Realism and Correlationism: Kant and the Short Argument

Meillassoux takes the correlationist to rely on the following argument:

thought cannot get outside itself in order to compare the world as it is ‘in itself’ to the world as it is ‘for us’, and thereby distinguish what is a function of our relation to the world from what belongs to the world alone. Such an enterprise is effectively self-contradictory, for at the moment when we think of a property as belonging to the world in itself, it is precisely the latter that we are thinking, and consequently this property is revealed to be essentially tied to our thinking about the world. (AF: 4)

This argument is a form of what Karl Ameriks calls the ‘short argument’ to idealism, which often gets attributed to Kant. However, Kant does not make this short argument. Ameriks traces this form of argument to Reinhold, and he notes that it does sometimes appear in the post-Kantian tradition. So, we find Reinhold claiming the following:

What is represented, as object, can come to consciousness and become represented only as modified through the form of representation, and not in a form independent of representation, as it is in itself. (Versuch: 240; quoted in Ameriks FoA: 129)

Reinhold takes it that a need to represent objects for them to be given to consciousness ensures that we cannot come into an epistemic relationship to those objects which could be disentangled from our representations:

The concept of a representation in general contradicts the representation of an object in its distinctive form independent of the form of representation, or the so-called thing in itself; that is, no thing in itself is representable. […]

[T]he object distinguished from the representation […] can only be represented under the form of representation and so in no way as a thing in itself. (Versuch: 244, 246)

So, for Reinhold, because we cannot get outside of our representations, then objects cannot be represented as they are in themselves.

If the correlationist — whatever ‘originary correlation’ they are meant to argue for, and whatever it means to say that they cannot consider its terms independently — has to rely upon this argument as it stands, they are in trouble. This is because the conclusion it argues for is trivial given the way key terms in the argument are understood. Reinhold is trying to prove that we cannot know things in themselves, where he takes knowledge to require that objects are represented to us. But if he tacitly understands ‘things in themselves’ just to be what is not representable, then the conclusion follows all too easily. Thus, on its own, this argument ought to convince no-one.

Meillassoux’s presentation of the argument proceeds in a similar fashion. It seeks to establish an (underspecified) ‘essential tie’ between thought and things in themselves. Like Reinhold, this is meant to undermine the possibility of an epistemic relation to the world as it in itself independently of thought (one that the realist requires to distinguish primary and secondary qualities). The way that it does this is by simply noting that we cannot think of features of the world in itself without the world in itself being the object of that thought. Thus, we must always factor in a correlation between thought and the world in itself when attempting to reflect on the latter. Again, the shallowness of this argument ought to be transparent. Knowledge of the world in itself, as required by the realist, is denied to us because thinking is always present when thinking about the world in itself. However, this is only because here we are to understand knowledge of the world in itself as knowledge where thought is not present. The opposition is simply defined out of existence. Nothing is demonstrated by this argument, and it is no more contentful than Reinhold’s efforts.

* * *

Even with Meillassoux’s distinction between weak and strong correlationism, and the specification of different possible correlates than simply thought and world, I am not yet clear in my own mind what the status of the correlationist’s claim that thought and world must be thought together is meant to be. So, I am hesitant to assert or deny that particular philosophers are correlationists. Besides, I am not sure how useful a discussion along the lines of ‘is x really a correlationist?’ would be. Still, insofar as transcendental idealism can be thought of as introducing some significant relation between thought and world, whether we understand this idealism as metaphysical, formal, methodological or whatever, then it may bear considering in this context.

However we understand the relation between objects and cognition in Kant, I have claimed that we do not find a ‘short argument’. Yet, Kant does claim that objects conform to the conditions of cognition. So, we can ask, how does Kant’s position differ from the ‘short arguments’ dismissed above? This ought not to be of mere historical interest insofar as it can furnish us with alternative arguments for either correlationism or a more plausible relative of it. Speculative realists have an interest in attending to other such strategies insofar as their own positions can develop in dialogue with a wider range of opposition than the colourless proponent of the short argument.

Transcendental idealism famously effects a Copernican turn. Instead of assuming that all our knowledge must conform to objects, Kant ventures a hypothesis: objects must conform to our knowledge. This claim has proven difficult to understand. It is clear that Kant is not asserting an empirical idealism, which holds that objects have a metaphysical dependence upon our epistemic activity or our ‘representations’. Kant denies this when distinguishing his position from what he calls Berkeley’s dogmatic idealism. In the Prolegomena, he calls his position formal idealism, and any dependence of objects upon our knowledge is restricted to the forms of our knowledge. In the Analytic of the first Critique, regarding the categories of the understanding, Kant denies he is engaged in a traditional metaphysical investigation of being qua being (A247=B303). However, it can appear that the Aesthetic claims that our forms of sensibility, namely space and time, are ontological conditions of objects (although Kantians such as Henry Allison and Graeme Bird forcefully argue against such a reading). Whatever the right interpretative approach here, obviously some important connection between formal conditions of knowledge and objects is being asserted. But why? The answer provides some possible motivations for something like a correlationist position which are not simply versions of the short argument.

Kant makes his speculative Copernican hypothesis because he is dissatisfied with metaphysics. When compared with mathematics, say, which also seeks knowledge which is not directly empirical, it can hardly be said to be on the ‘sure path’ of science. For Kant, this was illustrated by the hollowness of metaphysical inquiry into the nature of the soul, God and world, reflected in the the interminable debates in rational psychology, rational theology and rational cosmology which are diagnosed in the Transcendental Dialectic. The problem, he thinks, is that metaphysics has employed theoretical reason in illicit ways, beyond its proper bounds. Traditional metaphysicians have failed to take into account the anthropocentric forms of human cognition, and so constantly come to grief by asking of reason what it cannot deliver. However, this is merely a sketch of some of the territory. There is no swift move from registering the forms of human cognition and towards sealing us off from a non-human world. From the bare fact that it is our cognition, it does not follow that it cannot deliver things in themselves. To attribute such a short argument to Kant on this basis is to ignore the details of Kant’s examination of cognition and his lengthy inquiry into metaphysics.

If transcendental idealism does ultimately count as a form of correlationism, this will be on the basis of the determinate limits on knowledge explored in Kant’s inquiries. These include sensible conditions, intellectual conditions, cognitive conditions governing the relation of the sensible and intelligible (e.g. the discursivity thesis), and rational conditions pertaining to the proper use of practical and theoretical reason. Each is supported by argument and analysis, which vary in success. For example, the intellectual conditions on empirical knowledge include conformity to the categories of the understanding. These conditions on thought are backed by an examination of the forms of judgement, which many people have found problematic and dogmatic. This set of conditions will probably not be the most troubling for the speculative realist though (Kant allows that we can think the thing itself — though whether that is just as a limiting concept is debatable). Rather, it will be the sensible conditions which will be most problematic. These sensible conditions enable objects to be given. Thus, they provide the main receptive framework for cognition, where the understanding provides the main spontaneous framework. Objects are given to sensibility according to its forms, namely space and time. This can seem an unassuming empiricist move: we know about things through spatio-temporal experience. But it goes beyond this insofar as Kant’s Copernican turn makes an a priori pure form of intuition logically prior to objects. Objects are given according to this pure intuition, such that they have formal properties in conformity with this pure form. This can be understood in more or less metaphysical terms. It is where realists will doubtless demur though, since it can seem to impugn the independence of objects from our cognitive apparatus.

Why does Kant embrace something like correlationism here? Some reasons are arguably idiosyncratic. For example, Kant thinks that we require pure forms of intuition to help apply the categories of the understanding (such as existence or plurality) to sensible objects — they bind the a priori and the empirical together ‘schematically’. Also, given his understanding of geometry and arithmetic, pure forms are meant to explain the synthetic a priori status of mathematical knowledge.

What may have a wider resonance though is the role of forms of intuition in grounding Kant’s revised metaphysics. Kant thinks that reason can be shown to fail when, like the rationalists, it strays from the path of possible experience. This was what led metaphysics into darkness. But if objects have to conform to the forms of intuition, then their formal properties can be grasped a priori. So, for any object which is given to us, we can justify limited metaphysical knowledge of it with reference to the pure forms, since nothing can be given that does not conform to these forms. Kant sums it up like this: “reason has insight only into that which it produces after a plan of its own.” Now, by my lights, Kant’s specific appeal to pure forms of intuition is not ultimately successful. But it does give a substantive argument for a correlationist-like understanding of the relation between objects and cognition. Furthermore, it outlines a strategy which I think can be made to work, albeit in a heavily revised form, with respect to the normative bases of cognition (and which, in time, I hope to outline).

* * *

A final thought on the question of metaphysics. The metaphysics which Kant seeks to cut down to size is an unbridled rationalism. But speculative realism has typically championed a kind of empirical metaphysics. It seeks to be porous with respect to scientific discovery: it is science which is to be the leading-edge of ontology. I have some limited sympathy with this approach with respect to certain theoretical endeavours, and agree that on the whole there is no need for a metaphysical grounding for science, provided by philosophy. However, I wonder quite how speculative realism will come to understand the status of its own metaphysical claims.

Alexei has raised the problem of normativity in this area: does a radical materialism have the resources to account for its own justification? We are all naturalists now — after a fashion, at least. But speculative realists have adopted a particularly strident form, which does not seem to be friendly to normativity. Just witness Ray Brassier’s Nihil Unbound. Can it understand, or sufficiently redescribe, the context in which it puts forward its own theory, such that it can allow that such a theory is meaningful, justifiable and truth-apt, whilst cleaving to a sparse materialist metaphysics which admits values, if it all, only in an anti-realist fashion? I will have more to say about this at a later date.


12 thoughts on “Realism and Correlationism: Kant and the Short Argument

  1. Tom,

    Let me join in praising your clear and informative post.

    Knowledge of the world in itself, as required by the realist, is denied to us because thinking is always present when thinking about the world in itself. However, this is only because here we are to understand knowledge of the world in itself as knowledge where thought is not present.

    There’s no doubt that “nothing is demonstrated by this argument.” On the other hand, to oppose it, the realist will need to oppose correlationist’s definition of ‘knowledge of the world in itself.’ It seems to me that the realist will need to hold that thought is present in knowledge of the world in itself, but present in a ‘transparent’ way. And it’s difficult to see how to make such a claim without begging the question.

    Perhaps the solution lies in this direction: the very distinction between thought and world is unintelligible, and therefore, any attempt to give expression to such a distinction results in nonsense.

    • I think something like ‘transparency’ is probably the way to go. One example of this strategy is McDowell’s identity conception of truth. This takes facts to be true thinkables. A thinkable is meant to belong “just as much to minds as to the world. It should not even seem that we need to choose a direction in which to read the claim of identity.” (‘The True Modesty of an Identity Conception of Truth’) Hopefully, I’ll be able to come back to this idea in a later post to flesh it out in less sketchy terms.

      • It sounds like McDowell is following in the long tradition of Aristotle (De Interpretatione, I, 1 notwithstanding) and Aquinas (and early Wittgenstein, of course). For the latter, there is a formal identity between knowner and known (e.g., De Anima, III, 7: “Actual knowledge is identical with its object”).

        My only concern is a Wittgensteinian one: there is no place one can take up ‘outside’ of thought to determine its relation to anything ‘external’ to it. How, for example, could we tell that the structure of the world is not a ‘shadow’ cast by the form of our thinking.

    • This is something McDowell accepts in his criticism of what he calls a ‘sideways-on’ perspective — and this claim looks something like a form of correlationism. One line of support in this respect is the appeal to Davidsonian arguments against scheme-content dualism. This would preserve the distinction between thought and world but without making it into a gulf that needs to be bridged. I like the conclusion here, but not sure yet whether I buy the arguments.

  2. I would want to add in specificity of my praise, when I read something like this,

    “Kant does not make this short argument. Ameriks traces this form of argument to Reinhold”

    and a follow up on the consequences, I cannot help but feel that philosophy has been furthered. There is a ponderous tendency in philosophy to broad brush a position, especially one that you are going to “fight”, and to simplify it so that others can join you in the opposition. There is much to be said for a genealogy of a idea or position, the way that it comes to be characterized into an elementary error. Something of this also is rather widely done with Descartes, who likely did not hold a Theatre of the Mind conception of Ideas (something attributed to him by Reid). There is ever the need to narrate in philosophy, and to position oneself against a stream of thinking that is mythologized upon a founding Father, the source of the error. The good that this post does is parse out the mythology from the genealogy, just who proposed what.

    While NN’s proposal,

    “Perhaps the solution lies in this direction: the very distinction between thought and world is unintelligible, and therefore, any attempt to give expression to such a distinction results in nonsense.”

    is interesting, I might suggest that indeed it does make sense to make distinctions between thought and the world (we do it all the time), the Realist/Anti-Realist debate, which side you find yourself on is a bit of a defintional game, not philosophically resolvable. But as we define the world, so too our minds are directed to different places, to notice different things. In a sense, each tradition is engaging in elaborate metaphors in frame.

  3. Really terrific post, Tom. You develop an exceedingly thoughtful and balanced presentation of the issues, such that I come away not knowing whether you’re standing with Kant and correlationism or the realist… Which, I think, means that you’re genuinely grappling with all dimensions of the issue, almost as a Hegel, Leibniz, or a Whitehead who sees the kernel of truth in every position while also recognizing that the falsehood arises from overstatement or exaggeration. In the backdrop of a very heated and often less than generous debate, thanks for this.

    • Thanks — glad you liked it.

      Where I stand with respect to correlationism and realism is not yet clear to me. Part of what I hope to do in these posts is show that there may be more to the positions which have been labelled ‘correlationist’ than first meets the eye. This might be seen as a defence of a more sophisticated correlationism, but it might also be seen as a defence of a realism which ‘passes through’ a narrow correlationism, such as Meillassoux advocates. Either way, I think there is space for a middle position which can address a problematic surrounding access, yet without remaining obsessed with the mind-world relation to the exclusion of all else. Of course, it’s one thing to aspire to such a middle way but another to show that it is viable.

  4. Pingback: Short Arguments Reconsidered « Now-Times

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s