, in its turn, is presumably defined (at least partly) in terms of categorical properties. But the extent to which an individual organism actually possesses those categorical properties, and the way in which it concretely realizes them, depend on contingent factors about the individual organism and its environment (see WL: 6:517f.). This enables Hegel to accommodate the fact that there is a large variety of actual properties within a given biological kind—a fact that speaks against microstructural NKE (cf. Leslie Reference Leslie2013). (It bears emphasis that a teleological version of NKE for biology is still a viable option today. Thus, teleological accounts of biological kind essences have been recently proposed by neo-Aristotelians, who re-interpret traditional teleological accounts of organisms in evolutionary terms: see, e.g., Walsh Reference Walsh2006.)
While Hegel restricts the teleological character of natural kind essences to organic nature, he understands the essences of natural kinds in inorganic nature, too, as constituted by dispositional properties that fix how, and according to which natural laws, instances of those kinds behave under given circumstances. For example, when discussing chemical kinds in his Philosophy of Nature, Hegel claims that the ‘nature’ of a chemical kind ‘depends on its position with regard to the various processes in which it is the producing, determining factor [i.e., the reagent], or the product’ (E: §329A, 9:300).Footnote 22
V. The structure of Hegel’s argument
Hegel’s argument in favour of NKE connects NKE to considerations about the nature of understanding and, somewhat surprisingly, considerations of conceptual structure and subjectivity. It is clearly a very ambitious argument, and I claim neither that it is the only possible alternative to the arguments for NKE discussed above, nor that it commands assent. But still, I think that it is an argument worth exploring: it is free from the shortcomings of Putnam’s, Kripke’s and Ellis’s arguments; and it provides an unusual perspective on the metaphysics of explanation by addressing the very nature of understanding. Moreover, all steps in the argument can be made plausible also independently of Hegel’s system (or so I hope to show by briefly commenting on the plausibility of those steps, as I go along).
The basic structure of Hegel’s argument goes as follows:
(1) Hegel first gives an account of what it means to give an explanation, or understand something, at all. This account has the surprising consequence that constraints for an adequate metaphysics of explanation can be derived from our structure as thinking subjects.
(2) Hegel then gives an account of the structure of thinking subjects that, in terms of constraints for the metaphysics of explanation, supports a form of NKE.
As we will see, this result does not directly count in favour of NKE; but it does speak for an essentialist interpretation of science, and hence for a version of Ellis’s premise P1. It can therefore be combined with scientific realism (Ellis’s premise P2) to yield a sound argument for NKE. I address step (1) of Hegel’s argument in the next section, step (2) in sections VII and VIII, and I summarize the result in section IX.
VI. Hegel on the nature of understanding
The first step of Hegel’s argument consists in a reflection about what it means to give an explanation. For Hegel, it is natural to raise this question in the context of NKE because his defence of NKE is part of a project that enquires into the metaphysical preconditions of successful explanations (cf. Knappik Reference Knappik2016a). But, unlike other philosophers who have thought about the nature of explanation, Hegel attempts to answer this question by accounting for what it means to understand or (in his preferred term) comprehend (begreifen) something—where ‘understanding’ is to be taken not in the sense of semantic understanding, but of the understanding that is conveyed by an explanation, or an answer to a why-question. The following passage in ‘Of the Concept in General’, the introduction to the Logic of the Concept, contains a key statement about the nature of understanding or comprehension:Footnote 23
The comprehension of an object consists indeed in nothing else than in this: that the ‘I’ makes it its own, pervades it, and brings it into its own form […]. The object in intuition or also in pictorial representation is still something external, alien. Through comprehension, […] the ‘I’ pervades it as thinking. But only as it [i.e. the object] is in thinking, it is in and for itself; as it is in intuition or pictorial representation, it is appearance. […] The object has therefore this objectivity in the Concept, and this Concept is the unity of the self-consciousness into which it [i.e. the object] has been received; its [i.e. the object’s] objectivity or the Concept is therefore itself nothing else than the nature of self-consciousness, has no other moments or determinations than the ‘I’ itself. (WL: 6:255)
A first point that should be made about this dense passage is the following. To elucidate the nature of comprehension or understanding, Hegel uses metaphorical language—he talks about the I ‘bringing’ the object ‘into its own form’, and ‘pervading’ the object with its form. There are possible interpretations of such formulations which would make Hegel’s position very implausible, and hence should be avoided on a charitable reading. Thus, we should not understand Hegel here as describing an activity in which we literally change the explanandum itself—e.g., by physically manipulating it. Rather, he has in mind a characterization of a form of cognitive activity. What is changed in such activity is, strictly speaking, not the explanandum itself, but our representations of it. Furthermore, even if one grants this, one might read Hegel here as holding that explanations cannot go beyond the explanandum by introducing, as explanantia, principles that are distinct from the explanandum. This, too, would be an uncharitable reading; in other passages, Hegel clearly holds that comprehension does make reference to additional explanatory principles (e.g., WL: 5:176). Hence, we should assume that in the above passage, Hegel sees the cognitive activity of understanding as one that changes the form of (our representation of) the explanandum precisely by connecting our initial information about the explanandum not only to additional information about the explanandum itself, but also to information about other objects or facts, to hypotheses about explanatory principles, etc. (see E: §21A, 8:77f.). The result of such a process of understanding—or, in Hegel’s metaphorical language, of such a change in the form of the explanandum—is an explanatory account that substantially goes beyond the initial description of the explanandum, but still remains related to the explanandum because it offers an account precisely of this explanandum.—Bearing these clarifications in mind, I will myself use, for the ease of presentation, metaphors like the ones Hegel employs in the above passage, and talk about understanding as an assimilation, transformation etc., of the explanandum.
We can interpret now the claims that Hegel puts forward in the above passage, as well as the connections between them.Footnote 24 To begin with, Hegel can be seen as making an observation about processes of understanding that seems phenomenologically plausible: a situation in which we do not understand something is a situation in which the object or subject-matter that we are dealing with is alien to us, something we find strange and puzzling. This suggests that in a contrasting situation of understanding, the sense of alienness is replaced by a sense of coping with the subject-matter, or of finding our way around in it. Hence, we experience the process of understanding as a process in which what initially is alien to us becomes our own.
Hegel goes beyond this intuitively plausible point by interpreting it as follows: he sees the process of understanding specifically as a process in which the explanandum is assimilated to our own structure—that is, our structure qua thinking subject, or ‘I’. Hegel seems to presuppose two things here: first, that the right way to spell out the idea of understanding as a process in which something alien becomes our own is in terms of a transformation that makes (our representation of) the explanandum similar in some substantive respect to the understanding subject; and second, that the relevant respect is one of structure or form, rather than a material respect. Hegel does not argue for either of these points, and since the topic of the nature of understanding is so underexplored in the philosophical literature, it is rather difficult to evaluate his proposal. Still, this proposal can be made plausible by comparing it to possible alternatives. One such alternative would be to interpret the process in question as a process in which the subject gains ownership over something that she initially does not possess. However, this condition seems too weak for understanding. Possession or ownership with regard to a represented content does not require understanding in the relevant sense: we own the content simply by representing it in some way (it is my perception, thought etc.), even if we do not know why that content is the way it is. Similarly, it would not suffice to interpret the process in question as one in which a content is integrated into our belief systems, for we can integrate a content into our belief system without having an available explanation of it. Finally, one might interpret the process as one in which a subject gains practical mastery over something—as one makes a language or technique ‘one’s own’ by learning to use it. But this reading would both include and exclude too much: for example, many people are able to drive cars without understanding how they work, while other persons may have a perfect theoretical understanding of the internal mechanisms of cars without being able to use them. Compared to possible alternatives like these, Hegel’s proposal—that the idea of understanding as process in which something alien becomes one’s own—should be spelt out in terms of a process of assimilation is indeed much more promising.
Next, if understanding is a process in which the explanandum is rendered similar to the subject, it seems that the respect in which it is rendered similar should not be identified with some material feature, such as determinate mental states, physical properties, etc. For otherwise the range of possible explananda would be very much restricted—namely, to other human beings and their properties, or even to human beings that look, think and act like the understanding subject. If such material properties are excluded as relevant standards of similarity, only the form of the subject remains as a possible candidate for such a standard.
Hence, there are good reasons for Hegel to interpret understanding as a process in which the explanandum is assimilated to the structure of the subject. Of course, this is a very general characterization, and we have to find out what the relevant structure of the subject concretely consists in. But before we turn to that question, it is important to notice that in addition to assimilation, Hegel formulates a further condition for understanding, which relates to objectivity. Not only does the assimilation of given contents that is required for understanding have to match the structure of the subject; it must also do justice to the actual facts about the explanandum (and related parts of reality): rather than distorting reality, it needs to uncover its actual structure. For one thing, explanations are generally seen as veridical: the statement ‘p because q’ (with ‘because q’ expressing an explanation for p) commits the speaker to the truth of q. In addition, there is also a widespread view, adopted here by Hegel, on which an explanatory account of an object or phenomenon gives us access to what that object or phenomenon really is (as Hegel puts it in the above passage, the way it is ‘in and for itself’), whereas more immediate forms of presentation of the same object or phenomenon (intuition, pictorial representation) give us only some superficial knowledge about it.Footnote 25 The resulting two conditions of understanding—assimilation and objectivity—can be simultaneously met only if the process of comprehension makes accessible an objective structure in reality that involves the explanandum, and that is isomorphic with the subject’s own structure.
It follows from the objectivity requirement for understanding that Hegel’s account of understanding has important metaphysical consequences: if the account is correct, understanding is possible only if the actual structure of explananda (and other parts of reality related to it) is isomorphic with the subject’s own structure. So, in this view, one can derive from an account of the structure of thinking subjects constraints on the metaphysical structure that reality must display if understanding of it is possible. This derivation is the second step in Hegel’s argument.
VII. Hegel on the structure of thinking subjects
To unpack this second step, we must ask what account of the structure of the comprehending subject Hegel builds on in this context. In the passage on comprehension that I have been quoting from, Hegel himself gives an extremely dense answer:
The comprehension of an object consists indeed in nothing else than in this: that the ‘I’ makes it its own, pervades it, and brings it into its own form, i.e. into the universality which is immediately determinateness, or determinateness which is immediately universality. (WL: 6:255)
Hence, Hegel relies, for his account of comprehension, on a rather cryptic view of the relevant ‘form’ of the subject (or ‘I’) as ‘universality which is immediately determinateness’ or ‘determinateness which is immediately universality’. In the following, I will offer an interpretation of what Hegel’s technical characterizations of the structure of the comprehending subject amount to.
On this interpretation, there are two important points that are conveyed by these characterizations. First, as the context of the passage in question shows,Footnote 26 the formulations ‘universality which is immediately determinateness’ and ‘determinateness which is immediately universality’ allude to Hegel’s category of ‘the Concept’. (As we saw, that category is defined as unity of universality, particularity and individuality, and ‘determinateness’ in the present passage can be seen as referring to particularity and individuality together.) There is more than one reason why Hegel interprets the structure of the subject in terms of his category of the Concept. A first reason is that he describes, in the same context of the Science of Logic, self-conscious subjects as combining a dimension of universality—the subject is a numerically identical, common bearer of different mental states, and it is conscious of this—with a dimension of individuality—the subject is numerically distinct from other subjects (WL: 6:253). (Elsewhere, Hegel locates the third element of the Concept, particularity, in the determinate mental states of the subject; GPR: §7 R, 7:55.) But I think that, with regard to Hegel’s argument that I am reconstructing, another way in which he relates subjectivity to the structure of the Concept is of greater interest: the structure of the Concept is for Hegel also the basic form of conceptual structure, and conceptual structure is for him the form of thought—of the mental activity that is responsible, according to the passage from WL: 6:255 cited earlier, for comprehension.
We can understand why Hegel analyses the conceptual structure of thought in terms of the structure of the Concept as follows: in agreement with the Aristotelian tradition in logic, Hegel believes that concepts are hierarchically structured: more specific concepts are subordinate to more general concepts, while individuals fall under both.Footnote 27 Hegel uses the notions of universality, particularity and individuality—i.e., the elements of the Concept—to analyse such hierarchies of concepts. This offers a plausible account of the conceptual structure of thought insofar as the hierarchical order of concepts provides a model for the inferential structure of reasoning: if an individual falls under a more specific concept (corresponding to ‘particularity’) and this concept is itself subordinate to a more general concept (corresponding to ‘universality’), it can be inferred that the individual also falls under the more general concept.Footnote 28
Hence, a first constraint on the metaphysics of explanation is provided by our structure as thinking subjects. Thought is conceptually structured, which means for Hegel that it is articulated by the structure of the Concept; understanding requires an assimilation of the explanandum to such conceptual structure; hence, understanding is possible only if the explanandum itself (together with related parts of reality) is structured in a way that is isomorphic to conceptual structure, i.e., to the structure of the Concept.
In recent work by neo-Hegelian philosophers like Brandom and McDowell, the idea that our cognitive processes can only be successful if reality itself possesses something like a conceptual, inference-supporting structure has played an important role.Footnote 29 I think that these authors are right to attribute this point to Hegel, and that the same point applies to Hegel’s account of understanding or comprehension (indeed, proper cognitive success consists for Hegel in comprehension; cf. Knappik Reference Knappik2016a). However, taken on its own, this point is compatible with quite a broad range of metaphysical options. For instance, a quasi-conceptual structure of reality might be realized by real universals that form hierarchies without having essences, or by a nominalist picture that admits of a hierarchy of classes of individuals in different possible worlds. So, taken in isolation, the claim that comprehension requires us to identify a quasi-conceptual structure in reality, although important for Hegel’s position, does not impose sufficient constraints on the choice of metaphysical theories.
A second point that is needed to provide additional constraint, and that has not yet been appreciated by neo-Hegelians, is due to the role that subjectivity plays for Hegel in this context: for him, the structure to which we assimilate explananda in understanding them is partly defined by the structure of our thought, but partly also by our structure as the subjects of such thought. Hegel is not very explicit about this point, but he gives at least a hint when he ascribes to Kant, in the same context in the Science of Logic that I have been quoting from, a path-breaking insight into the relation between the subject and the concepts it uses, or ‘the Understanding’ (meant here in the sense of the ability for concept-use in general, not of understanding-why or comprehension):
When one speaks ordinarily of the Understanding that I have, one understands by this a faculty or a property that stands in the same relation to the I as the property of a thing to the thing itself—an indeterminate substrate that is not the true ground and that which determines its properties. According to this notion, I have concepts and the Concept, as I also have a coat, skin colour, and other external properties.—Kant went beyond this external relation of the Understanding as the faculty of concepts and of the Concept itself to the I. It is one of the deepest and truest insights that there are in the Critique of Reason that the unity which constitutes the essence of the Concept is recognized as the original-synthetic unity of apperception, as unity of the ‘I think’ or of self-consciousness. (WL: 6:254)
As Hegel presents Kant here, he abandons a view on which the relation between a subject and its concepts is a relation between a substrate and its properties. Indeed, Kant’s notion of a synthetic unity of apperception can be read as referring to a unity that we bring about, through synthetic activity, among the given manifold content of our representations, and that is a necessary condition of conscious representation (and for the ability of self-ascription that conscious representation famously requires for Kant).Footnote 30 Moreover, Kant makes clear that the agent of such synthetic activity is the subject: according to him, the relation of a representation to the identity of the subject ‘does not obtain because I accompany each representation with consciousness, but because I add one to the other and am conscious of their synthesis’ (KrV: B133, my emphasis).
So for Kant, there is a sense in which all our (conscious) representations issue from a synthetic activity whose agent is the subject. Of course, this agency of the subject, like other forms of agency, does not exclude responsiveness to externally given conditions and contents. Nevertheless, the resulting representations are, for Kant, to an important extent, rooted in ourselves. In this sense, our synthetic activity is a form of self-determination. As Hegel paraphrases the point: instead of having them as simple properties or possessions, we relate to our concepts (and our representations in general, we may add) as their ‘true ground’ or ‘that which determines’ them (WL: 6:255).
It is important to realize that Hegel’s points about the connection between subjectivity, self-determining agency and understanding can also be made plausible independently of Hegel’s technical terminology and analyses. For, as far as the argument for NKE that I am reconstructing is concerned, it is sufficient to assume that subjects who engage in processes of understanding are necessarily self-determining agents. This weaker claim suffices to show that agency is a part of the structure of subjectivity, and hence that understanding as assimilation to one’s own form has to detect in reality a form of, or analogy with, agency. And the weaker claim is highly plausible: even if we could imagine beings for whom agency is much less important than it is in our lives, it is hard to see how such beings could engage in inquiries that aim at understanding without engaging at least in a form of mental agency that is directed to a goal and sensitive to reasons; and this would already count as a form of self-determining agency in many accounts.
We can sum up our results by saying that, for Hegel, the structure to which we need to assimilate explananda in order to understand them is the structure of thinkers—subjects who are self-determining agents of conceptually structured thought. This yields two constraints on the metaphysics of explanation: understanding is possible only if the relevant parts of reality display, first, a quasi-conceptual structure, and second, an analogon of self-determining agency.
Before proceeding, we need to address a possible objection: even if it is granted that assimilation to conceptual structure and to the structure of subjectivity is what it means for us to understand something, this is merely a contingent fact about our cognitive set-up. If we are not allowed to take our essentialist bias as a basis for the selection of a preferred interpretation of science, we should not base such a selection on facts about what it means for us to understand something either.
By way of response, it should be emphasized that nothing in the Hegelian account of understanding entails that it is limited to the actual cognitive set-up of human beings. Rather, the account aims to uncover what understanding is as such—hence, it claims validity for all possible subjects who engage in explanatory projects. This claim to general validity is supported by the following consideration: the form of understanding that Hegel’s account deals with is understanding-why. A mind (be it human or not) that possesses understanding in this sense needs to know the answers to relevant why-questions. Both why-questions and the answers to them have propositional contents, and grasp of propositional contents requires the possession and application of concepts. Moreover, we have already noticed that any mind that attempts to gain understanding in the relevant sense will need to engage in some form of (at least mental) agency. Hence, any mind that works towards the understanding of an explanandum in the relevant sense needs to share with us at least those features (namely, agency and conceptually structured thought) that define for Hegel the structure to which explananda need to be assimilated.
VIII. Natural Kind Essentialism and the possibility of understanding
We now need to turn to the question why, for Hegel, NKE is the best candidate to fulfil these two conditions on an adequate metaphysics of explanation.
(a) On Hegel’s account, natural kinds with essences display a structure that is isomorphic to conceptual structure in at least two ways. First, they are ordered in hierarchies that support inferences (e.g., from something being a tiger, to it being a mammal).Footnote 31 Second, Hegel describes the explanatory relations between a natural kind and its instance in terms of an analogue of conceptual structure. Thus, he writes: ‘All things are a [sc. universal] genus … in an individual actuality with particular features’ (E: §179, 8:330). The relation between kind, individual instance and particular features is presumably to be understood in the sense that membership in the kind entails (directly or indirectly) the particular features of the object. This consequence relation between kind membership and determinate properties, too, supports inferences (e.g., from an individual being a tiger, to it striving to realize the properties that make up the kind-specific standard of tigerhood).
(b) While we saw that alternative metaphysical frameworks could satisfy the need for a quasi-conceptual structure of reality, Hegel’s form of NKE is also able to fulfil the second constraint: the need for an analogon to the agency that characterizes subjectivity. To see why, consider first Hegel’s version of NKE for biology—a version on which, as we have seen, biological kinds have teleological essences. In the case of such kinds, the essence of the kind fixes a specific standard that members of the kind necessarily strive to realize. Hence, the essence of the kind grounds an activity that brings about the particular properties of the organism. There are two reasons why this activity can be seen as an analogon of self-determining agency. First, like the self-determination of subjects, it is rooted in a dimension of universality or relative indeterminacy—namely, the dimension of the organism’s kind—and brings about more concrete properties of the organism. Second, and more importantly, the organism’s activity is analogous to self-determining agency because it has its root in the organism’s own nature (as member of its kind), rather than being imposed by an external source (even though, like the self-determination of subjects, it is an activity that is sensitive to external influences and conditions). As we saw, physical and chemical kinds are not teleologically structured for Hegel. Nevertheless, physical and chemical kinds can be seen as approximations to the structure of biological kinds, and as a consequence, to the structure of subjectivity. Even though no proper activity takes place at these levels of reality, we can find here the structure of a universal nature—the essence of the kind—that grounds a form of behaviour of the individual object (the manifestations of the dispositional properties which it possesses qua member of its kind), such that the particular categorical features of the object result from that behaviour (under the given circumstances). Again, this is a behaviour that is rooted in the universal dimension of the individual, its kind, and brings about more determinate properties; and it is based on the individual’s own nature, rather than being imposed from the outside.Footnote 32
By contrast, rival accounts of the metaphysics of explanation do not simultaneously fulfil the two constraints that have resulted from Hegel’s account of understanding. Laws, for example, satisfy the need for quasi-conceptual structure, as they support inferences (e.g., predictions), but not the need for self-determining activity. Causal relations, by contrast, can be seen as displaying (at least on some accounts) a form of activity. But either this activity is governed by causal laws. In that case, the causal relation displays quasi-conceptual structure, too (because of its participation in a law); but since the causal activity is ‘imposed’ on the causal relata by something outside of them, namely the law, it is not analogous to self-determination. Or the causal relation is anomic, as in singularist causation. In this case, the causal activity would be analogous to self-determination, but it would not display quasi-conceptual structure, as it would not support inferences.Footnote 33
IX. The result of Hegel’s argument
If successful, Hegel’s argument, as we have reconstructed it in the previous sections, shows that if a genuine understanding of some part of reality is possible, that part of reality is ordered in natural kinds with essences—i.e., NKE is true. A full argument for NKE results if this conditional conclusion is combined with Ellis’s version of the miracle argument. For the conditional conclusion of Hegel’s argument suffices to motivate the premise P1 that we found unwarranted in Ellis’s argument: the view that scientific explanation is ultimately essentialist. (While Ellis had restricted P1 and the conclusion of his argument to physics and chemistry, Hegel’s argument supports these claims for biology also.) I had objected to Ellis’s argument that this premise presupposes an essentialist interpretation of science. What Hegel’s argument, if it is successful, establishes, is precisely that we should, as long as possible, adopt such an essentialist interpretation of science. It may ultimately turn out that such an interpretation is not feasible, and this would be sufficient reason to reject NKE; from Hegel’s viewpoint, this would mean that it is, after all, not really possible to gain proper comprehension of reality. But as long as an essentialist interpretation of science is at least possible, Hegel’s argument gives us reason to prefer it over its rivals. In this case, we will postulate the existence of natural kinds with essences as theoretical entities. If we combine this with Ellis’s premise P2 (scientific realism about the entities postulated by causal theories, based on a miracle argument), a complete argument for NKE results.Footnote 34
