I actually like Kant's distinction between noumena and phenomena. But I have a nagging doubt.

If we look at modern physics, appearances can be explained by entities such as atoms, electrons and quarks which are not themselves picturable. I can't say that this reality is the noumenon because by Kant's definition the noumenon cannot be known.

But the approach of scientific rationalism may undermine the idea of the Ding an Sich. Maybe we don't need it. We just keep exploring. Kant had no knowledge of modern science.

Any help here would be much appreciated. I have searched Google in vain. Has anyone written on the matter?

  • 1
    I'm not sure if you're equating the noumena with the Ding an Sich, but if you are, be aware that not everyone agrees they are equivalent concepts. See 'Noumenon and the thing-in-itself' on This page. Commented Aug 26, 2021 at 13:40
  • 2
    Science deals with what we can perceive (empiric knowledge = empiric truth), not with the Ding-an-Sich. We don't have access to it, and reaching it is not the goal of science, it is impossible. "We don't need it": correct, we don't need it IN SCIENCE, because science deals only with experience. The Ding-an-Sich is, on the contrary, a key topic in philosophy (the mother of all sciences), specifically, it is essential in metaphysics.
    – RodolfoAP
    Commented Aug 26, 2021 at 14:02
  • 1
    @RodolfoAP: Saying 'we don't need ding an sich in science' is misleading. Science is modeling, which means there must be a 'thing' we model for the model to be meaningful. If I make a model of the Grand Canyon, that model ins't the Grand Canyon; if no such thing as the Grand Canyon exists — no ding an sich — that 'model' is at best a flight of fancy, without relation to science. Commented Aug 26, 2021 at 14:58
  • 1
    Please do not confuse the noumenal realm with Ding an sich. The latter is only a friendly reminder that what we represent as object of experience is formed by our cognitive faculties, ie. the human way of representing the object in question. Ding an sich was never intended to be understood as any kind of reality. To be precise, it is a contradictio in adjecto, see this answer of mine, under further evidence. The noumenal is real in Kant though.
    – Philip Klöcking
    Commented Aug 26, 2021 at 15:56
  • 2
    Well, do we need Kant at all? Scientific rationalism can probably get by without him. It is unclear what you are looking for. For the question to make sense you need to work from within Kantian perspective and find internal arguments for undermining Ding an Sich. An outside perspective does not undermine one piece in particular, it "undermines" the whole doctrine, if anything.
    – Conifold
    Commented Aug 26, 2021 at 20:03

3 Answers 3


The thing-in-itself isn't the same as theoretical entities, i.e. entities that we cannot directly perceive but can infer from our observation. Kant doesn't, unlike some empiricists, make a sharp distinction here. When he discusses the Third Analogy of Experience in the Critique of Pure Reason he says that, although we cannot directly perceive magnetic fields due to the particular nature of our senses, we can know it perceptually due to an immediate inference from the presence of movement of iron filings. Kant, in other words, would deny that there is any ground for skepticism about our knowledge of magnetic fields just because they cannot be, strictly speaking, directly perceived by us ("directly perceive" is my term, not Kant's - I hope it's intuitively intelligible what I mean).

You significantly underappreciate human ingenuity before "modern science" (let's say, XX-th century science). Atoms and other such entities were postulated long before Kant to explain various phenomena. And, yes, there were some who insisted that this practice is scientifically unsound (although I see no grounds for this) or (a view which is still common) that these are merely 'convenient fictions'. But Kant holds neither of these views.

The thing-in-itself doesn't have anything to do with empirical knowledge. It can be defined the object of knowledge of pure reason, entirely independently of affection by an object which is the cause of experience. And Kant denies such knowledge, metaphysical knowledge, but not scientific explanations which reference entities like atoms etc. which cannot be observed.

It must be said, however, that Kant himself wasn't a proponent of atomism.


"Kant had no knowledge of contemporary science" is I think a more accurate statement than "Kant had no knowledge of modern science," because "modern" science" generally refers to science since the scientific revolution of Newton and his approximate contemporaries in other fields of empirical inquiry. Kant's project was more "grounding" science than inventing something beyond the reach of science with the agenda of keeping philosophy above science (although a case can certainly be made for the latter). I don't think he considered his metaphysics a necessary and hitherto missing piece for the continuing progress of science; I think he considered it a pat on science's back.

But the lasting value of Kant's metaphysics is most evident in what future philosophers have developed off of it. The Ding-an-Sich concept isn't transformed by contemporary science; it's transformed by Hegel, who tries to demonstrate that it's not a thing that exists but a thing in us that defines our investigations of existing things (much like your sentiment that we just keep exploring and finding out more). I recommend Robert Brandom's reading of Kant and Hegel. And also Quine's "Two Dogmas of Empiricism" (that explodes Kantianism as much as anything does, at the analytic/synthetic distinction pivot-point of his theoretical edifice). (Rorty on Quine in Philosophy and the Mirror of Nature for a more readable and less technical rendering of the point.) But my overall point is just that there's a lot more to Kantian metaphysics than a surface reading of the phenomenal/noumenal distinction.

  • Good reply. I read Quine's Two Dogmas of Empiricism many years ago. As far as I remember he tried to demolish the analytic/synthetic distinction. I will get back to it, and also check out Hegel's view. I will even check out Rorty.
    – Marek
    Commented Aug 30, 2021 at 6:22
  • Rorty, no, no. He carries pragmatism to a ridiculous degree.
    – Marek
    Commented Sep 6, 2021 at 13:57
  • But. if we are investigating particles, would it make sense to say the there is a thing in itself behind the electron. According to much of contemporary physicists the electron is a purely relational entity.
    – Marek
    Commented Sep 26, 2021 at 13:00

The key to this problem is that the physical sciences produce models of the world; they do not access the world as such. When we talk about subatomic particles — proteins, electrons, neutrons, quarks, or what you will — we are applying concepts defined within a model that largely works to describe the events we see. We cannot see the particles directly, but we can see indirectly see effects on the world that we can attribute to these particles within this model, Therefore the particles are phenomena. In other words:

  • We assert that a table is an 'object' because we can place a cup on it, and the cup won't fall through to the floor (as opposed to, say, a hologram of a table). If we were to make a table out of a perfectly transparent material, we could still assert it was an object, because we could still place a cup on it without the cup falling to the floor. The cup not falling as a phenomenal experience that points to the existence of an object.
  • We assert that a quark is an 'object' because we can perform certain (highly technical) actions and produce consistent with the model that defined quarks. It doesn't matter that we can't 'see' them directly; like the transparent table, we see an effect that implies the existence of an object.

We don't really 'know' what's happening 'in reality' on the subatomic level. That's all noumena, beyond our grasp. We see certain phenomena; we build models that attempt to encapsulate that phenomena systematically; those models define arguable/argumentative objects that ostensible encapsulate meaningful elements of Ding an Sich. But there is always an insurmountable gap of knowledge between those modeled objects and Ding an Sich itself.

  • All current physical models are understood as approximate, but the goal of many physicists is a final, most fundamental "theory of everything" that would not be merely approximate in this way, at least as far as prediction is concerned. If we had such a theory, presumably it would still deal with purely mathematical properties of fundamental units (particles or strings or whatever), but would there be any good argument for thinking these units must have additional properties (metaphysical or otherwise) beyond those included in the theory?
    – Hypnosifl
    Commented Sep 28, 2021 at 15:32
  • @Hypnosifl: Physical models work because physical models whitewash irregularities under the error term. For instance, to 'truly' know the inertial moment of an object, we ought to know the composition and distribution of every single atom within that object: replace a carbon atom with a lead atom anywhere, and the inertial moment changes ever so slightly. But for most cases we can ignore such issues, modeling a point mass with uniform distributions and properties, because the results we get are within the error tolerances we set for ourselves. Commented Sep 28, 2021 at 15:45
  • @Hypnosifl: The more 'Grand' our Grand Theory gets, the more it becomes a whitewashed abstraction, divorced from the idiosyncratic details of the real world. Even if we get an exact understanding of how the fundamental forces interact with each other, we still have to apply it to practical uses, where the detail devils hold sway. As Ram Dass once said "I you think you're enlightened, go home a spend a weekend with your parents"; likewise, if you think you have perfect understanding, try putting it to use. Commented Sep 28, 2021 at 15:50
  • 1
    The fact that all models so far are approximations isn't a strong argument for the impossibility of an exact fundamental theory. Just as a hypothetical, imagine a world that does have such an exact mathematical rule underlying its behavior (like a cellular automaton), but where intelligent beings would need many centuries of technological development before they could get the necessary empirical evidence to demonstrate which fundamental theory is correct. Until then, wouldn't you expect all their models to be approximate, with error terms?
    – Hypnosifl
    Commented Sep 28, 2021 at 16:24
  • @Hypnosifl: You're missing the point. An exact model is still a model. If I had an exact map of the surface of the earth, it still wouldn't be the surface of the earth. It would tell me how to get to Chicago, but I'd still have to put feet on the ground to go there. Commented Sep 28, 2021 at 16:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .