Brandom’s Circular Semantics?

The aspects of Brandom and McDowell’s projects that I am focusing on — namely, freedom and rational constraint — sit alongside sustained attempts to understand intentionality. Both are aware that from a certain vantage point, phenomena like meaning can appear mysterious. An approach to language that typifies such an attitude would be the ‘Augustinian’ model that Wittgenstein sets out in the Philosophical Investigations in which, confronted with a name abstracted from the pattern of its usage, its purported ability to designate its bearer seems inexplicable.Whilst sharing an acknowledgment that an air of mystery can hang over semantic properties, Brandom and McDowell adopt different strategies for dispersing it. Adumbrating these strategies will set the stage for a hesitantly advanced objection to Brandom.

Brandom opposes his semantic theory to a position he calls representationalism. The representationalist treats representation as an explanatory primitive, attempting to understand the concepts of reasons and truth in terms of it.This specific explanatory move between representation and the concepts of truth and reasons, he thinks has met with some success. Starting with a set of representational primitives corresponding to subsentential expressions (such as singular terms and predicates), the representationalist can proceed to specify truth conditions for sentences in which those expressions feature. Then, on this basis, correctness of inference can be explained by specifying the truth-preserving routes between such sentences. Brandom’s main objection to such accounts is not concerned with the strategies employed by the representationalist once representational primitives are in play; rather, he thinks that no congenial understanding of this sort of representational content and our relation to it is to be had.

The rival approach that Brandom champions is inferentialism, which adopts inference as primitive in semantic explanation. As I’ve discussed previously, he thinks that he can give an account of inference which is built upon the social practice of assessment of discursive activity. This account attempts to understand what inference is by appeal to our practice of treating the commitments-whether doxastic or practical-which people acknowledge as committing them to further things. With this account of inference in place, Brandom embarks on an ambitious project that seeks to reduce appeals to representation to something that can be constructed from an account of the social activity of holding each other to the consequences of the commitments that we have undertaken. To put it in Brandom’s idiom, we are to ground an inferential semantics in a normative pragmatics.

For Brandom then, the representationalist has got things backwards. By trying to reduce inference to a function of the prior relations between representational contents, she will have deprived herself of the very resources needed to adequately characterise talk of representation: namely, a conception of inference. In motivating and defending a prior account of inference, Brandom hopes to clarify the foundations of his explanation of semantics in a way that he thinks the representationalist has been unsuccessful at doing. Thus, he begins with a conception of inference understood in terms of the social practice of playing the game of giving and asking for reasons rather than in terms of representations in which subsentential expressions designate objects or properties. Then, he exploits this non-representational conception of inference to explain how we can retain our ascriptions of intentional content to sentences, along with many of the other central features of language that representation had been invoked to explain.The problematic, designational conception of reference attaching to the representationalist model of semantic explanation can thus be avoided; reference as traditionally conceived is “explained away.”So, we are meant to be reassured that the semantic properties of language, as something whose intentionality may have quite understandably struck us as mysterious on the representationalist’s necessarily incomplete story, are not threatening after all. We just need to reverse the order of explanation that we pursue, showing how semantic features are grounded upon the familiar but nonetheless crucial activity of treating people as having a set of doxastic and practical commitments.

McDowell accepts that the strategy of Brandom’s representationalist is unworkable but thinks that reversing the order of semantic explanation will fare no better.What McDowell questions is whether, faced with an initially puzzling concept like representation, the way to remove this puzzlement is by reconstructing that concept-say, by showing that its explanatory role can be mimicked without appeal to it-before being entitled to resume our use of it. The problem with the representationalist model was supposed to be that it lacked the resources to explain how subsentential expressions could designate things, what this designation consisted in and how this linked up with the practice of using language. McDowell thinks that the problem here is treating representation as a primitive explanatory item, intelligible apart from the wider context in which it has an intelligible semantic role. It is this failure which generates the confusion, not simply the idea that names can designate objects. As such, we do not need to reconstruct a conception of representation by appeal to practices of social assessment that may be able to do a similar job, but to provide the necessary context in which representation makes sense. In this case, we can bring in an account of inference not as a competitor to the traditional conception of representation but rather as the needed supplement to make that (unreconstructed) conception intelligible. Thus, McDowell thinks that he can agree that semantics must be understood with reference to inference, and thus as inseparable from the practice of treating some things as reasons for others, without endorsing the inferentialist order of explanation.

Brandom retorts that he is aware that he has alternative options open to him that do not consist in focusing upon different primitive explanatory categories, and that instead he might have tried to explain semantic properties by appeal to both representation and inference.However, he thinks that if his more ambitious project, employing a more austere set of tools, is successful, then it should be preferred. This is because he thinks that it would have the advantage of being a reductive explanation and so would have the theoretical virtue of parsimony on its side. It is not clear, however, that this sort of response will fully address McDowell’s worries, since McDowell’s target includes both the idea that a reductive account is needed and whether we can give a good explanation of semantic intentionality by suspending our appeal to the concept of representation until it can be reconstructed in different terms. We need not follow this debate further though, since enough has been said to allow us to be able to connect these approaches to semantics to the issues of freedom and rational constraint.

The debate as sketched above between Brandom and McDowell centres on the adequacy of Brandom’s inferentialist strategy in semantics. The main issue was whether semantics could be understood without its traditional representational dimension and with merely a normative-pragmatic one. I do not intend to address this potentially wide-ranging question but rather an offshoot of it. This further question is whether rational constraint from extra-social reality can be accommodated by a semantics based upon a normative pragmatics without a representational dimension.

We can begin to frame this worry about rational constraint as follows. Brandom’s semantic inferentialism is a rejection of representationalism in favour of a normative pragmatics that understands inferential relations as products of a social practice of keeping track of what we take our fellows’ doxastic and practical commitments to be. What we want to know is whether this sort of normative pragmatics can accommodate a specific kind of norm: rational constraint by extra-social reality. This is the sort of norm that will be in play when objects exert authority over our thinking; and this is something that is required to make sense of the idea that the thoughts generated in the course of our inquiries are answerable to what we are inquiring about.

The problem arises from the order of explanation that Brandom pursues. Eschewing a more modest approach that would propose to understand semantics in terms of both representation and inference, he seeks to explain it only in terms of inference, where this latter notion is explained in terms of the social practice of giving and asking for reasons. This appears to require that this normative pragmatics does not stand in need of explication by appeal to the sort of intentionality that its role in grounding semantics is meant to explain. But it would seem that this normative pragmatics cannot itself be elucidated independently of the semantics that it is called upon to explain, which Brandom will require it to be. This is because norms for rational constraint, such as those determining the propriety of empirical judgements, are not fully intelligible as such prior to an account of intentionality. So, for example, an empirical judgement is essentially something that is subject to a norm which determines whether it is correct or incorrect; and if we fail to grasp this fact, we will mischaracterise judgement. The crucial aspect of the rational constraint exercised by such norms though is not just that they determine whether a judgement is correct or incorrect, since any arbitrary norm could sort performances into correct or incorrect ones according to some standard. What is distinctive of such norms is that they tie the propriety of performances like judgements or thoughts to how things stand in the world: specifically, what such judgements and thoughts are about. Therefore, given the strictures that he himself places upon it, Brandom’s account appears circular. The reason for this is that he appeals to a supposedly self-standing social model of normativity to explain intentionality, but to explain a crucial subset of norms that compose this social model we must rely upon an account of intentionality that by Brandom’s lights we ought to be precluded from appealing to.

A defender of Brandom might object that understanding rational constraint in this way is anachronistic given the new inferentialist project which rejects both representationalism and the weaker two-pronged approach including representation and inference. So, unlike someone like Rorty, who rejects the idea that the world can exert a rational constraint on our practices, we might think that Brandom’s accepts this sort of constraint because he believes himself to have found a way to domesticate the idea. One of Brandom’s advances over Rorty could be seen to be his realisation that rational constraint upon our practices exerted by how things stand in the world is achievable from within the bounds of sociality rather than requiring some problematic, non-causal relation to obtain between us and empirical objects.In other words, recognising the importance of intersubjectivity as the locus through which normative claims must pass for their authority to be intelligible does not forestall the possibility of objectivity but is the conduit through which that objectivity is realised. So, given this social basis for rational constraint, alongside the rejection of traditional representational relations, is it not mistaken to accuse Brandom’s account of normative pragmatics of presupposing semantic features that it is meant to explain? For it may seem that it is only the old, representationalist conception of rational constraint that needs to invoke a semantic relation between thought and judgement and objects to which they are answerable.Therefore, in keeping with Brandom’s demanding order of explanation, the normative pragmatics that explains semantics would not require an appeal to semantic notions to explain it.

This would be to tell only half the story though. Whilst Brandom is opposed to a representationalist understanding of intentionality, modelled on a designational relation between a name and the object bearing that name, his positive project wants to understand a different sort of intentionality, namely the “propositional contentfulness of attitudes.”This sort of intentionality is the sense of ‘aboutness’ in play, for example, when we say that to suspect that the cat is on the mat is to hold an attitude about the cat being on the mat. If understanding the normative pragmatics that is called upon to explain Brandom’s inferential semantics depends upon a prior understanding of this sort of intentionality, then it will be circular. Although understanding the norms for rational constraint that Brandom invokes does not rely on the representationalist, designational model of intentionality, it seems that it does rely on an understanding of propositional content; therefore, it will lead to circularity. To establish that for Brandom rational constraint implies intentionality of this sort, consider what Brandom says about how rational constraint is to be secured:

If a suitable story is told about how [non-inferentially elicited judgements] are rationally criticizable by those who key their correctness to their correspondence to the facts reported, and about their entitlement to the reliability of the noninferential process that elicits them, then rational constraint by how things actually are is secured.

Thus, on this account, it seems that to fully understand norms for rational constraint then we must already grasp the intentionality of thought and judgement as things that purport to correspond to the facts, even if we do not give a representationalist account of this purport. But for Brandom, our entitlement to this sort of semantic concept is meant to be secured only after it has been demonstrated that an independently intelligible normative pragmatics can be shown to have made it respectable.

Again, an objector might claim that Brandom’s project has been misunderstood here. For when Brandom sets out to construct an inferentialist semantics based upon a normative pragmatics, he does not have to show that all norms are intelligible prior to the employment of semantic concepts. Rather, he sets out a formal account of how the game of giving and asking for reasons is structured in such a way that moves within that game can be ascribed intentional content. It is central to this project to give an account of the social practices that confer intentional content in virtue of our activity of tracking and instituting normative commitments. But this is very far from giving a full-blown theory of normativity, and arguably nor does this seem to be required to complete Brandom’s project in semantics. He is happy to admit the limited scope of his ambitions, at least in Making It Explicit: “The methodology of MIE requires appealing to the very weakest, most primitive sort of social normativity possible.”This appeal, he continues, is not meant to serve as a basis for understanding all aspects of normativity. This suggests the possibility that an account of norms for rational constraint can be understood in a different way to the norms undergirding semantic content. So, if the account of normative pragmatics that underlies Brandom’s semantic inferentialism does not have to include an account of norms for rational constraint then the charge of circularity can be avoided because semantic concepts will no longer be required to be employed in the normative pragmatics that is supposed to explain them.

The problem with a response of this kind is that there is good reason for thinking that the intelligibility of Brandom’s inferential semantics is in fact dependent on an account of norms for rational constraint. This is because it is the connections outside of the practice of giving and asking for reasons, whereby how things are in the world can exert an authority on how things ought to stand within the practice, which prevents the practice from becoming a self-contained game. It is in light of this sort of normative (rather than exclusively causal) significance which worldly happenings can have that it makes sense to suppose that the performances which make up a linguistic practice have a bearing upon those worldly happenings. One way to put this point would be to say that what it is to have the world in view (to borrow a McDowellian phrase) is to be connected to something that essentially, and not merely accidentally, reveals to us what it would be correct to say about it. Absent this presumption that what we do with words is appropriately responsive to what we take to be the case, it is hard to see what the why this linguistic activity should also count as semantic activity-as  something that qualifies as being about the world rather than simply generated in interaction with the world.

The charge against Brandom is not that he denies the connection between semantic content and rational constraint in general. The objection is rather that the order of explanation that his pursues, subsuming semantics under a normative pragmatics, cannot accommodate it at the right point. Specifically, on his account we should be able to explain why the activity of keeping track of the changes in normative commitments brought about by the performances of our fellows allows us to ascribe intentional content to some of those performances; and we should be able to do this without already invoking semantic concepts. However, it seems to be a condition of those performances being able to have a semantic content at all that they be subject to norms of rational constraint. But to explain what is involved in rational constraint then we must already invoke semantic content, which is the very thing that we wanted to explain. Thus, Brandom’s account will contain a pernicious circularity in virtue of the role that rational constraint must play within it.

Advertisements

Hegelian Glee-Watch: Scholasticus Goes For A Swim Edition

In my recent post on Kant’s idealism, I noted that part of Kant’s strategy was to give a critique of our powers of cognition so as to identify the conditions for epistemic access to appearances. According to Hegel, insofar as this critique must be prior to the exercise of cognition, so as to determine how our cognitive capacities are to be applied (e.g. only to sensible objects, conforming to the forms of intuition and a fixed set of general conceptual categories) then he thinks Kant has got himself in a bind. With somewhat uncharacteristic perspicuity, Hegel sets out his objection like so:

A very important step was undoubtedly made, when the terms of the old metaphysic were subjected to scrutiny. The plain thinker pursued his unsuspecting way in those categories which had offered themselves naturally. It never occurred to him to ask to what extent these categories had a value and authority of their own. If, as has been said, it is characteristic of free thought to allow no assumptions to pass unquestioned, the old metaphysicians were not free thinkers. They accepted their categories as they were, without further trouble, as an a priori datum, not yet tested by reflection. The Critical philosophy reversed this. Kant undertook to examine how far the forms of thought were capable of leading to the knowledge of truth. In particular he demanded a criticism of the faculty of cognition as preliminary to its exercise. That is a fair demand, if it mean that even the forms of thought must be made an object of investigation. Unfortunately there soon creeps in the misconception of already knowing before you know — the error of refusing to enter the water until you have learnt to swim. True, indeed, the forms of thought should be subjected to a scrutiny before they are used: yet what is this scrutiny but ipso facto a cognition?

Hegel, Shorter Logic, s.41

Draft of Chapter 2: Brandom on McDowell on Freedom and Rational Constraint

Here is a preliminary draft of the second chapter of my thesis. The aim of the chapter is to explicate the relation between freedom and rational constraint in both Brandom and McDowell. I don’t try to assess either position at this stage but simply to try and frame their projects as involving attempts to develop Kantian accounts of autonomy that bear upon my project of exploring the relation between the senses in which we are both free but nonetheless subject to external constraints. The first half deals with Brandom, reworking and revising some of the things that I have written here on him. The second half is new material on McDowell and needs the most work still to expand upon and sharpen up. As ever, any comments are warmly appreciated!

Update:Two years and many revisions later, this material is no longer the basis of a thesis chapter, but dismembered parts have still found their way into the final draft.

Hegel and idealism I: The case of Kant

Hegel’s idealism is a tricky issue to get a handle on. In this post, I’ll try to lay the ground for a short series that picks up on one strand running through it, relating Hegel’s idealism to Kant’s, as I have done in brief previously. This will be only a very partial picture, sidelining a consideration of the important influence of the idealisms of contemporaries like Fichte and Schelling, and those of the ancients like Plato and Aristotle. Nevertheless, I do not think it simplifies the picture too much. We can start, then, by considering Kant.

In what sense was Kant an idealist? My brutally short account begins as follows: In his oft-quoted 1772 letter to Herz, Kant says that in his previous work, “I still lacked something essential, something that in my long metaphysical studies I, as well as others, had failed to pay attention to and that, in fact, constitutes the key to the whole secret of hitherto still obscure metaphysics.” This key comes from an answer to a further question: “What is the ground of the relation of that in us which we call ‘representation’ to the object?”

It is the dogmatic failure to ask this question that Kant thinks has led all of his predeccesors (even Berkeley) to assume a form of realism, which he calls ‘transcendental realism’. The transcendental realist takes the concept of objectood to be independent of epistemic conditions. So, for this type of realist, the question of what we take objects to be is not dependent on what must be in place to know these objects. For them, first, we have some concept of objecthood; then, we go on to ask how we come to know the things that this concept picks out. (Or at the very least, the transcendetal realist thinks that these two questions are in principal seperable.)

Kant believes that the problem with this is, had his question been asked, it would be apparent that pursuing or merely assuming an answer to the ontological question of objectood, apart from the epistemic conditions for knowledge of such objects, was insufficient. According to him, there is a lacuna in any such approach. This is because it will fail to explain how an object comes to be for us — how we can come to represent it, or otherwise be in a meaningful cognitive relation to it. To say that objects affect us, say, by causally impressing themselves on us, would not yet be to explain what it is about both us and the object that allows this affection to form a representation connecting us and the object. Conversely, saying that we represent objects through our affection of them (actively interacting with them), is once again to fail to explain how it is that objects are available to us so that we can grasp them in this way.

To give an adequate explanation, Kant thinks we must take the ‘Copernican turn’. This turn has two closely related moments: one methodological and another substantive.

The methodological component involves making a distinction between the old, transcendental realist, conception of objects, and a new epistemically-inflected conception. So, Kant wants to retain some idea of reality as composed of things in-themselves, which are as they are independent of our capacity to know them. But he thinks that this conception is of no use to us in explaining what our cognitive connection to reality is. The new conception of objects is as things standing under conditions (not yet specified) of knowability; thus, what Kant calls ‘appearances’ (as opposed to ‘things-in-themselves’) are objects insofar as they are essentially available to the subject.

Tied to this methodological move, is a further substantive thesis about features of these objects. Following the methodological distinction, the required sense of objecthood for explaining our connection with reality will be one dependent upon the epistemic conditions that enable us to know objects. If, so conceived, objects must conform to the cognitive capacities of subjects, Kant thinks that it would be remarkable if we were just presented with objects of exactly this type, as if God had set up some harmony between object and cognition. So, he thinks that the subject must have some role in establishing the conformity of objects to the conditions under which we can know them. This role is to actively constitute objects, but only in respect of the conditions for them to be known by us. (Kant also thinks the failure of previous metaphysics justifies this as a tentative experimental hypothesis, but that need not concern us here.)

Kant’s idealism, then, consists in this: objects of knowledge are dependent upon knowing subjects for those features that enable them to be known. Obviously, what this then hangs upon is what features enable objects to be known. Kant argues that our only knowledge of objects is through sensible experience, and attempting to know objects apart from such experience simply leads reason into interminable confusion and contradiction. So, the conditions for knowledge of objects are those related to sensibility. But Kant does not think that all sensible properties of objects are directly dependent upon knowing subjects. Instead, he argues that it is the form of sensible experience, along with a set of conceptual structures in which the objects of experience are relatable, that are the relevant conditions of knowledge. This means that the subject only provides a spatio-temporal framework for objects, along with a number of principles (such as causality) that we must apply to organise experience in such a way that it can present objects graspable in thought.

Once a set of very general a priori conditions are in place, we have no role in shaping objects. Thus, Kant can claim to hold onto a qualified form of realism at the level of ordinary empirical properties. So, for example, we contribute the forms of space and time that an object must appear within, but that does not mean it must be our contribution where and when a given object appears. Thus, Kant marries a transcendental idealism, at the level of the a priori conditions of appearances, with an empirical realism, which holds at the cognition of those appearances.

Now, this is a very schematic account, and which fails to incorporate some of Kant’s central concerns (such as his attempt to explain synthetic a priori knowledge and his arguments against the reality of space of time). Nevertheless, I hope the general picture is clear. First, Kant seeks to explain our cognitive relation to objects. To do so he begins with a novel distinction between things-in-themselves and appearances (i.e. objects which are essentially knowable). He goes on to suppose we can make better sense of our relation to reality if we undertake the hypothesis that objects must conform to cognition rather than cognition to objects. This involves a critique of cognition to determine what the conditions for knowing objects are. The result is a conception of objects as having some formal features contributed by us a priori, whilst having their material empirical properties more robustly themselves.

Obviously, the correct interpretation of Kant’s theoretical philosophy is controversial, and the nature of his idealism particular so. I am somewhat attracted by a reading that moves away from language of the subject ‘constituting’ objects, of them conforming to the conditions of its experience, replacing this with the idea that objects must conform to the conditions of experientiality in general. This is the sort of Kantianism that, arguably, is latent in the Tractatus, though I am not yet sure to what extent it is Kant’s. Either way, I think the picture of Kant presented above is not too far from Hegel’s understanding of Kant — encapsulated in his claim that Kant was a ‘subjective idealist’ — and it is that understanding that will be paramount in the following posts.