Natural kinds

3rd December 2020

A premise on Modal Logic.

If all entities of type A share property p, (all pieces of gold have a specific atomic structure) are


Grouping things in kind is important in science, so that we can expect behaviors, inductive reasoning work accordingly; a great grounding for predictions

+++ role in scientific laws and explanations

Questions to settle

  • what make us pick certain properties
  • what make certain entities appear sufficiently similar (see [[ Introduction to Scientific explanation#The raven Paradox ]])

For example, ions, positrons, protons are all three positive charges, but this is not enough to make us state that they belong to the same kind.


Essences: properties which make an object belonging to that kind. Such kind is natural if it corresponds to recurrent properties which are in nature independently of human classifications. Essentialists defend this kind of reasoning, even though anti-essentialists may object that having mind-independent classifications is quite hard if not impossible.

One way to capture the essentialist intuition in a way that perhaps is a little bit less exposed to the objection of the anti-essentialist comes from a well-known discussion concerning the so-called semantics of natural kind terms. This has to do with the very famous debate in the 1970s (1979-1980) carried out by both a philosopher of science and a philosopher of language: Saul Kripke and Hilary Putnam.


According to descriptivism, the meaning of a term is determined by the verbal formulation on the properties relative to these terms; these verbal formulations are named definite descriptions.

For example, we describe water as being liquid, colorless, odorless, drinkable and so on. All of this fix the meaning of the term “water”. These descriptions, according to this view by Kripke and Putnam, are taken to be necessary and sufficient conditions for sorting out any candidate to the natural kind “water”.
In other words, we need the meaning in order to refer to the kind. This, supposing we all know the proper description before hand. When we use the term “water” in a proper manner is because we know this meaning, we know what we’re talking about.

Against Descriptivism

Against this view, Kripke and Putnam make us notice that in fact the above are not necessary and sufficient conditions. Why? Because we might know the meaning while referring to the wrong kind. How is that possible? To prove this, Putnam suggests the “twin earth” thought experiment.

Twin earth experiment

There’s a planet where there’s an element which looks like water, called by its inhabitants “water”, but with different chemical properties. When inhabitants of both planets refer to “water” they mean different substances.
If this is the case, then there is more to reference than what meaning alone describes.

In a modal form:
∀ x, x is an inference of the kind K iff it falls under the description D.

The above is a necessary and a priori kind of sentence.


  • it is possible that there are instances of K that do not possess D. This does not seem to be such an easy possibility to accept, for if we know a priori that to be an instance to K then D is needed in order to say that x is an inference of K, then this possibility seems to be hold out from the start.
  • It is possible that there are items that share D but are not instances of K. This might be more of a natural possibility, for we cannot rule out that distinct kinds might share the same manifest property and yet not being of the same kind. And in actual fact, even in our world we can see gold and what is called Fool’s Gold/pyrite: they look the same, but only one is gold. We might imagine that gold might look blue rather than yellow in specific circumstances, but then again, this is something that you learn/put in context/learn a posteriori, it is not an a priori definition, it is not a definite description.

Essentialism in science

The periodic table is the perfect example of natural kinds. Elements are neatly and precisely defined and distinguished, not by men but by nature. Furthermore, there is only one correct way to divide out the world in kinds. The kinds are natural precisely for this reason.

From this stems the definition given by Robin Hendry:


According to microstructuralism, a kind is identified by the microstructural properties of each element – for ex. the atomic number of their nuclear structure.

Open issues

Even granting that water has a particular microstructure, on what does it depend? Interactions might change, they’re not always the same; even if we focus on similarities on a microstructural level we may think of or come across to, the situation is less settled than one might expect. In addition to this, sometimes properties are not as important as their function, so the focus is on how certain things actually work within an aggregate rather than what are the things that are in there.

It is as if essentialism in chemistry might go in the direction of admitting some kind of different forms of pluralism, concerting the very nature and role of microstructure, depending, as we said before, what we are targeting and what for. But if the whole point of assuming a microstructure is to decide what the kind is and what I means to belong to that kind, admitting that there might be different essences or different ways of doing this might actually depict the purpose of this. It might give us on one side enough information to understand specific types of entities, but it doesn’t give, in the end, enough information to demarcate one kind from another. It doesn’t play that role, microstructure doesn’t play that role. This is a situation that does not only occur in chemistry. Another typical example is biology. The case of biological species is quite interesting, because a species might appear as a classical example of natural kind (what else if not a species?), but if we try to figure out whether species are natural kinds, then again we sees to be struggling with using essentialism criteria. For example, let’s take the Darwinian theory of evolution as applied to species. It tells us that there are no properties of species that all the members of a species actually share, and even if they do share this properties, evolutionary history tells us that in fact these properties can and do change. All this, again, seems to be the antithesis of essentialism. So, how can we make use of this idea of essence if it is something that mutates, that all the members of the kind share in a way that we can say “this is the criterion for the belonging to the kind”.

This has made some essentialists claim that essentialism perhaps adapts better to certain types of science or branches rather than others. Other essentialists went in the direction of rethinking essentialism in a different way, in a way that is able to handle at least some of these objections. There are 2 main amendments of essentialism in the literature in philosophy of science (they are not only 2, but they are the main). The first one is called “property cluster theory” of natural kinds: this is a more relaxed view about natural kinds, for it only requires members of a kind to share a subset of properties that might end up clustering together due to some underlying cause. There are different types of properties: some properties are more proper to assemble than others. These are actually properties that cluster together, and it’s enough for a kind to target these properties in order to say “well, such and such is an instance of that kind”. There is no necessity, in other words, that the properties are always the same, the properties that identify a kind are always the same, but it is enough that some of them are shared in the way that that kind describes. Take again the case of biological species: members of a species may share many (not all) properties caused by various mechanisms to belong to the species (for example, sharing a common ancestor is one candidate, or sharing a particular ecological mesh, or coming under a particular mechanism of development or a particular gene exchange. Some of these properties can get to cluster together, but then this leaves open a wider possibility of variation of difference. So, the clustering is a more relaxed criterion than essence in the way we’ve been talking about.

Natural kinds by Tommaso Tonello

This doesn’t occur in chemistry only, also in biology.

The case of biological species again may be used to confute essentialism.

Two main ways of rethinking essentialism:

  • proper cluster theory, Boyd
  • promiscuous realism, J. Dupré:
    in his perspective, Essentialism is part of an epistemological packet, composed also by
    • determinism
    • reductionism

Disunity of sciences: reality as different levels of descriptions. Not an epistemological monism, but a plurality of methods, epistemic measures we use to depict our theories.

Promiscuous realism

The metaphysics of modern science is taken to be deterministic in some way: +++.

The metaphor which represents it is the clock:
a mechanism that, once represented, +++.

The above is useful in particular to understand the thesis of reactionism.

Complex mechanisms which becomes simpler and simpler.

Besides individual things, +++


Essentialism is in part defended by Dupré: there actually are basic divisions naturally in the world, but they aren’t enough. The context of use chenges things, but this doesn’t mean that there are no natural kinds.

The promiscuity of kinds +++

How this promiscuous position positions itself with regard to traditional metaphysics?


A natural kind must include anything which is fundamental to make that kind.

The Ontology of common sense, which is


4th December 2020

By means of n.k. we can classify entities. Grouping should not be accidental. Classification in a universal manner, not contingent.

Essences might be a way to classify natural terms.

How can we envisage a correspondence between artificial classification and actual qualities of entities.

These qualities themselves are the result of a theoretical speculation, not a self-evident characteristic.

Classifications are meant to have a purpose and a role.

Not an argument per se against realism, but the temptation is to say if we take what the world is to be a description of our words, +++

Anti-realism perspectives: our world is more what we label than what it actually is.


Applied Metaphysics

There is a difference also in social kinds, according to Hackings.

+++S quarks are not aware

Natural kinds are defined as “indifferent”: stationary, not passive, targets. They don’t do any of the things that we do because they’re not aware of the classifications we give them

about human kinds:


Social kinds are “interactive”: their definition is strongly related to the label it’s attributed to them. They rethink themselves accordingly.

There is a kind of feedback, looping effect, involved when we classify people.

For this (+++) to be

+++ example of autism

John Searle

This is the last of Logic and Philosophy of Science I notes



edit this page

The following is a graph containing all the notes and every topic in the website