Response to Levi (part 2)

Continuing the post from yesterday, here are sections 4-6 of the response, dealing with the place of knowledge in OOO, the points of convergence and divergence between myself, Levi and Graham, and my criticisms of Levi’s accounts of meaning and knowledge. Levi already has a brief counter-response up (here). I don’t want to address his counter-points in great detail here, as I’m still finishing the final part of the main response that will deal with some of these issues. I would like to pick up on one of them though, as I think it can be addressed fairly quickly.

Levi has misinterpreted my challenge to his notion of translation. He thinks that my claim is something like: we must in each particular case be able know what is being translated in order for the notion of translation to make sense. He then claims that this argument illegitimately places epistemological criteria on a metaphysical point, and that the whole point of translation is that we can’t know what something is like prior to translation. This is not the claim I made though. My claim was that we must have a general understanding of what is being translated in order for the notion of translation to make sense. We must be able to make sense of the very idea of direct contact between entities in order to make sense of the very idea that they can only encounter one another indirectly. I take the last post to have shown why the ‘translation’ of perturbations into information, and of information into system states, doesn’t provide us with the resources to think such directness in general, and thus why all talk of indirect access is at best metaphorical. This has nothing to do with knowledge, and everything to do with the coherence of metaphysical concepts.

Anyway, onto the main course…

4. Knowledge and Relations

I’ve now made clear my problems with the notion of translation, and I’ve also explicated the notion of withdrawal to some extent. There is more still to be said about the latter, and I have yet to tackle any of the issues regarding the notion of knowledge properly. However, in order to deal with these things, it will be helpful to further outline the common ground I share with Graham and Levi. This is somewhat complicated, but I’ll do my best to make it clear.

In my comments on Jon’s blog (here) and in the above quotes from my email to Graham, I’ve mentioned that I accept some form of what Graham has called the R7 thesis (which is meant to be an addition to Braver’s set of realist positions). This is the idea that the relationship between man and world should be understood in the same metaphysical terms as the relations between other things within the world. On Jon’s blog I suggested that we distinguish weak and strong forms of this thesis. The weak R7 thesis simply holds that man should not have any kind of privileged metaphysical status which creates a metaphysical distinction between the relations we enter into with objects from the relations other objects enter into with one another. The strong R7 thesis holds that knowledge must be understood in metaphysical terms, and thus that relations of knowing must be understood in the same terms as all other relations. I endorse the weak form of the thesis, but deny the strong form. This stems from the commitment to the strong version of the principle of univocity I inherit from Deleuze (here).

Before clarifying this further, I’ve got to address a terminological issue that’s been bugging me. I find that in many of the debates surrounding OOO, regarding the thesis that objects are independent of their relations to other objects, that the term ‘relation’ gets used in too narrow a fashion. Of course, it’s obvious that knowledge, perception, and causation can be understood in relational terms: ‘x knows y’ (or preferably, ‘x knows y about z’, or ‘x knows that P’), ‘x perceives y’ (which is distinct from ‘x perceives that p’), ‘x causes y’ (where x and y are events). Despite the differences between Levi and Graham, I think that OOO collapses these three different relations into one metaphysical category. Following phenomenology, knowledge gets understood primarily in terms of perceptual knowledge (i.e., perceiving that), and then perception and causation get collapsed into one another, insofar as all perception involves causation. All three can thereby be understood in terms of the excessive dimension withdrawal (e.g., in terms of indirectness). This is the move which underlines their commitment to the strong R7 thesis.

Now, I have a problem with the priority given to perceptual knowledge here, insofar as it tends to ignore knowledge gained by non-perceptual means (such as testimony and inference), as it seems like there is nothing analogous to a causal interaction with the object of knowledge in these cases. Leaving this issue aside though, what’s been bugging me is that there are relations that aren’t amenable to description in these kinds of terms. For instance, their are relations such as ‘x is to the left of y’, ‘x is taller than y’, and ‘x is the brother of y’ that don’t seem to involve anything like an interaction between the related objects. If this is the case, then it makes no sense to talk about the excess of one object over the other in terms of either knowledge or access. I’m not here denying that objects are independent of these relations – it seems relatively sensible to maintain that the fact that I’m to the left of my bookcase does not say anything important about me – just that the two halves of withdrawal (independence and excess) seem to pull apart here. I asked a question about these kinds of relations at Dundee, and Graham clarified his position somewhat by saying that whereas intentional relations between objects are themselves real objects, the kinds of relations I’ve just mentioned are simply sensuous objects. This is a perfectly good answer, but it basically admits that these kinds of relations are to be talked about in slightly different metaphysical terms. It remains a question as to how Levi incorporates these kinds of relations in his system, but I suspect it would have to admit the same basic metaphysical difference, even if it had to situate them in a different way (e.g., at the level of local manifestation rather than information).

What I’d make of this is that there are at least two different levels going on in OOO’s account of relations: the genus of relation, and then at least two species, causal relations and non-causal relations. Independence is accounted for at the level of the genus, and then excess is accounted for at the level of the species of causal relations (among which knowledge seems to be situated). This might not be an entirely accurate account, but it raises some very interesting questions about how to understand the connection between independence and excess, and thus the unity of the notion of withdrawal I discussed earlier. However, I won’t address these questions further here.

What’s important is that I can now state very clearly something that both myself, Levi and Graham fundamentally agree on: we should not posit some special type of metaphysical relation simply in order account for the possibility of human knowledge. Where I disagree with them is that I take it that knowledge is not a metaphysical matter at all, and thus not only that there is no need to posit a special metaphysical relation to account for it, but that there needs to be no modifications to our metaphysics of relations in general or of causal relations more specifically in order to account for it. However, what’s interesting about the way this disagreement plays out is that it reveals that I share some additional common ground with both Levi and Graham that they don’t seem to share with each other. To make this clear I first need to address some of Levi’s comments about meaning and normativity (here), and to outline precisely where I differ from them.

5. Meaning and Representation

Levi gets his account of meaning from Luhmann, and, unfortunately, I don’t have the time to go away and familiarise myself with Luhmann’s work. I don’t think this should necessarily count against me, as I’m sure Levi doesn’t have the time to go away and familiarise himself with the major issues in analytic philosophy of language and Brandom’s philosophy more specifically (the relevance of which should be seen shortly). I must admit that I find Levi’s presentation of Luhmann’s ideas very confusing, but will do my best to explicate it before commenting on it. Let’s start with the major claim that Levi makes in his post (here):-

Now, it seems to me, coming at these issues from my Luhmannian perspective, that the concept of meaning is necessarily more basic and primordial than either notions of correctness and incorrectness, or issues of rationality.

I’m happy to accept that one can’t give a full account of the notions of correctness and rationality without appealing to the concept of meaning. The norms governing what it is correct to do in a given case obviously have content. To make a norm explicit one must be able to formulate it as a rule, and this must obviously mean something, or else it is no rule at all. This is doubly true of rationality, insofar as the norms governing it dictate what one should do on the basis of the meaning of both one’s own claims and those of others. However, I also think that the converse relation holds. The notion of meaning is intimately bound up with the notions of rationality and correctness, and can’t be properly explicated without them.

Before this can be properly explained, we must clear up what we mean by ‘meaning’ here. This word is notoriously polysemic, and can be used to indicate a variety of things, from the meaning of my words and the sentences I compose out of them, to the intentions underlying my actions and the relative importance of events. Levi is very clear that Luhmann’s notion of meaning is distinct from the notion of information he deploys, despite being understood in terms of it. However, despite this I find it difficult to pin down precisely what Levi means by meaning here:-

Luhmann argues that meaning is a structure in which each actuality points beyond itself to other potentialities or possibilities. As a consequence, meaning is the unity of a difference such that every actuality points to other potentialities. For example, I make a cup of coffee (actuality) but could have had an orange fanta or a glass of water. What is specific to meaning events is that each actuality is haunted by a penumbra of other possibilities from which certain actualities are selected.

It seems that what Levi means by meaning is more than just linguistic meaning. As the above example indicates, he is not talking about the meaning of a sentence about the making of a cup of coffee, but rather about the meaning of the event itself, or rather, the meaning that the event has for someone. In this respect, this idea of meaning is closer to the phenomenological approach of Husserl and Heidegger, insofar as they take it that it is experience which is primarily meaningful, and that anything like linguistic meaning is derivative of this (Levi does actually make the point that Luhmann is indebted to Husserl). Moreover, the meaning of events seems to be linked to the idea of possible actions, rather than possibility in general, adding a level of further similarity to Heidegger. This can be seen from the fact that Levi still talks about meaning in terms of the way it helps systems navigate (or act within) their environment, and the risks this always involves. There seem to be three other important facts Levi gives us about meaning: it is specific to certain kinds of system (psychic and social systems), the meaning of an event (e.g., a vocal utterance) can be the same for two systems (or the same system at different times) even if the information it produces for them is different, and that whatever enables a system to grasp events as meaningful plays a role in constituting the information those events exhibit for the system (i.e., the apparatus of meaning forms part of the ‘subjective form’ of psychic and social systems). I’ve got a number of issues with this basic model.

First, what is it specifically about the subjective form of psychic and social systems that allows them to encounter events as meaningful? Of course, it seems to be something to do with having a grasp of the variety of possible actions they can perform and relations of incompatibility between these possibilities (i.e., that if I only drink coffee I can’t also only drink orange juice). However, we’ve not been given an idea of what such a grasp would consist in, and how it would qualitatively differ from the way non-meaning based systems select their states on the basis of information. To put this in a different way, the notion of meaning is generally correlated with the notion of understanding, and we’ve not been given any account of what such understanding consists in. By contrast, Heidegger provides quite an in depth account of how such understanding of possibilities for action is constituted, but this is an account that is thoroughly dependent upon the idea of common functional norms for the use of equipment. There is nothing analogous in the brief account Levi presents here.

Second, despite indicating that this broad notion of meaning can be used to provide an account of how systems communicate with one another (even if such communication is fundamentally always miscommunication), he doesn’t provide a solid account of how this works (e.g., how something like linguistic meaning can be derived from experiential meaning). In short, we’re not given an account of what would distinguish something like a communicative meaning event from a non-communicative one. He does talk about Luhmann’s problem of ‘double contingency’ in relation to this, which he explains is a matter of how individuals can develop expectations about each other’s expectations. However, I find this discussion strange because it doesn’t say anything about how anything like a shared apparatus of meaning plays a role in structuring communication (such as the role das Man plays in Heidegger’s norm based account). This apparatus is what is supposed to provide the possibility that the same event could have the same meaning but different information on the basis of structuring the individual’s grasp of their possibilities for action, but he doesn’t relate this to the ‘double contingency’ problem at all. Then again, we haven’t been told what this apparatus is, and thus how it could play such a role.

Third, there’s a more specific form of linguistic meaning that it’s not clear Levi’s approach is able to account for, namely, the kind of meaning exhibited by declarative sentences. These kind of sentences are principally used to make assertions, which express propositions that represent the world as being one way or another (e.g., ‘My mother makes nice cakes’, ‘Nitrogen is the most common gas in the earth’s atmosphere’, ‘Unicorns can interbreed with horses’, etc.). Such sentences are also essential for forming ‘that’ clauses, which have many important linguistic functions, including explicitly formulating certain non-assertoric speech acts (e.g., ‘I desire that you make me a cake’, ‘Is it the case that nitrogen is the most common gas in the atmosphere?’, ‘I stipulate that unicorns can interbreed with horses’, etc.). In essence, it’s unclear that Levi’s account of meaning has any room for the notion of propositional content.

Now, it is a very mainstream position within the analytic philosophy of language to hold that propositional content (or whatever one calls the content of declaratives) is central for understanding all other forms of linguistic meaning. Many, including myself, take propositions to be a constituent part of the meaning of many non-assertoric speech acts, which modify these propositions insofar as they have a pragmatic force distinct from that of assertoric speech acts. So, for instance, the difference between the assertion ‘I have climbed Everest’ and the declaration ‘I will climb Everest!’ (not to be confused with the prediction that I will climb Everest, which is just an assertion about the future) is to be found in the pragmatic force they add to a proposition that represents a certain state of affairs (i.e., that I have climbed Everest). The former undertakes a theoretical commitment with the content of that proposition – it expresses that I take it to be true. The latter undertakes a practical commitment with the same content – it expresses that I intend to make it true. However, one needn’t adopt this approach. There are approaches to meaning which don’t take assertion to be the primary speech act (and thus take the meaning of declaratives to be central). Heidegger’s account of discourse (Rede) is one such approach. Nonetheless, even if one doesn’t take this kind of meaning to be central, one still needs to be able to give an account of it (and Heidegger does, or at least tries to, with his account of apophantic discourse). The important point here is that in order to do this one has to give an account of how speech acts such as assertions purport to represent states-of-affairs. If one can’t do this, then one can’t give any kind of account of how it is that we take assertions to be true or false, let alone how they could actually be true or false.

This is a good point to address another of Levi’s rejoinders, namely, that I illegitimately deploy the notion of representation in presenting the various epistemological problems I’ve been posing for OOO:-

Second, I note that Pete repeatedly refers to questions of knowledge in terms of representation. Clearly OOO cannot embrace a representational model of knowledge of the sort that Pete seems to advocate precisely because substances withdraw and are only encountered by other objects under conditions of closure with respect to substances. It is thus curious for Pete to repeatedly pitch these questions in terms of representation, for representation refers to an adequation or mirroring between a substance and another substance.

The important thing to note about this paragraph and the other places Levi talks about representation is that he always talks about it in metaphysical terms. It is obvious that such a strong notion of representation as a kind of metaphysical mirroring is problematised by OOO’s commitment to withdrawal. However, representation does not need to be talked about in these terms. Indeed, I never have. Moreover, I find appeals to such metaphysical notions of representation as problematic as Levi, Deleuze, Heidegger and Rorty among others. There are a great number of approaches in epistemology and the philosophy of language which take the idea of representation as a primitive, as something which requires no further explanation, and these approaches often fall back upon illicit metaphysical assumptions of the kind Levi is criticising. This is not the approach I take, however. I follow Brandom in trying to explain representation (both purported and successful) in terms of the notion of inference. The aim of this project is to describe what it is one must do to count as representing something. This is a thoroughly pragmatist strategy (not unsimilar to Heidegger’s), albeit one which describes the requisite practices in irreducibly normative terminology.

I won’t recapitulate the whole of Brandom’s account of representation here, but I will reiterate that this two pronged approach of normative pragmatics and inferentialist semantics is supposed to explain what representation is in non-metaphysical terms, thereby avoiding appeals to the notion of representation as a primitive (which tend to collapse into illicit metaphysics). It is commitment to this approach which underpins the claim I made earlier about the essential connection between the notion of meaning and the notions of correctness and rationality. On this approach, the propositional content of assertions is not only understood in terms of a certain practice that is described in normative terms (i.e., in terms of how it is correctly engaged in), but this practice is that of giving and asking for reasons (which is the basis of rationality as such). When combined with the idea that the meaning of assertions is to be taken as primary in relation to the meaning of other speech acts, this produces a commitment to full blown linguistic rationalism. This is the idea that all linguistic meaning is to be understood in terms of the core discursive practices of giving and asking for reasons. When this is in turn combined with the Sellarsian/Wittgensteinian idea that there is no thought without language, or that there is no meaning prior to language (even if there is information prior to it), we get the full claim: the nature of meaning cannot be understood independently of the normative structure of rationality.

In contrast to this explanatory strategy, Levi seems to want to jettison the notion of representation entirely. For him, we provide a metaphysical account of the basis of representational content or we provide none at all. It’s metaphysics or bust, and we’re bust. We can now see where Graham and Levi start to pull apart, insofar as there is some ground that I share with Graham that he doesn’t share with Levi. This is because Graham’s account is based around a notion of intentionality which Levi rejects, and this is an intrinsically representational notion. I imagine this will confuse some people, as they will wonder how Graham can deploy such a representational notion and still advocate the thesis of withdrawal. What they will find more surprising is that Graham’s appeal to the notion of withdrawal is actually dependent upon this appeal to representation, even while it is meant to undermine other such appeals. I will do my best to make this clear.

Graham’s account of the way one object encounters another is a stripped down version of the phenomenological account of how a consciousness (or Dasein) encounters an object. This means that the first object’s experience of the second has something like a phenomenological content. Of course, Graham thinks that the kind of phenomenological content found in our experience is quite different from the kind of phenomenological content found in a billiard ball’s experience, but the basic structure remains the same. This is to say that such relations exhibit intentional directedness and an as-structure. My experience of a tree is of that tree, and not some other object, and I experience it as a tree, and not another kind of thing. In the same way, when one billiard ball is struck by another, its experience is of that ball as a ball. This basic structure is the twofold structure of representational content. Propositional content is representational just insofar as any proposition shares this twofold structure (e.g., that expressed by ‘My girlfriend is at work’) has a referential aspect (it is about my girlfriend) and a predicative aspect (it predicates of her that she is ‘at work’). Now, there is a legitimate question as to whether Graham takes the phenomenological content present in encounters between objects to be propositional. This is because the latter is often taken to be a linguistic matter, which is understood as derivative upon the more primitive phenomenological content of experience (see here and here). Nonetheless, on this account, both phenomenological content and propositional content would still be forms of representational content, it would simply be the case that the representational character of the latter dependent was upon that of the former.

This is a bit of a simplification, as the issue of how a given content picks out its intentional object is quite complex, insofar as it often seems that we grasp what is represented (referential aspect) in terms of how we represent it (predicative aspect). We can leave these issues for now though. What is left to be explained is how Graham’s account of withdrawal is dependent upon the representational character of intentionality. This is because, as I’ve shown above, the excessive dimension of withdrawal is characterised in terms of a certain global failure. The thing to recognise is that this is a failure of representation, albeit a very stripped down form of representation common to both our experience and that of everything else. An object withdraws insofar as all other objects are incapable of grasping it, which means insofar as the content of this grasp always misrepresents the object. However, this kind of failure depends upon a different kind of success. In order for us to misrepresent an object, we must nonetheless manage to pick it out. For example, in order for my claim that ‘Bill Clinton is a paragon of chastity’ to be false, I have to succeed in picking out Bill Clinton, who provides the standard which my claim fails to meet. If I say ‘Xartath Gblar is a paragon of chastity’, there’s a good sense in which I’ve failed to represent anything (unless there’s some unfortunate child out there with very cruel parents, or it just happens to name a character in Jon’s D&D campaign). In short, for the predicative dimension of representation to fail (or the ‘as’ in the phenomenological case), the referential dimension must in some sense succeed. It must be this tree which withdraws from my grasp of it, and not another, or else we’re back to the kind of Zizekian real of the remainder that Graham so dislikes.

The upshot of all this is that Graham appeals to some primitive notion of representation in setting up his approach. Furthermore, it’s not only primitive, but understood in explicitly metaphysical terms. This is precisely the kind of position that both Levi and myself have a problem with. Moreover, I don’t think it’s actually a good account of how representation works. It sidesteps a bunch of crucial issues, of which criticism (2) above is only one. In essence, the problem is that I don’t think his approach actually explains representation.

All of this throws new light on my earlier criticisms of Levi’s attempts to articulate the notion of withdrawal within his system. It is because Levi denies himself the kinds of intentional (and thus representational) resources Graham deploys that he must take the opposite approach to withdrawal (independence to excess, rather than vice-versa), and this is ultimately what prevents him from properly cashing out withdrawal in terms of translation. Information and representational content are two different things.

6. Knowledge and Pragmatism

We now have a somewhat interesting picture of the common ground I share with Levi and Graham. I agree with both of them that knowledge should not be understood in terms of some special metaphysical relation. However, I agree with Levi (against Graham) that encounters between entities should not be understood in intentional (or more broadly representational) terms (and even that they should be positively understood in informational terms), while nonetheless agreeing with Graham (against Levi) that knowledge should at least sometimes be understood in such terms. As I’ve already noted, I can only hold this middle ground because I deny that knowledge should be understood in metaphysical terms at all. It should now be clear that this is because I think that propositional content should not be understood in metaphysical terms. This position needs clarifying further, as to many it looks like a dodge. However, in order to do this, it’s helpful to further examine Levi’s claims about knowledge in the second part of his response (here), and the problems that his radical anti-representationalism gets him into.

Let’s start with Levi’s anti-representational counter-proposal:-

From the onticological standpoint (and here I don’t know if Graham’s object-oriented philosophy is in accord with my position… Maybe he’ll speak up), knowledge is not a representation of objects, but is rather a construction and an action. In this respect, knowledge is closer to a recipe in cooking than a reflection in a mirror. A recipe does not represent a meal. It doesn’t tell you what a meal is “like”, nor does it maintain a relation of adequation to the dish. Rather, a recipe is a series of operations involving various implements and ingredients for producing something. A recipe says do this and do this and do this and you will produce this. Knowledge is thus not a representation or mirroring of something, but is a list of actions for producing a particular local manifestation. It says that if you do this, then you will produce this local manifestation.

I’ll delve deeper into what Levi means by the fact that knowledge is “a construction and an action” further on, but I’d like to make an initial point about what Levi says here. On the account presented above, knowledge consists in instructions for action. The problem with this idea is that it entirely presupposes the notion of representation it’s supposed to undercut. If we are to be able to communicate a set of instructions to someone, we must be able to represent to them the individual actions that they are supposed to perform. If we generalise the schema Levi gives us: ‘If you do P then you will produce Q’, we should be able to see that P and Q are propositions which represent possible states-of-affairs. If you make true P you will make true Q, but in order to understand what it is to make true either you must understand what it is to take them to be true. You can’t say ‘do X’ without being able to say ‘X is the case’.

This returns us to the earlier point about the relationship between norms and meaning. Instructions are just rules, which is to say explicit norms. One can follow them correctly and incorrectly, insofar as the propositional contents determine success criteria for whether one has genuinely brought about the right states-of-affairs. Of course, Levi wants some kind of non-representational way of conveying these kinds of instructions, but he needs to actually provide us with a viable account of such conveyance.

I think that what Levi means by the claim that knowledge is an action is actually a more familiar point, namely, that knowing-that is really just a form of knowing-how. However, he combines this claim with some other problematic claims about the importance of those who produce knowledge as opposed to those who merely report it:-

However, whether we’re talking about the know-how of craftsmen such as artisans or engineers or the investigations of experimental scientists, knowledge just never takes this form. For the actual knowledge producers (rather than knowledge reporters), knowledge takes the form of a set of procedures for producing local manifestations or actualizations, a set of techniques for producing particular outcomes, a set of translations for producing certain results, rather than a set of representations.

The particular claim made in this quote is just false. At best, it is true not of those who produce knowledge, but those who use it. This could be interpreted as the claim that although knowledge can be communicated in linguistic form, you only grasp this knowledge if you can actually put it to use. However, this would need to be reconciled with the claim that much newly produced knowledge has no immediate use at all, at least by ordinary standards. Theorems discovered at the forefront of mathematics are understood prior to the discovery of any application they may have, and this means that understanding them cannot be a matter of understanding their application. Similarly, at the forefront of physics their are often different theories that have no difference in real world practical applications (to the extent that we can’t even experimentally choose between them), but we can nonetheless understand how they differ in content.

The important point here is that if one is going to try and reduce knowing-that to knowing-how, you need to actually specify precisely what kind of knowing-how the former consists in. If you don’t do this, then you’ve simply abandoned the distinction rather than explaining it. By contrast, both Brandom and Heidegger take broadly pragmatist approaches to knowing-that which try to explain it in terms of a specific kind of knowing-how. Both of them take it to be a matter of knowing-how to use linguistic tokens and the way these practical linguistic abilities tie in to our broader practical abilities to engage with the things these tokens refer to. They differ insofar as Brandom thinks that one cannot have linguistic abilities without having the ability to track the inferential connections between sentence tokens (which thereby gain the significance of expressing propositions). However, this doesn’t mean that understanding the content of a knowledge-claim is only a matter of understanding its inferential significance. These abilities must appropriately link up with both perceptual capacities to make assertions on the basis of non-inferential causal inputs and practical capacities to produce actions (or non-inferential causal outputs) on the basis of assertions and practical commitments (there are also social capacities for tracking the anaphoric and deferential relations between linguistic tokens, but this is quite complicated).

I endorse this broadly Brandomian story, while nonetheless thinking that there are still lessons to be learned from Heidegger’s approach. What this means is that I have a story about what it is to know that something, insofar as I have a story about what it is to grasp the meaning of the declarative sentence that makes up the ‘that’ clause (i.e., the something), or the content of the proposition it expresses. Loosely, this means having some combination of abilities to navigate the inferential role that sentence plays within a discursive context, to perceptually confirm it or other sentences it is appropriately inferentially related to, to practically act to make it true (or other appropriate sentences), or to identify and defer to others with relevant abilities. In short, it involves knowing how to use the sentence, and knowing how to use it correctly. Levi has no comparable account of the meaning of declaratives and thus no account of how knowing-that is grounded in knowing-how.

This brings us to what Levi means by the fact that knowledge is a construction and therefore local:-

Here “construction” should be taken in the most literal possible sense. “Construction” does not denote “making up whatever you like”, but rather refers to working with actual substances to produce particular local manifestations. Just as the cook is powerless to cook without kitchen equipment such as stoves, ovens, grills, various utensils, a lemon squeezer, and his ingredients, the would-be knower (supposing that he were distinct from a cook) requires his implements, his tools, his substances, and so on. The scientist is nothing without telescopes, beakers, centerfuges, ingredients, etc., etc., etc. Thus as Haraway already has observed, knowledge is always local knowledge. Knowledge is the set of procedures or operations for producing particular local manifestations under highly specified contexts. Substances could manifest themselves differently were they perturbed in different contexts by other substances and other implements. There is nothing representational here. As for the objectivity of these doings, it suffices to simply repeat the operations to determine whether they produce the same local manifestations.

This is a serious overgeneralisation. It does not follow from the fact that inquiry is dependent upon context specific factors such as the kind of equipment that is used that the knowledge it produces is somehow necessarily relativised or restricted to that context. The whole point of experimental inquiry is to produce generalisations that hold beyond the experimental context. This is indeed what the content of the theories produced says. When Einstein used his theory of general relativity to derive predictions about the curvature of space-time, he claimed that they would hold up to methods of experimental verification not yet invented (which they did). This was not a claim Einstein made in addition to the theory, but a consequence of taking the theory to be true.

This isn’t to say that results can’t be context dependent. There are many cases in which previous results have been shown to be dependent upon contextual factors that were not appreciated at the time. However, the reason Levi’s claim is an overgeneralisation is that it assumes a priori that this is true in all cases. In truth, such dependence must be demonstrated on a cases by case basis: equipment must be shown to produce distortions that can be eliminated with improved equipment, background theoretical assumptions must be shown to be unwarranted on the basis of new ideas with greater explanatory power. However, this is nothing other than the rational process through which science develops itself, divesting itself of limited understanding in favour of more expansive theories. In essence, this epistemic dependence of knowledge on the conditions of its production must be understood from the perspective of the rational process of argument and assessment, rather than the other way around (compare here).

Finally, we can turn to Levi’s answer to objection (2). Levi says that although all knowledge of objects is in some sense inadequate (due to withdrawal), there can nonetheless be normative assessment of knowledge as better or worse, if not necessarily as true or false. In his initial quick response to this objection he said that this is because we can distinguish between successful and failed translations. He elaborates this as follows:-

This also provides the means of responding to Pete’s questions about determining the difference between translations that don’t work and those that do. Insofar as knowledge is a construction, insofar as it’s something that needs to be built, there is little difference between knowledge and a recipe that doesn’t turn out well. Poorly constructed knowledge is unable to stand or produce particular local manifestations. I’m not sure what the mystery is here. The substances fail to hold together in the production of a local manifestation. That’s all there is to it. If I don’t give a general answer then this is because the answer pertaining to norms is system or substance-specific and there are a variety of different substances. There isn’t, as I argued in my last post, one answer to this question.

Now, this answer is obviously subject to the first problem I pointed out: how can knowledge be characterised as like a recipe or a series of instructions for producing effects if we have no account of how those instructions or effects are to be conveyed? However, there are deeper problems here. For the moment, lets forget about the problems of knowing-that and just talk about pure knowing-how. Assume that in doing so we’re simply talking about capacities for producing certain effects, and that the process of learning such know-how is a straightforward matter of adaptation in relation to one’s environment. If we treat all knowledge in these terms, then Levi’s solution looks very much like that of classical pragmatism. The success of knowledge (i.e. ‘truth’) is understood in terms of something like utility, as the effectiveness of a capacity for achieving a certain end. There are a series of problems for this solution.

First, it recapitulates the major problem of classical pragmatism, namely, that truth and utility tend to pull apart (e.g., simpler theories are often more useful than more accurate ones, depending on context). This is because, as Levi seems to indicate, there is no single end in relation to which utility would be determined in every case. The end in each case is determined by the desires of the knower. This pretty much annihilates the unity of truth. This is a regulative ideal governing inquiry which holds that there aren’t several incompatible correct theories about the world, thus forcing us to give and ask for reasons for our respective positions if we disagree. Rejecting the unity of truth undermines the impetus of rational debate and the force of reasons as such. One ends up in situations where one says ‘that may be true for you, but it’s false for me‘. The problems with classical pragmatism have been elaborated at length by other commentators, so I won’t provide further details here.

Second, there is the question of precisely how we determine the end against which the effectiveness of capacities is to be judged. The end is what provides the criteria of success for such judgments. The problem is that ends are themselves normally understood in terms of propositional content, i.e., in terms of a proposition that is to be made-true by bringing about the state-of-affairs it represents. On the classical pragmatist model this is supplied by desires, but these are themselves usually characterised in terms of a proposition that the knower wants to be made-true. Of course, there have been attempts to produce accounts of desire which treat it as something other than a propositional attitude (e.g., Deleuze & Guattari’s metaphysics of desire). However, it remains to be demonstrated that such notions of desire can play the role required of them. The benefits of moving away from determinate propositional contents are balanced by problems for providing determinate success conditions.

Finally, the quote above seems to indicate that this solution is not simply meant to apply to some specific kinds of substance, but to all substances. So far I’ve been trying to interpret Levi’s account of knowledge as distinct from Graham’s on the point of whether all interactions between objects should be understood in terms of knowledge. However, even if this is Levi’s intention, his solution to the problem of how knowledge can be assessed says nothing about the structure of any specific substances. If Levi interprets the notion of desire required to supply normative criteria in a system-specific way then he simply retains all of the above problems. If he interprets it in a way which applies to all systems (like Deleuze & Guattari) then he’s got additional problems. This is because he’d not only need to account for the desires of billiard balls, but to do so in a way which is still sufficient to determine success conditions. The only viable way of doing this is to posit something akin to Spinozan connatus, which indexes desire to systemic integrity (or some other suitably general systemic notion). This creates its own problems, because it fails to account for many ordinary desires (e.g., trivial things that have no relation to systemic integrity) and runs counter to others (e.g., suicidal desires). In short, I don’t think any of the ways of cashing out this solution are viable.

To be continued…