The Perils of Representation

Following up yesterday’s the day before yesterday’s post, I should probably just add a few notes in response to Graham (see here). I’ll try and not let this spiral into a 7 message exchange though. I’ve always wanted one of those epic sounding monikers you have to put in quote marks, and I think Pete “the relentless” is the best suggestion I’ve come across so far (I take everything Graham says in good spirit). Before leaping in though, I should perhaps say a little bit about the way I approach philosophical disputes like this.

Graham does rightly note that I have a habit of trying to point out what I see as confusions or insufficiently precise uses of terminology, and this does reflect a bit of my more analytic background (although I’m sure if you ask those analyticians who know me I’d get accused of being wildly speculative!). This kind of nitpicking can come across as pedantic, or as labelling the other person as ‘confused’, ‘muddled’ or something similar. I’m trying to avoid this, as such things can be quite offensive, and if there’s one thing I’ve learned about the internet, it’s that it’s much easier to offend people by accident here (there’s so little way of modulating one’s tone that attempts to do so can wildly backfire), so it’s at the very least good practice to be careful with one’s words.

However, I’d like to defend my nitpicking to some extent. It’s all well and good to say that we should just take our (and by this I mean more than just myself and Graham) differences to be disagreements in all cases, and then to try and resolve them directly, but I think that it’s often the case that it’s not entirely clear what exactly we’re disagreeing about. Sometimes you have to do a bit of preparatory work in order to figure out where a given disagreement lies. It can be very frustrating for all involved, but it can pay us back for this frustration tenfold if done right. Of course, perspicuity can devolve into pedantry, and this can lead to obscuring what is genuinely important in a discussion. Perspicuity is a virtue, and as such, the Aristotelian in me strives for a golden mean. I don’t always find it, but I try regardless.

Brevity, on the other hand, isn’t a virtue I have any success with. This post will be quite long (8,000 words or so), largely because it includes the additional material hinted at in the last post, which is quite in depth. To Graham: don’t think I’m forcing you to ratchet up a current account deficit in our discussion. You’re a much busier man than I, and you can feel free to get back to me whenever you like (or not to at all). As our Australian friends say: no worries.

1. Reductionism and Materialism Revisited

Bearing everything just said in mind, I’m going to try and put the main points of the last post in more careful terms. Only some of what I said about reductionism and materialism was really directed at Graham, and I should have been more clear about this. I’m just somewhat frustrated by the way these terms can be used in the blogosphere, and I try to take any opportunity I can to give them better definitions, with the hope that it raises the overall quality of debate on these issues. As an additional point, I’m assuming that what Graham called ground floor materialism is in all important respects equivalent to undermining, and that first floor materialism is similarly equivalent to overmining (even though I understand that there can be non-materialistic forms of both). As such, I’ll just substitute them here. If I’ve gotten this wrong, I’d be happy to have it pointed out.

With regard to reductionism, I just wanted to point out that this term usually tends to be used to indicate a certain methodological, and ultimately metaphysical position taken up by certain paradigms in the natural sciences. When I say that the presocratics’ metaphysics can be characterised independently of any relation to the natural sciences, what I’m saying is that we can do so without discussing any of the contemporary debates about the relations between the domains of the different sciences, and specifically about the derivability of the laws of one domain from another. Yes, there is some sense in which Anaxamander and Parmenides are scientists, but their metaphysics can be discussed independently of most of the major metaphysical debates surrounding modern science (if not all of them). In essence, I think the sense of reductionism as used by Ladyman and Ross, where it is taken to be the antonym of emergentism, is the right way to use the word, and that we should distinguish it from what Graham is trying to get at with his notion of undermining objects (even if these two things sometimes overlap). Importantly, what I was calling strong reductionism, exemplified by physicalism, seems to undermine objects in Graham’s sense by default, in virtue of positing a fundamental substratum of reality to which all others may be reduced.

It’s quite likely that Graham didn’t use the word ‘reduction’ in relation to undermining at Dundee, and so I apologise for reading it in to his paper (he’s right that the Bristol talk probably coloured my experience of the Dundee one). I suppose my real motivation for trying to separate out reductionism from undermining is that it has been used in this sense (i.e., as an antonym of emergentism) by others in the same context as Graham’s critique of undermining (I’m thinking specifically of Levi here, though I can’t find a reference off hand). As Graham points out in his critique of Ladyman and Ross, even card carrying emergentists can be underminers, and so we need to make sure we properly delineate the debate between reductionists and emergentists on the one hand, and underminers, overminers, and OOO enthusiasts on the other. The debates might overlap, but, if we want to pinpoint exactly where they do, we need to understand their broader divergence.

With regard to materialism, I also have to apologise for not making clear that I understood entirely why Graham identified Zizek and Badiou as materialists (i.e., in terms of their relation to the Marxist tradition), despite his initial reticence to do so. I should have really split my point in two: 1) There is in my opinion a proper way of identifying materialism, and this involves a core metaphysical commitment to the manifestation of all things within some material substrate, i.e., to some notion of matter (and a corresponding notion of form), and 2) That even though I think Graham would be quite entitled to count all such materialisms as undermining objects, I don’t believe that the specific arguments he offered against undermining in his Dundee paper (esp. the one’s levelled at Ladyman and Ross) necessarily work against them all.

On the first point, my terminology policing wasn’t really meant to be directed at Graham, but at the more broad discussions of materialism that have been inspired by his work (particularly Levi and Ian Bogost), as well as at the Zizekian/Badiouian materialist axis. I will mention a specific post this time, namely, Ian Bogost’s post on materialisms, which I didn’t get to respond to at the time. Ian generally writes very interesting stuff, but I find it hard not disagree with his definition of materialism more strongly. As far as I’m concerned, if you don’t think that there’s some single sense in which everything is made of matter, then you’re not a materialist. This isn’t limited to OOOntologists (?) though. When I was at the SR II event, I asked the panel whether materialism needed to develop a concept of matter, and the materialists present (Ray Brassier, Iain Grant and Alberto Toscano) all seemed to agree that it was unimportant. Against this, I’m claiming that materialism without matter makes no sense. This doesn’t mean that one needs to be a physicalist, i.e., that one needs to take one’s concept of matter wholesale from the sciences. It’s perfectly possible to have an original metaphysical conception of the relation between matter and form, and this is precisely where I think Deleuze is exemplary (even if he doesn’t necessary address his own work in these terms).

Here’s a quick rule of thumb of the style of Graham’s fire and cotton criteria for idealism: if you think that numbers are entities, then you’re not a materialist. This is because there’s no good sense in which they are manifest in a substratum, and no good sense in which they are composed out of parts (well, maybe David Lewis’ freaky set-theory as mereology approach (he’s almost the anti-Badiou here) can manage this, but I still don’t think you’d call it materialism). Of course, this is necessary but not sufficient. One can deny the existence of numbers and nonetheless be something other than a materialist (e.g., a correlationist). Nonetheless, I think it’s a decent rule of thumb. For instance, by this criteria, I’m a materialist and Graham is not. There’s a straightforward disagreement for you!

On the second point, although I haven’t read the Zagreb paper, and although I think we can fairly easily point to which positions are underminers in Graham’s sense, I’m still not entirely sure we’ve got a completely clear idea of what binds them together. Indeed, they all in some sense ‘reduce’ objects to a deeper metaphysical principle or fundamental substratum. However, what I want to know is: i) what exactly is being reduced in each case, and ii) what exactly this ‘reduction’ (or whatever better word we can come up with) consists in.

Taking the first point, we can’t say that what is being reduced is Graham’s radically discrete and totally withdrawn objects, because this would basically amount to defining any position which disagrees with Graham as overmining/undermining objects (and I’m certainly not claiming that Graham thinks this). We thus need a weaker sense of object to describe what is being undermined by Grant, overmined by Meillassoux, and which Graham takes himself to be genuinely elucidating or explaining with his full blooded conception of objects. I suspect Graham would appeal to some stripped down, phenomenological notion of the object here. This would then link us back to the ongoing debate over the viability of phenomenology as a method for ontology (see here and here). The concern I have here parallels the problems I have with Levi’s argument that eliminativism turns certain entities into epiphenomena (here).

Taking the second point, I posed Graham a sort of challenge, which was to say “what Ladyman and Ross, Anaximander, Ian Hamilton Grant, and others have in common”. Graham has responded to this challenge, but I think it’s an inadequate response. To quote it in full:-

Furthermore, I don’t think it’s all that difficult to see what “Ladyman and Ross, Anaximander, and Iain Hamilton Grant” have in common. That’s pretty clear, actually: all are committed to a real, but a real that is not carved up into distinct objects. In fact, that’s what drives all three of these philosophies. For Ladyman and Ross the world is a relational structure and objects are replaced by real patterns that exist pragmatically (while also somehow being “real,” in thoroughly unconvincing fashion) at different scales. Anaximander’s apeiron is broken into pieces historically through rapid rotation and vibration. Grant’s real is a productive force that breaks into object only through retardations. All three philosophies are committed to a mind-independent reality devoid of what we know as individuals. They seem to have a tremendous amount in common.

The problem with this response is that it doesn’t address the “others” I was referring to. This seems like a little gripe, but what I was after was a conception of what binds together all of the positions Graham would classify as undermining objects, i.e., of precisely what their ‘reduction’ of objects consists in. As Graham notes in an earlier post, this broader category includes those who claim that “objects are surface-effects, folk manifest images, mere aggregates, etc.”. For instance, I’ve always thought that Graham would count physicalism as a species of undermining, and it seems that, from the above definition, that he would. However, physicalism is perfectly fine with a discrete cosmos. It does ‘reduce’ objects to a fundamental substratum, and thus in some way turn them into aggregates of fundamental physical entities, but this isn’t an apeiron, in the sense of some primal metaphysical continuum out of which individual objects would emerge. I imagine there’s a lot of further arguments we could have over Graham’s claims about the discrete and the continuous (I don’t buy his claim that DeLanda, following Deleuze, is cheating when he says there is such a thing as a heterogeneous continuum), but those can wait for now. The important point here is just that we need to know exactly what’s involved in the ‘reduction’ common to all undermining positions. I’m not claiming that Graham can’t give an account of this (or even that he hasn’t, I might have just missed it), just that such an account would help to pinpoint exactly where the important disagreements lie.

The final point to make is just to reiterate that the arguments Graham levelled at Ladyman and Ross, although good arguments, don’t really hit thinkers like Deleuze, who would nonetheless count both as a materialist and an underminer. This is because Deleuze’s position is not based on epistemological considerations about the relation between science and ontology (which isn’t to say he doesn’t say anything about science). Of course, Graham would want to deploy his arguments about discreteness against Deleuze, so I’m not claiming he’s without ammunition here, just that it’s got to be of a different sort. All I’m trying to do is delineate a little space where there is still a debate to be had. Now, there is a further argument of Graham’s remaining, namely, the one I hinted at at the end of my last post.

For those of you just interested in the most recent back and forth, you can stop now (this includes you Graham!). It gets more technical from here on in.

2. Representations and Objects

Here is the follow up to that post, explaining my problems with the additional argument in Graham’s paper. To repeat, this argument is meant to refute all realisms that do not subscribe to Graham’s radical conception of the real as that which is entirely withdrawn from all possible experience of it. Now, I’m not claiming that this argument is anything like the central pillar of Graham’s work, but he has deployed it a couple times, once on his blog in a response to one of my earlier posts. I’ll quote it from the relevant post:-

And people who claim that direct contact is possible are going to have to claim that one thing can adequately grasp another through contact. But this is impossible: no amount of knowledge about a tree, even if it were God’s own knowledge, is itself a tree.

The basic principle put forward here is: for knowledge of an object to be adequate, it would have to be the object itself. The inference that Graham seems to make here is that, because it is absurd for any knowledge of a thing to become the thing itself, it must therefore be equally absurd that we could have such adequate knowledge. Therefore, we must hold that there is something about all objects which is completely withdrawn from all possible knowledge, lest we be committed to the possibility of such an absurdity. It should be noted that this is not a straightforward reductio ad absurdum inference, as it does not identify a genuine contradiction that must be eliminated. There is one way of construing the argument in such a way as to do this, which is to say that in the absurd case the knowledge would in some sense have to be both numerically identical with the thing and distinct from it at the same time. To make this stick, one has to maintain that knowing always involves some essential difference between itself and its object. The case of adequate knowledge would then be a matter of straightforward contradiction, and it could be eliminated by the usual reductio inference. We’ll keep this possibility in mind for later.

I think this is a bad argument in either form. I suspect many will have the same intuition. However, showing precisely why this is a bad argument is actually quite a delicate matter. It involves showing precisely what is wrong with the basic principle posited above. The problem centres around the notions of sameness, identity, difference and representation. I should mention in advance that I am not going to argue that we can have a completely adequate knowledge of things, I’m just going to argue that Graham doesn’t effectively show that we cannot, let alone that this implies that objects have the kind of withdrawn molten core he is after.

First of all, we have to recognise that what is meant by adequate knowledge involves a certain kind of completeness. So, my knowledge of my sister’s hair colour is not adequate knowledge of her, not because it is incorrect, but because it is not a complete knowledge of my sister. When Graham talks about knowing an object in this way, it seems that he is not talking about isolated individual bits of knowledge about a given object (e.g., my sister’s hair colour), but rather the totality of our knowledge about it (e.g., everything I know about my sister). We can imagine this as a kind of conjunction of all the particular things we know about the object. The basic principle is then that for this conjunction to be a complete knowledge of the thing, it would have to be identical with the thing. The issue is then the status of incomplete knowledge, and whether its correctness has any interesting relation to the notions of identity and difference.

The pertinent question is: precisely what are we saying is either identical or different with the object of knowledge? We could say, the knowing, the knowledge, or the grasp, but none of this is very informative, as we have no good criteria of identity and distinction for these. Understanding the sense in which knowing is representing is very helpful here.

In talking about our knowledge, knowing, or grasp of the object, we are in fact talking about a representation. Indeed, we are talking about specifically discursive representation. Regardless of whether one thinks the content of such representations needs to be understood in terms of the normative structures of discourse (as I do), or in terms of the phenomenological structures of consciousness (as I think Graham does), one must nonetheless recognise that we are dealing with something like propositionally articulated contents (i.e., of the form ‘x is F’; e.g., ‘Pete’s sister has blonde hair’). I can know that my sister has blonde hair, and I can express this by saying that my sister has blond hair, and the content of this knowledge is the same as the content of my assertion, insofar as the same proposition is involved in each case. Both the knowing and the saying are representations, and they represent the same thing, as being the same way.

A related point here is that there are two ways of understanding what I know, if I know that ‘my sister is in Newcastle’: as a matter of which object I know (my sister), and as a matter of what I know about it (that she is in Newcastle). These articulate the two inseparable dimensions of representational content: the intentional and the predicative. One can’t have one without the other. If my father knows that ‘Pete’s brother is in Newcastle’, there may be some more limited sense in which he knows the same thing about my brother that I know about my sister, but in the proper sense we do not know the same thing. The contents of our knowledge are different precisely insofar as they are intentionally directed at different objects, and thus representational content must be intrinsically intentional.

Moving on, let us draw an additional distinction between the vehicle and the content of a representation. To give an example of this difference, it is the difference between the saying and what is said in language. One can actually make more fine grained distinctions here, as one might distinguish between the act of assertion (or the utterance), the sentence which is tokened in this act, and the proposition which is expressed by the use of the sentence (e.g., there is my concrete act of asserting ‘Snow is White’, the type-sentence which can be asserted in other acts of assertion, both vocal and non-vocal, and the proposition which can be expressed by different sentences, such as ‘Schnee ist Weiss’). This is a threefold distinction between the pragmatic, syntactic and semantic dimensions of linguistic expression, respectively. However, we don’t really need this threefold distinction, all we need is the distinction between the semantic content and the vehicle which bears this content, be it an act, a sentence, a brain state, or whatever other entity or state of an entity (or aspect of an entity) one wishes to pick out. This is particularly important given that Graham is unlikely to endorse the primarily linguistic account of semantic content that I do.

Nonetheless, Graham should accept that there is some distinction between the content of my knowing and the knowing itself, be it conceived as an act of knowing, or as some other aspect of myself. This should be obvious insofar as two people can know (or at least claim to know) the same thing. If I claim that ‘Snow is White’, someone else must be able to utter the same sentence and mean exactly the same thing as I do, they should be able to express the same proposition. This holds for the utterance of suitably different sentences in the same language, and suitably inter-translatible sentences in other languages (there also the possibility that the same sentence can be used to express different contents, but we’ll leave this to one side). Similarly, if myself and my brother use litmus paper to test whether a liquid is acidic, and we both become aware of the positive result, then it seems sensible to say that we both thereby know the same thing, namely that ‘this liquid is an acid’. Our knowledge has the same content. Moreover, this should in principle extend to ‘knowing an object’ in the conjunctive sense we outlined above. If it is in principle possible for myself and my brother to share each given bit of knowledge, then it should in principle be possible to share a larger conjunction of such knowledge.

After that little aside, we can now genuinely pose the question of what is supposed to be identical or different from the object of knowledge in knowing: is it the vehicle of representation, however this is construed, or is it the content?

I think it should be obvious that it can’t be the vehicle. This is because the characteristics of the vehicle can only be relevant to the correctness (and thereby the adequacy) of the knowledge insofar as they play a role in specifying its content. The difference between one assertion made in spoken English and another made in written English does not make any difference to their content. If one is true the other must be, and so this difference can play no role in making it true or false. Or, to take an example that can be understood non-linguistically, we can at least say that the knowledge of my sister would be adequate whether or not it is my knowledge or my father’s knowledge, which definitely amounts to some difference in the vehicle of the knowing. In truth, it is a trivial fact that the vehicle of representation is numerically distinct from what is represented, in all but artificially constructed self-referential cases (e.g. ‘this assertion is true’), and even in those cases identity is certainly not equivalent to correctness (see ‘this assertion is false’).

If any sense is to be made of the basic principle, then it must be interpreted as relating to the representational content of knowledge. So we must understand it as saying that the adequacy/inadequacy of our knowledge has something to do with the identity/difference between the knowledge’s representational content and its object. This is still an odd claim, but it’s less obviously problematic than the vehicle option. It’s real oddness lies in the fact that it’s not clear how we are to understand anything like an identity claim between a representational content and something which isn’t itself a representational content. As we’ve seen, we can easily understand two representations having the same content. Indeed, that two different representations can have the same content is constitutive of the very notion of representational content. The idea of sameness and difference of content is straightforward. We can even to some extent understand the idea of a representational content being different from something that isn’t a content, insofar as we can understand that the vehicle and the content are distinct. However, what it would be for a content to be identical to a non-content, specifically its intentional object, is unclear. This is analogous to the old chestnut regarding the claim that Julius Caesar is identical to the number 9. We can easily understand the claim that Caesar and the number 9 are different, but the converse seems to lack any good sense.

Of course, Graham thinks the idea of knowledge actually being an object is absurd, but this doesn’t matter. For his argument to work we need to be able to provide a coherent interpretation of the basic principle, so that we can deploy it in something like a reductio of the possibility of adequate knowledge. As such, we must be able to provide a coherent interpretation of how identity and difference are related to correctness and incorrectness, and by extension adequacy and inadequacy.

3. Representation and In-Itselfness

Having set the scene quite extensively, I think we can derive the basic principle from  series of other ideas:-

i) For any individual bit of knowledge of an object (such as knowledge of my sister’s location, or her height, weight, etc.) to be correct, the object must in some sense be the same as it is represented as being. For instance, my knowledge of my sister’s hair colour represents it as being blonde, and my sister’s actual hair colour is blonde. My sister as represented, and my sister in herself, are the same in the relevant respect.

ii) For our ‘knowledge of an object’ to be correct, every bit of knowledge that it contains must be correct.

iii) For our ‘knowledge of an object’ to be adequate, it must be both correct and exhaustive.

iv) Our ‘knowledge of an object’ is exhaustive if there is no feature of the object that does not correspond to some individual knowledge of it.

v) Thus, for our ‘knowledge of an object’ to be adequate, the object as represented and the object in itself must be the same in every respect (from i-iv).

vi) By the principle of the identity of indiscernibles we may thus infer that for our ‘knowledge of an object’ to be adequate, the object as represented and the object in itself must be identical (from v and PII).

I think this is a plausible reading of the thought underlying the basic principle. However, it still leaves us with two issues: 1) What exactly is the object as represented, and what is it to say that it is numerically identical with something? 2) Is there a structural difference between the object as represented and the object in itself from which we could derive and contradiction, and thereby carry out a genuine reductio ad absurdum of the possibility of adequate knowledge?

I’m going to take the second of these questions first, as I think doing so sheds light on the former. To recap what was said earlier, we suggested that there were two forms of the argument against the possibility of adequate knowledge on the basis of the basic principle. The weaker form takes the idea of the identity between knowing and object to be absurd, but not in a straightforwardly contradictory way. The stronger form finds a genuine contradiction. In each case, the absurdity is used to leverage a rejection of the idea of adequate knowledge, albeit only in the form of a strict reductio in the latter case.

Now, if we are to pursue the weaker form of argument, the absurdity produced needs to be something quite serious and undesirable. If we hadn’t differentiated between the vehicle and content of representation, we might be tempted to see an obvious absurdity, such as the idea that a book which adequately represents a tree would have to be the tree, or that a person who adequately knows his sister would have to be his sister, or at least contain his sister as a part of him. Because we know that the identity can’t be a question of the vehicle of representation, we know that these kinds of absurdities don’t follow from the basic principle. However, since we haven’t answered the first question just posed, neither do we have a good idea about what other non-contradictory absurdities would follow from it. So, in the absence of a compelling weak form of absurdity, I’m going to try and pursue the strong form.

There is indeed a structural difference between the object as represented and the object in itself, in virtue of the structure of representation as such. This difference constitutes the in-itselfness of the object represented, insofar as it indicates the independence of the object from our representations of it. The idea (call it (vii)) is that representation involves the necessary possibility of error, or incorrectness. In order for this to be the case, it must be possible that the object as represented by us (or the object for-us) can be different in some respects from the object in-itself. Put in a different way, the possibility of error requires that the way the object is in itself must not be determined by the way that we represent it. If the object is represented correctly it must not be because the representation itself sets the standard by which it is to be judged. That the object sets an independent standard in terms of which representations of it can be judged correct or incorrect is just what its in-itselfness consists in.

One might think that we can now combine this with the basic principle to produce a contradiction, and bob’s your uncle we have a reductio ad absurdum. However, it’s not clear that the sense in which the object for-us and the object in-itself are different here, is the same sense in which they are the same in premise (i) above. For any given individual bit of knowledge, the way it represents the object can be the same as the way the object is (i.e., it can be correct) and nonetheless it could still be possible that this hadn’t been the case (i.e., that the knowledge could have misrepresented the object), but this possibility doesn’t seem to have anything to do with the fact that this knowledge is incomplete. Yes, it is the case that the difference between the object for-us and the object in-itself makes error possible in this and other cases (if they were identical, error would not be possible), but it’s not clear how the completeness of knowledge would somehow render error impossible. If we can make sense of us having individual bits of correct knowledge about an object, then we have to also be able to make sense of how having an exhaustive set of such correct knowledge would invalidate the possibility of incorrectness, of how it would somehow negate the independence of the object, or remove its in-itselfness. What is it about adding that last bit of information which puts us over the edge? If we can’t make sense of this, then rather than rejecting the possibility of adequate knowledge, we should reject (i), or reinterpret it in a way that makes it impossible to combine it with the principle of the identity of indiscernibles. Indeed, I think such a reinterpretation also involves us reinterpreting what this structural difference between representation and its object (vii) consists in.

This returns us to the first question posed above: What exactly is the object as represented or the object for-us, and what is it to say that it is numerically identical with something?

To tackle this question is also to address the issues we raised earlier about the sense in which a representational content can be identical with something that is not a representational content. The notion of the object as represented, or the object for-us, is just another way of parsing this notion of representational content, albeit one that lends itself better to the kind of totalisation involved in the idea of ‘knowledge of an object’. Whereas in the case of content, we have to talk about a kind of slap dash conjunction of various different propositional contents into a single representation of the thing, in talking of the object as represented, we are already talking about something that is in principle singular, which is simply more or less detailed. However, what this way of talking does is to make it seem as if we are in fact talking about two different kinds of objects: the object-for-us and the object-in-itself. Once we start thinking in this way, the whole problem of how a representational content could be identical with its object seems to slip away, and with this, the necessity of denying the possibility of adequate knowledge in order to maintain the possibility or error as such becomes apparent. I think the parallels between Graham’s distinction between sensuous objects and real objects should be fairly apparent here, though I’m not going to elaborate on it, because it isn’t quite exact.

My claim is that it is this elicit transition from talk of representational contents (or the content of intentional acts if you’re a phenomenologist) to talk of a special kind of object that is the problematic premise here, and that, because of our inability to make sense of the relation between the possibility of error and the impossibility of an exhaustively complete knowledge, this is the implicit premise (contained in the formulation of (i) above) that we should reject. Representation (and knowledge more specifically) does not aim at any kind of intermediary object between it and the object in-itself, it just aims directly at the object in-itself. The content of our claims is intrinsically about something independent of itself, even if there is the constitutive possibility that it misrepresents it.

4. Identity, Difference and Object-Types

To make this clear, lets just unpack (i) a bit. In order for the inference codified in (vi) to work, it has to say that when we say that some object has a given property, we are in some way appealing to an object-for-us that has this property. For the principle of the identity of indiscernibles to work here, the sense in which such an object-for-us has the property must be the same in which the object-in-itself has the property, if our representation is correct. This is ontologically spurious unless we appeal to some kind of Meinongianism (and it’s no coincidence that Meinong is famous for his theory of objects), wherein objects-for-us may have such properties in the same sense as objects-in-themselves (in Meinongian terms, they both have Sosein), whereas they differ insofar as only the objects-in-themselves exist, or are in some sense real. This would amount to presupposing some kind of ontological difference between the two kinds of object. Now, on this basis, one could legitimately say that there is something specific that ‘knowledge of an object’ can never grasp, namely, its existence, reality, or whatever ontological characteristic it is that distinguishes the object-for-us from the object-in-itself. This is fair enough, but it amounts to grounding the epistemological thesis that there cannot be adequate knowledge of objects-in-themselves in a very strong ontological thesis. If this were Graham’s approach, then it would be impossible for him to use the argument from adequacy to justify his ontological conception of objects.

The way to reformulate (i) and (vii) so as to avoid such ontological commitments is to pay a bit more attention to the structure of identity. We earlier noted a bit of an asymmetry between certain kinds of claims about identity and difference. Paradigmatically, it seems easy to understand the claim that Julius Caesar is not the number 9, but not so easy to understand the claim that Julius Caesar is the number 9. Similarly, we can easily understand the claim that ‘the direction in which I am pointing my finger’ is not Hamlet’s left foot, but the converse is much harder to make any sense of. The way to make sense of this asymmetry is to note that we have different kinds of identity criteria for different kinds of intentional objects. We have very definite criteria governing what it is for two numbers to be identical, such that we can easily decide whether ‘the number of the planets’ is identical with the number 9, and similarly for ‘the square of 3’. We have entirely different kinds of criteria for working out whether the direction in which I am pointing my finger is the same direction you are pointing your finger, and so on. The criteria we have for identifying fictional entities (does Clark Kent = Superman?) and worldly entities are again different, although I would say that these cases are far less cut and dry, and involve their own philosophical problems. It’s also important to point out that this idea is not equivalent to Peter Geach’s claim about the relativity of identity, wherein identity always involves some kind of sortal term (i.e., x is the same F as y; e.g., this sculpture is the same clay as my previous one, but it is a different sculpture). This is because it is only certain specific meta-sortals which provide criteria of identity (number, proposition, fictional entity, etc.), not all sortals. In essence, although this picture doesn’t advocate absolute identity (not to be taken in the Hegelian sense), it doesn’t completely relativise it either.

The idea then is that the asymmetry arises from the fact that claims about the difference between two objects that are of different types make sense because there can be nothing like identity outside of the context of the criteria of identity that constitute each type, whereas the same fact prevents the corresponding identity claim from making any sense. On this basis, we can understand the kind of structural difference exhibited in (vii) differently. The fact that the content of our representations and their objects must be numerically distinct can be conceived as one of these asymmetrical difference claims. We have a set of criteria for individuating the contents of our representations which are distinct from whatever criteria we have for differentiating what we are talking about. The exception to this is when we make claims about the representational contents of other representations, such as claims about the meaning of other people’s claims, or construct bizarre self-referential claims like the liar paradox. To make a nice analogy, whether or not we are saying the same thing (expressing the same content) is on par with whether we are pointing our fingers in the same direction. Propositions are objects only in the same sense that directions are.

5. Epistemology, Ontology, and Deontology

At this point, it is likely that many will want to accuse me of grounding my own interpretation of these epistemological issues on ontological assumptions about differences in types of entities, making me no better than the hypothetical Meinongian I discussed above (hypothetical Meinongians are the best kind!). It will also strike anyone who knows how passionate I am about the Deleuzian commitment to the univocity of Being (see here) as completely incongruous that I am appealing to such differences in type at all. Don’t worry, I’m still all for univocity, and I’m not making elicit ontological distinctions. Rather, this variety of types of object is generated by the pragmatics underlying the structure of quantification and existential commitment. I’m not going to go into this in too much detail, but I’m getting most of it straight out of Brandom’s account of existence (and types of existence) in Chapter 7 of Making It Explicit. All of this relates to the comments I’ve made before about the distinction between generic existence and real existence (here).

To reiterate this quickly, we have a formal notion of existence (and of identity, predication, essence, truth, etc.) which underlies the structure of our talk about everything, be it fictional entities, numbers, propositions, or black holes. These various formal notions are generic, in that they don’t discriminate between the different kinds of things we talk about. Taken together, they constitute our pre-ontological understanding of Being, or the formal structure of the intentional object as such, or what Kant would call the transcendental object or the object=x. The content of these formal notions is implicit in our grasp of the fundamental norms of rationality (see here and here), which are the rules governing discourse. Doing ontology, or answering the question of the meaning of Being, is a matter of giving content to this formal understanding of Being. In brief, this means moving from our pre-ontological understanding to an understanding of Being in-itself, which means getting at the objective structure of Being. This is a two stage process, starting with the delineation of the formal structure of objectivity, and then proceeding to the objective structure of objectivity (explained in more detail here and here). The relevant point is that this first stage essentially amounts to the development of a formal concept of reality, and the application of this concept to the various generic notions. This produces a purely formal distinction between real beings and pseudo-beings, real properties and pseudo-properties, real essence and pseudo-essence, etc. The formality of this distinction means that it is still not yet ontological, but merely preparatory for ontology. Doing ontology amounts to providing a systematic account of what real beings, their properties and essences really are (e.g., material processes situated in a spatio-temporal causal-mereological nexus, actualising virtual capacities and tendencies that are localised in various overlapping self-stabilising continuities; see here and here). That was a lot of information to cram into a paragraph.

The essential point is that we have various kinds of norms that set up stable ways of talking about kinds of entities that in truth have no real Being (i.e., they are pseudo-beings). Fictional entities are the best example of this, as the way they are is not entirely independent of the way people represent them as being. Yes, whether Thor has long hair is independent of whether you or I think he does, and thus it is possible for us to be wrong, but it is not independent of the norms for talking about Thor established and maintained by the Norse people, through the ways they represented him across their history. Criteria of identity are structural norms that underlie particular kinds of discourse, and in doing so establish particular kinds of pseudo-entities (e.g., norms, numbers, propositions, fictional entities, etc.). Where I differ from Brandom is that I don’t think that there are a priori norms governing the way we individuate real entities. He would want to stipulate criteria for individuating physical entities in terms of spatio-temporal relations, whereas I think that working out such criteria is a matter for ontology. Specifically, it is a matter for the move from generic identity to real identity (even if my own intuitions are that real identity ultimately turns out to be little more than processual continuity).

So, I’m not basing epistemological considerations on ontological assumptions, but rather upon deontological ones about the normative structure of discursive representation (as far as I am concerned deontology is first philosophy). The essential problem is that the original interpretations of (i) and (vii) given above suggest an ontological reading, whereas I think they should be described in normative terms.

As such, (vii) should be cashed out in terms of the way in which we give up authority over the correctness of our claims in such a way that we grant something like authority to the object itself, as mediated by empirical observation or the authority of others (e.g., the Norse people in the case of Thor), thereby making error possible. I call this the withdrawal of authority. My gamble is that it is only in discourses structured around empirical observation (e.g., the natural sciences) that authority is completely withdrawn from everyone, in which a kind of formal objectivity has been attained, that we actually deal with real entities, which is to say those entities that are genuinely in-themselves. This idea is what underlies the first step of my two stage approach to ontology outlined above. Objectivity, in-itselfness, reality, and Being are concepts that are thoroughly intertwined on my view.

Things are trickier in the case of (i), where it is effectively impossible to talk about the sameness of the object and our representation of it in any literal way without lapsing into some kind of ontological interpretation of the matter. This is the big problem in the history of the correspondence theory of truth, which Heidegger points out so deftly in §44 of Being and Time. The only way out is to paraphrase this kind of talk as being about relations between contents, rather than between contents and things. This has been worked out in many different ways in the analytic tradition, in the form of deflationary theories of truth. As will come as no surprise, I pretty much accept Brandom’s prosentential account of the truth predicate (which explains what it is to say that a claim or representation is correct/true; see MIE Chapter 5), with the qualification that I think it only gives us the generic structure of truth. I actually think there is a taxonomy of different forms of truth based on the various ways the withdrawal of authority (which in making error possible, also makes truth possible) can be set up, at least amounting to a distinction between objective and non-objective truth. A fuller description of any of these issues can wait for another time.

6. Conclusion: Varieties of Excess

In conclusion, I think I’ve provided some good reasons why we should reject Graham’s general argument against forms of realism that claim we can know the real. I think the most consistent account of the impossibility of adequate knowledge that Graham can provide would be based upon his ontology, and thus would be unable to be used to justify it against other forms of realism. Of course, he might have better arguments for the claim that adequate knowledge is impossible (I’m pretty sure he does), but I think that this particular one doesn’t work. I don’t think this is that surprising given that Graham really thinks that we can’t have any knowledge of the real object, let alone an adequate one. It seems to me that he’s better off trying to establish this stronger claim directly.

All of this isn’t to say that I actually think we can have complete knowledge of the real. I don’t. In fact, I think this claim is fairly mainstream in modern continental philosophy, ‘speculative realism’ included. The problem is always how to articulate this intuition, and justify it, without collapsing into some form of correlationism. Where I agree with Graham is that if we’re going to make this kind of claim, it’s going to have ontological ramifications, not merely epistemological ones. If there is something about Reality as it is in itself which is in excess of us, then it’s going to have to be in excess of everything. You can’t deny this and accept the strong principle of the univocity of Being that I advocate. Where I disagree with Graham is how precisely one should go about articulating this excess and interpreting it ontologically. In accordance with my wider methodology, I think we’ve got to give an account of the structure of representation first, and if we uncover some kind of structural deficit in the structure of representation (which I believe we do), then we can use this (in conjunction with univocity) as a starting point for understanding the real structure of entities. However, I think there are a lot of methodological hoops we’ve got to jump through to get to this point. I also think that the kind of excess we find is somewhat different to Graham’s austere segregation of the sensuous and the real; something akin to an excess of the Real as a whole over each individual entity, out of which a more limited excess of entities in relation to eachother emerges. You might call this a distinction between a local excess (Graham) and a global excess (me, Deleuze, and possibly Ray Brassier).

Anyway, time to wrap this monstrosity up. There’s a lot here to read, but writing it has been most helpful. Sometimes trying to pick apart why you think something is wrong can be more productive than analysing something you think is right. I suppose one could write a whole book titled In Praise of Disagreement. It’d be well worth it.