1 Introduction

Uniformism is the view that one and the same epistemology should apply for all modal knowledge: there is just one basic route to modal knowledge about things as intuitively diverse as persons, knowledge, or gold. In contrast, a non-uniformist takes there to be more than one route to it. One of the main arguments for non-uniformism is that, given the “metaphysical heterogeneity” of the things that modal knowledge is about, and in particular given the fundamental difference between concrete and abstract entities, we should be “suspicious about any uniform attempt” to explain how such knowledge is possible “that is not sensitive to this heterogeneity” (Roca-Royes, 2007: 126; cf. also Sjölin Wirling, 2020). The “assumption” that we need different epistemologies for concrete and for abstract entities seems to be, on the contrary, very plausible (Roca-Royes, 2018: 245).

Sònia Roca-Royes argues in this vein that her induction-based modal empiricism– according to which “we know about some concrete entities’ unrealized possibilities by extrapolation from (largely a posteriori) knowledge about some other, similar entities’ realized possibilities”– “is not fit to explain our de re modal knowledge of abstract entities”, which requires another epistemology (ibid.). This is mainy due to the fact that abstracta, unlike concrete entites (paradigmatically: apples, tables, rocks, etc.), lack spatiotemporal location and causal powers: because they are causally inert or inactive, it is difficult to understand how we could acquire empirically-based knowledge of them, and, a fortiori, how we could acquire knowledge of their essences and of the modal facts that are true of these entities. This naturally leads to the idea that there must be two routes to knowledge of essences: an a priori route and a posteriori route.Footnote 1 As Tuomas Tahko argues, “[o]n the face of it, it is difficult to see how a posteriori essentialism could account for the essences of abstract objects […], so unless we can explain our epistemic access to the essences of concrete objects in terms of a priori essentialism as well,Footnote 2 then a hybrid view [i.e., combining a priori and a posteriori elements] looks unavoidable” (Tahko, 2018: 103; see also Hale, 2013: 254).

According to the neo-Aristotelian turn championed by Kit Fine, Bob Hale and E. J. Lowe, modal knowledge should be accounted for in terms of knowledge of essences,Footnote 3 where essences are conceived as that in virtue of which x is what it is, and more precisely, the kind of thing it is.Footnote 4 What I would like to do in this paper is show that, whether or not the neo-Aristotelian turn is correct, uniformism about knowledge of essences is untenable– just as uniformism about modal knowledge is untenable according to (Roca-Royes, 2007, 2018), (Tahko, 2018), or (Sjölin Wirling, 2020). I shall do this by contrasting two lists of things and showing that the essences of those on the second list are not empirically discoverable (§ 2.2), unlike the essences of those on the first list (§ 2.1). I shall then advance a metaphysical ground for this epistemic difference: the difference between having, or not having, an empirically discoverable essence turns out to exactly mirror the metaphysical difference between being, or not being, uniquely realisable. This makes the uniquely realisable–non-uniquely realisable distinction a better metaphysical candidate for grounding the epistemic difference than the concrete–abstract distinction since, as I shall argue, at least some of the things that make up the second list clearly do not belong to the category of abstracta (§ 3). I shall conclude by suggesting that it is the task of philosophy, not empirical science, to identify the essences of things on the second list (§ 4).

2 A new way of arguing against uniformism

Let us start with a basic and seemingly trivial observation: take the case of water, gold, and lemons, on the one hand, and of belief, knowledge, religion, personhood, causation, free action, and democracy, on the other. To the question “Are the things that, more or less intuitively, we take to be beliefs, knowledge, religions, persons, causes, free actions, and democracies really so?”, we certainly would not answer: “This depends on what they are made of”. That would be, on the contrary, a natural answer to the question “Are the things that, by looking at their observational properties, we take to be water, gold, and lemons really so?”. And it would be natural for us to add: “And this, only empirical investigations will tell us”. The same would go for the question “Are these two things in the basket we take to be two lemons, these two ingots on the desk we take to be two gold ingots, and these two glasses on the table we take to be two glasses of water, really identical in nature?”. It would be natural for us to answer to this question: “This, only empirical investigations into what they are made of will tell us”. But this is not how we would answer the question “Are these two beliefs we take to be cases of knowledge, these two practices we take to be cases of religion, and these two beings we take to be persons, really identical in nature?”– and rightly so, as I shall explain below.

2.1 More on water, gold, lemons, tigers, etc

The idea that what they are made of, or their underlying material constitution, determines whether things that, by looking at their observational properties, we take to be water, gold, and lemons are really so grounds the micro-essentialism advanced by Kripke and Putnam in the seventies. This idea may also be seen as grounding Locke’s claim that real essence is “the real internal, but generally unknown […] constitution of things”. In this section, I shall first indicate what micro-essentialism in general– as opposed to Kripke and Putnam’s specific variety of it– does and does not mean. I shall then turn to the respective epistemic pedigrees of (i) the idea that what they are made of, or their underlying material constitution, determines whether things that we take to be water, gold, and lemons are really so, and of (ii) micro-essentialism in general, the truth of which does not imply the truth of any specific micro-essentialist view of natural kinds, such as Kripke and Putnam’s.

According to micro-essentialism, for the two ingots on my desk to be two gold ingots, and so two members of the same natural kind or two instances of the same substance, it is necessary and sufficient that they have the same atomic constitution– more precisely, that their atomic number is 79–, which makes it the essence of gold. Similarly, for the two glasses on my table to contain the same liquid, water, it is necessary and sufficient that both liquids share the same, or a sufficiently similar, microstructure (that “being composed of molecules of H2O” captures only roughlyFootnote 5), which makes water the chemical substance it is. As a consequence, if a sample of liquid is microstructurally much different from water but possesses its superficial properties (being odorless, transparent, thirst-quenching, etc.), this is not sufficient for this liquid to be a sample of water. The same goes, mutatis mutandis, for biological kinds, in the case of which DNA is the relevant microstructure: for the things that, by looking at their observational properties, we take to be lemons, or tigers, to be instances of the same biological kind, it is necessary and sufficient that their DNA is sufficiently similar.Footnote 6

Correlatively, the microscopic properties in question are responsible for the observable macroscopic properties and behaviour of these different things in such-and-such environmental conditionsFootnote 7– e.g. the thermal conductivity and tensile strength of samples of gold qua instances of that element, and this or that observational phenotypic trait of our two tigers. It may be important to underline that this is perfectly compatible with the fact that “phenotypic traits arise from complex interactions between an individual’s genes and its environment” (Leslie, 2013: 123)– so that “differences in the genetic level need not translate into differences in the biochemical processes” from which phenotypic traits result (ibid.: 138). More generally, the microscopic properties in question should not be described as having the power to cause their bearer to have certain macroscopic properties “more or less independently of [its] environment” (ibid.). What micro-essentialism implies is, among other claims, that a) entities with the same relevant microstructure cannot have or acquire, in identical environmental conditions, different macroscopic properties; and that b) any microstructural difference between two entities will or would cause them to have different macroscopic properties in certain possible environmental conditions (whether common or physically rare).

This does not imply that there must be “one specific unifying micro-constitution […] shared by most same-grouped samples” (Häggqvist & Wikforss, 2018: 921) that underlies and is responsible for their common observable characteristics on the basis of which these samples have been so grouped. In other words, the idea that the identity of things like water, gold, lemons, or tigers is determined by their underlying material constitution, as micro-essentialism articulates it, is perfectly compatible with the possibility that, as happened with jade, things with different micro-structures have been grouped together on the basis of their being similar in appearance. Similarly, micro-essentialism does not imply the possibly empirically false claim that the reason Indian rhinos grow just one horn while African rhinos grow two is that all Indian rhinos have in common a certain characteristic of their DNA while all African rhinos have in common another genetic characteristic that their Indian cousins lack. In other words, micro-essentialism can admit the possibility that, say, half of the Indian rhinos grow one horn because of a distinctive genetic characteristic they have in common while the other half grow one horn because of another.

The idea that what they are made of, or their underlying material constitution (whether we can know it or not) determines whether things we take to be water, gold, lemons, and tigers are really so is clearly not the product of an empirical discovery: we have not empirically discovered that it is what it is made of, or its underlying material constitution, that determines whether this or that ingot really is gold. What we empirically discovered more than one century ago is that this underlying material constitution consists in its microstructure. This makes it misleading to say that it is an a priori conceptual truth that gold has its actual microstructure essentially.Footnote 8

And while it is through empirical means that we can hope to discover it, this truth is not itself the product of an empirical discovery: we have not empirically discovered that it is empirically (rather than, for instance, by analysing our concept of gold) that we can hope to discover what this piece of matter is made of. This is why we would not say “I hope we are not wrong when we judge that it is what this ingot is made of that determines what it is”, nor “I hope we are not wrong when we judge that it is empirically that we can discover what it is made of”, while it makes sense to say “I hope we are not wrong when we judge (as we have done for one or two centuries) that its microstructure tells us what this ingot is made of”, or “I hope we are not wrong when we judge (as we have done for one or two centuries) that its microstructure is so-and-so”.

Therefore, when it comes to things like gold, water, lemons, and tigers, it is from empirical investigations into their underlying material constitution that we can hope to discover their essence. Correlatively, when science has still not taught us what their essence is, there is no hope that, without further empirical investigation, any conceptual analysis or understanding of a real definition could ever reveal it to us.

It follows that it is neither entirely a priori nor entirely a posteriori that we can know the essence of such things: it is because we know a priori that the underlying material constitution of the ingot determines its essence, and that we can hope to discover this constitution empirically, that we know what to look for (and where to look at, so to speak): underlying material properties of this piece of matter that determine its observable macroscopic properties in such-and-such environmental conditions. More strongly, it is because we did not have to empirically discover what to look for to discover its essence that we could discover it. Had this not been the case, then, when empirically discovering the different properties of the material entity in question, we would have had no way of determining which are essential and which are accidental to it, because their being possessed essentially or accidentally, contrary to their being possessed tout court, does not make any observational difference.Footnote 9

2.2 More on belief, knowledge, religion, personhood, etc

Let us now turn to things like belief, knowledge, religion, personhood, causation, free action, and democracy. I shall assume here that it makes sense to speak of essences of such things– i.e. that there is something in virtue of which a particular instance of, e.g., knowledge or causation is so, just as there is something in virtue of which a particular instance of gold is soFootnote 10. I shall also assume that there is prima facie no reason to think that, in trying to capture the thing in virtue of which a particular instance of knowledge or causation is so, one is is trying to capture the essence of a concept or of an abstracta.

Let us consider the idea that, as happened with water and gold, it is from empirical investigations that we can hope to discover the real essence of knowledge. This idea means that we could empirically discover that something we take, given our concept of knowledge, to be an instance of knowledge is not in fact so, and hence that it would make sense to be afraid of being wrong when taking this thing to be an instance of knowledge (where this would not simply consist in being afraid of being wrong when judging that it satisfies the conditions that, according to our concept of knowledge, something must satisfy to be knowledge).

In order to see why this idea– and this fear– in fact make no sense, let us consider whether empirical discoveries could entitle us to say “There are certain conditions C that, according to our ordinary concept of knowledge, must be satisfied for something to be knowledge, but this concept does not capture what knowledge is. Indeed, for something to be knowledge, some other conditions C* must be satisfied. And we have empirically discovered that what satisfies C does not always satisfy C*”.

Let us suppose that we have empirically discovered that what satisfies C does not always satisfy C*. Why should we draw from this the conclusion that what satisfies C is not, in fact, knowledge (contrary to what we thought before this discovery) rather than the conclusion that conditions C* do not have to be satisfied for something to be knowledge? To be rationally bound to draw the former conclusion from this discovery we would have to know that something must satisfy C* to be knowledge. But this is not something we can learn from any empirical discovery. Correlatively, we should not hope to learn from the empirical study of cases of knowledge what knowledge is– i.e. what conditions are necessary and sufficient for something to be knowledge– because this strategy would come down to saying, absurdly: “Because we know that these are cases of knowledge since they satisfy certain conditions, we can hope to learn from the empirical study of these cases what makes them cases of knowledge”. Thus, there are no empirical truths to discover about the things we take to fall under our concept of knowledge that would put us in a position to decide whether these things really are cases of knowledge and not just things that satisfy the conditions for falling under this concept. And this also goes for belief, religion, personhood, causation, free action, and democracy: the conclusion of the reasoning just developed would have been the same had we substituted any of these things for knowledge.Footnote 11

Here is another, complementary way of arguing for this point. Suppose people from another planet were to tell us that the two ingots they put in front us are made of a metal unknown to us, which they call kyxa. Suppose they leave the Earth just afterwards. How could we know what kyxa is? Only by studying its micro-constitution empirically. Suppose now that we are wondering what knowledge is, and some epistemologists from the future, who have finally succeeded in elucidating the conditions that are individually necessary and jointly sufficient for something to be knowledge, pay us a brief visit. Suppose they say, just before disappearing: “We won’t tell you what knowledge is. We don’t want to make things too easy for you. But this should help: take these two people in the room, Esther and Paul. Esther’s belief that p is knowledge, and Paul’s belief that q is knowledge”.

In this situation, our philosophical inquiry into the nature of knowledge could– and certainly should– be partly empirical: in order to elucidate the nature of knowledge, we could– and certainly should– study empirically whether the cognitive mechanisms involved in the formation of the beliefs in question are reliable, whether Esther and Paul are capable of articulating sophisticated reasons for their beliefs, whether they are cognitive achievements, whether the environments in which they formed these beliefs were such that they could easily have been false, and so on. Why? Because this would enable us to exclude some of these famous conditions for knowledge that have been advanced in the epistemology literature. But from this empirical study we could not learn which of the empirically discovered properties of these beliefs are individually necessary and jointly sufficient conditions for knowledge, while from the empirical study of kyxa we could learn which of its empirically discovered properties are individually necessary and jointly sufficient conditions for being kyxa.

Let us now turn to the (very different) situation in which we currently find ourselves when trying to elucidate the nature of knowledge. This situation can be seen in two different ways.

Suppose we assume that, in the situation in which we currently find ourselves, we do not know– because we do not know the conditions that are individually necessary and jointly sufficient for knowledge– whether this or that actual belief of this or that actual subject really is an instance of knowledge. What then could we possibly learn about the nature of knowledge from the empirical study of the beliefs highlighted by the future epistemologists? We could learn whether the cognitive mechanisms involved in the formation of these beliefs are reliable, whether the concerned subjects are capable of articulating sophisticated reasons for these beliefs, whether they are cognitive achievements, whether the environments in which they formed these beliefs were such that they could easily have been false, and so on. But this cannot teach us anything about the conditions for knowledge if we are right to assume that we do not know whether the beliefs in question really are instances of knowledge.

Suppose instead we assume that, in the situation in which we currently find ourselves when trying to elucidate the nature of knowledge, we do know that this or that actual belief of this or that actual subject is an instance of knowledge– e.g., we know that Hana’s perceptual belief that her keys are on her desk, Juan’s belief that his name is Juan, and Ivana’s belief that the Earth did not start existing five minutes ago, are all instances of knowledge. Suppose that the reason we assume that we know these beliefs are instances of knowledge is that we assume that we know these beliefs to satisfy the conditions that are required for beliefs to be knowledge. Then we would not have anything to learn from an empirical study of these beliefs into what makes them instances of knowledge. So, suppose instead that the reason we assume that we know these beliefs are instances of knowledge is that we assume that our (largely tacit) mastery of the concept of knowledge led us to correctly categorise these beliefs as instances of knowledge. What then could we empirically discover about knowledge? Suppose we discover through the empirical study of these beliefs that (e.g.):

  1. i)

    The cognitive belief-forming mechanisms that have been activated when these subjects have formed these beliefs is largely truth-conducive (in similar environments);

  2. ii)

    when asked what reasons they have to believe these different things, these subjects just respond: “I can see that my keys are on my desk!”, “Well, I just know that my name is Juan…”, and “If I know anything, it’s that the Earth is more than five minutes old”.

Would we learn from the first empirical discovery that, if our initial assumption is correct, then the activation of such belief-forming mechanisms is necessary (or sufficient) for knowledge? Not at all obviously– and the same would go if we had studied one billion similar instances of knowledge and observed that the same belief-forming mechanisms were always involved. What about the second empirical discovery? Would we at least learn from this discovery the negative truth that being capable of articulating sophisticated reasons is not necessary for knowledge, as we would in the earlier scenario with the future epistemologists, where we empirically discovered that Esther and Paul are incapable of articulating sophisticated reasons for their respective beliefs that p and that q? Not even, as we would just learn from this discovery that, if our initial assumption is correct, then being capable of articulating sophisticated reasons is not necessary for knowledge.

This does not mean that there are no important empirical discoveries we can make about knowledge, belief, religion and the like. In particular, the discovery that there is something in the world that satisfies the non-empirically discoverable conditions that must be satisfied for these things to be instantiated must be an empirical discovery, as must the discovery of how these conditions causally came to be satisfied if they are in fact satisfied. For example, the discovery that there are countries in the world that satisfy the conditions that are required for being democracies cannot but be empirical; and, supposing there are such countries, the same goes for discovering how they came to acquire or possess the properties in virtue of which they are democracies.

The moral of the foregoing is that the conditions that are individually necessary and jointly sufficient for these things to be what they are can no more be empirically discovered than it can be empirically discovered that the underlying material constitution of what we take to be gold, water, lemons, and tigers determines their essence, and hence whether they really are so. But since it is empirically that we can hope to discover their essence, contrary to the essence of belief, knowledge, religion, personhood, causation, free action, and democracy, it follows that, if it is admitted that all these essences can be known, then uniformism is false: they cannot all be known in the same way.

3 How to explain this non-uniformism?

Our epistemic access to the essences of certain things then differs from our epistemic access to that of others. It seems that two options present themselves here: either non-uniformism is a brute epistemic fact– i.e. it is a brute epistemic fact that our ways of acquiring knowledge of essences differ–, or this epistemic difference corresponds to, and results from, a difference in nature between the things in question. While I find the former option most implausible and intellectually unsatisfying, the latter seems to me most intuitive. I shall therefore suppose here that it is correct. (Note that this supposition is not the reverse assumption that any important metaphysical difference or heterogeneity within the domain of the truthmakers of one’s modal and essential beliefs is epistemically relevant– i.e. calls for different epistemologies that mirror this heterogeneity.Footnote 12)

Is the abstract–concrete distinction capable of entirely explaining the difference I have underlined between how we can hope to know the essence of water, gold, lemons and tigers, on the one hand, and of belief, knowledge, religion, personhood, causation, free action, and democracy, on the other? This distinction is not capable of explaining this because religions, persons, causes, free actions, and democracies clearly do not belong to the category of abstracta: unlike sets and numbers, which are generally taken to be indisputably abstract, these things clearly are spatiotemporally located, and clearly possess causal powers.Footnote 13Abstracta, on the other hand, are neither “denizens of space-time”, due to their lacking spatiotemporal properties and relations, nor “subject to causality”– which “perhaps amounts to the same thing”, as Lowe suggests (1998: 51).

Let us now explore the hypothesis that the multiply realisable–uniquely realisable distinction is a better candidate for the job of metaphysically grounding the distinction between things the essences of which are empirically discoverable (water, gold, lemons, tigers) and things the essences of which are not empirically discoverable (belief, knowledge, religion, personhood, etc.).

Multiple realisability can be minimally defined as follows: a property P is multiply realisable if and only if different kinds of physical properties can give rise to instances of P that do not differ in P-ness in the way instances of different determinates of a same determinable D differ in D-ness. These different kinds of physical properties are the multiple realisation bases, or the multiple realisers, of P. (This minimal definition goes against Stephen Yablo’s suggestion that multiple realisability just is a species of the determinable–determinate relation (Yablo, 1992). Against this suggestion, it can be argued, as Funkhouser (2014) and Haug (2010) have, that there is a fundamental difference between multiple realisability and the determinable-determinate relation: if a given property is multiply realisable when different kinds of physical properties can give rise to instances of this property, it makes sense to wonder how this is possible and to look for a physical explanation of this fact– e.g. of the fact that different neurophysiological and neuroanatomical properties can give rise to pain. But it does not make sense to look for a physical explanation of the fact that instances of scarlet and vermillion give rise to instances of red, and, more generally, to wonder how this is possible.Footnote 14)

The canonical example of multiple realisability is pain. A mental property such as pain is sometimes said to be multiply realisable on the grounds that it is possible that living organisms such as tigers, alligators, or ducks, whose physical properties– neurophysiological and neuroanatomical in particular– strongly differ from ours, nevertheless experience pain. Since we cannot be in pain without having C-fibres firing, and since these animals, being neurophysiologically and neuroanatomically different from us, have no C-fibres, it follows that, if they experience pain just as we do, then the physical property of having C-fibres firing is not necessary for experiencing pain: a variety of physical properties– of physical states of living organisms– can give rise to instances of the same kind of pain, which means that it is multiply realisable.

Let us now consider once again our two lists of things. When it comes to gold, water, lemons, and tigers, their physical microstructure is essential to them. And this makes these things uniquely realisable. Conversely, as I have underlined, whether or not something is a belief, an instance of knowledge, a religion, a person, a cause, a free action, or a democracy does not depend on its underlying material constitution. Correlatively, such things seem to be multiply realisable: it seems that different kinds of entities with very different physical properties– human beings, non-human animals, advanced silicon-based systems, aliens, etc.– can be persons, act freely, have beliefs and knowledge. And, obviously, things with totally different physical properties can be democracies, religions, and causes of other things. This does not mean that, for example, different persons or democracies can be so in virtue of totally different physical properties, but rather that their being different persons or democracies does not depend on their physical properties, in the following sense. Whether a given individual really is a person, or whether a given country really is a democracy, does not depend on what they are made of, in the sense that their being a person, or a democracy, is compatible with their having very different physical properties.Footnote 15

One could then suggest that the multiply realisable–uniquely realisable distinction grounds the fact that we can hope to discover the essence of things like gold, water and lemons empirically, whereas this is not the case for the essence of things like persons, beliefs, and knowledge (which is obviously not to say that no empirical knowledge is necessary for knowing them). This idea, however, must be refined.

First, let us consider the essence of God, nothingness, and numbers: just like the essence of things like persons, beliefs and knowledge, these essences cannot be discovered empirically. And God, nothingness, and numbers (or, more generally, abstract entities) are not multiply realisable, but non-realisable.Footnote 16 So a better candidate for grounding non-uniformism is the uniquely realisable–non-uniquely realisable distinction, where something is non-uniquely realisable either when it is multiply realisable or when it is non-realisable.

Second, the idea that the uniquely realisable–non-uniquely realisable distinction grounds non-uniformism must be expressed in a more precise way. If multiple realisability basically consists in the fact that different kinds of physical properties can give rise to instances of the same property P, the claim that belief, knowledge, religion, personhood, causation, free action, and democracy are multiply realisable must be understood in the following way: their essential properties are multiply realisable, which means that different physical properties can give rise to instances of these essential properties, unlike the essential properties of water, gold, lemons and tigers, which are uniquely realisable. In other words, there cannot be different physical properties that give rise to instances of the essential properties of such things.

This not only goes for subtances and for objects that possess, as Lowe puts it, “the kind of unity” that makes them, “at least in principle, countable” (Lowe, 2007: 518). This goes for all things we have grouped together on the basis of similar observational features, and about which we think that empirical investigations will tell us whether they really share a common essence. For instance, let’s take the case of events and processes we have grouped together on the basis of their observational features, such as lightning flashes in the Earth’s atmosphere: it is only if we discover that they all consist in a distinctive series or pattern of physical causes and effectsFootnote 17 that is grounded ultimately in the underlying material constitution of the entities involved– and hence that is uniquely realisable– that we will conclude that these events or processes share the same essence, and that we have discovered itFootnote 18.

Returning once again to our two lists of things, the metaphysical difference between being uniquely realisable and not being uniquely realisable turns out to coincide exactly with the epistemic difference between having an empirically discoverable essence and not having an empirically discoverable essence. Perhaps surprisingly, there is a 1–1 correspondence between possessing (or not possessing) the metaphysical property of being uniquely realisable and possessing (or not possessing) the epistemic property of having an empirically discoverable essence. Because I could not identify any other potential metaphysical ground of this epistemic difference that would imply this 1–1 correspondence, I conclude that the hypothesis that the metaphysical difference in question grounds the epistemic difference in question is the best potential metaphysical explanation at our disposal of the form of non-uniformism discussed in Sect. 2.

One may however doubt that the uniquely realisable–non-uniquely realisable distinction can really have the power to metaphysically ground the fact that some things have an empirically discoverable essence whereas other things do not. One may doubt it on the grounds that being non-uniquely realisable is a negative property shared by so many different things that possessing it cannot be a satisfactory explanans of (most of) what is true of the things with this negative property. Take for example the property of not being an elephant: this negative property is possessed by so many things of very different types– numbers, minds, chairs, storms, battles, paintings, rocks, universities, redness, etc.– that it cannot explain, e.g., why numbers are not located in time, why minds are not spatially extended, why storms necessarily are temporarily extended, why rocks have mass, etc.

In order to assess whether the fact that being non-uniquely realisable is a negative property shared by very diverse objects is, in itself, sufficient for preventing this property from being a satisfactory explanans, consider the following case. The reason why Helium-4 turns into a superfluid when the temperature is close to absolute zero is that Helium-4 is a many-particle Bose and Fermi system, and the reason why numbers, minds, chairs, storms, battles, paintings, rocks, universities, redness, etc., do not turn into superfluids when the temperature is close to absolute zero is that these various things are not many-particle Bose and Fermi systems. The negative property of not being a many-particle Bose and Fermi system intuitively is a satisfactory explanation of the fact that these very heterogenous things do not turn into superfluids when the temperature is close to absolute zero. Moreover, I do not see any compelling reason why things should be different when it comes to the property of not being uniquely realisable and the property of not having an empirically discoverable essence.

That being said, in order to make more intuitive the claim that the uniquely realisable–non-uniquely realisable distinction is capable of grounding the empirically discoverable essence–non-empirically discoverable essence distinction, more can be said about the properties of being uniquely realisable and of being non-uniquely realisable. To the question of what makes it the case that something is uniquely realisable, a quite uncontroversial answer is that something is uniquely realisable when its underlying material constitution makes this thing the kind of thing it is, and so is the essence of this thing. Now, if it is admitted that the underlying material constitution of something is empirically discoverable, and empirically discoverable only, it follows that the essence of uniquely realisable things is empirically discoverable, and empirically discoverable only. When it comes to the question of what makes it the case that something is non-uniquely realisable, a quite uncontroversial answer is that something is non-uniquely realisable when its underlying material constitution does not make this thing the kind of thing it is, and so is not the essence of this thing. This makes it possible that the essence of non-uniquely realisable things is not empirically discoverable, if it is admitted that the underlying material constitution of something is empirically discoverable onlyFootnote 19.

4 Non-uniformism, science, and philosophy

Let us now consider one last time our two lists of things. Intuitively, discovering the essence of water, gold, lemons, and tigers belongs to science, while discovering the essence of those of belief, knowledge, religion, personhood, causation, free action, and democracy belongs to philosophy. This, combined with what has just been said about the metaphysical grounding of non-uniformism, suggests the following twofold thesis:

  • Discovering the essences of things whose essential properties are uniquely realisable– because what is essential to these things is their microstructure– belongs to science, and cannot but be empirical.

  • Discovering the essences of things whose essential properties are not uniquely realisable belongs to philosophy, and cannot be empiricalFootnote 20.

These claims are supported by two observations. First, among the different things mentioned in this paper (democracies, persons, causation, etc.), those things whose essential properties are not uniquely realisable are also those things whose nature it is intuitively up to philosophy to elucidate. Second, the things whose essential properties are uniquely realisable are also things whose nature it is intuitively up to science to elucidate.

In order to briefly illustrate these claims, let us consider the case of knowledge and the case of pain. If one empirically inquires into whether certain phenomena one has grouped together and tagged as instances of knowedge have in common a certain causal profile, one is not looking for the essence of knowledge in the sense in which one does when one wonders whether knowledge is justified true belief. Indeed, when one inquires into this, one is not looking for the conditions that are necessary and sufficient for something to be knowledge because, as underlined above (cf. § 2.2), it would not make sense to say, when looking for them, “Since this and that are instances of knowledge, let’s now empirically inquire into whether they really are so, and what makes them so”.

Similarly, suppose one has grouped together, on the basis of observational similarities, certain (verbal and non-verbal) behaviours, and their associated neural activities, as “pain behaviours and neural activities”. Inquiring into whether all of them are elements of a distinctive causal pattern grounded ultimately in the microscopic properties of the entities involved is clearly doing science. And if one discovered that all of these behaviours and neural activities indeed are elements of such a pattern, one then could argue that one has empirically discovered that they share an essence, composed of uniquely realisable properties– just as if we had discovered that a certain group of things we observationally group together as “lemons” all share certain distinctive genetic characteristics. That would not be, however, having inquired into, and discovered, the essence of pain, i.e. the properties that are necessary and sufficient for something to be pain, and that are multiply realisable. In order to discover these properties, and hence to grasp this essence, one has to inquire into, e.g., whether one can have someone else’s pain, whether there can be unconscious pain, or whether pain can be pleasant. This is doing philosophy, and there is no empirical discovery to be made that would answer these questions.