Robert Brandom’s long-awaited A Spirit of Trust sets out to tell us what is living in the philosophy of Hegel. The answer is – unsurprisingly – ‘a lot’, with much of it having been kept on ice by Wilfred Sellars, Donald Davidson and the later Wittgenstein. Several of the book's core claims will be familiar to readers of Brandom's earlier works, Reason in Philosophy and Tales of the Mighty Dead, while readers of German will have had a sneak-peak of the introductory chapters in Brandom's Wiedererinnerter Idealismus. Footnote 1 This isn't to say that the book doesn't make novel contributions. Hegel scholars might say that the book makes many novel contributions but this review won't assess those claims. The book provides a broadly naturalist analysis of the primitive representational capacities implicated in the attitude of desire, an elaboration of how social-recognitive relations can be constructed out of these, and a sophisticated, diachronic, recollective approach to Frege's sense-reference distinction. Each of these is worth the entry fee. Rather than attempting to list every original claim, this review will pull at one thread of the tapestry to try to give the reader some idea of the book's core themes as well as their significance for contemporary debates in metaphysics and the philosophy of language.
One of the big ideas in A Spirit of Trust is that we've been approaching the relationship between objecthood and modality from the wrong direction. This wrong direction, the ‘extensionalist order of explanation’ (p. 147), tries to account for the content of our modal claims in terms of the relationships between objects, typically objects in different possible worlds. In doing so, it assumes that we can speak of distinct and distinguishable objects prior to our grasp of the modal notions being defined.
With the extensionalist approach we take a domain of objects, define properties extensionally as sets of these objects, then define modal properties as relations between objects in different possible worlds. I am ‘hungry’ if I'm in the set of hungry things and ‘possibly hungry’ if I have a counterpart in the set of hungry things in an accessible possible world. If it's possible that I have a pet lion, then there is an accessible world in which I have a pet lion. From the very start of this story, we assume that we have a grasp of distinct objects before defining their modal properties. These objects are viewed as ‘merely distinct’ in the sense that they are different but needn't have incompatible properties (p. 148). This method gives us the indiscernability of identicals for free. We get the identity of indiscernibles by stipulating that every set of objects determines a property. By stipulating that certain sets are disjoint in all worlds, we arrive at the idea that the corresponding properties are incompatible. The order of explanation begins with objects and ends with a modally significant notion of incompatibility.
Of course, things aren't so simple. Quine argued that, while we can make sense of the distinctness of objects in the actual world, we don't have clear identity conditions for objects in other possible worlds: ‘Take, for instance, the possible fat man in that doorway; and, again, the possible bald man in that doorway. Are they the same possible man, or two possible men?’.Footnote 2 Much ink has been spilled trying to make sense of the relations between identity and modality in general, and in arguments about the nature of transworld identity and the ontological status of possible worlds in particular. Some philosophers work around these issues by positing essences or capacities as fundamental properties of reality. David Lewis notoriously thought that we should accept that other possible worlds are spatiotemporally-extended concrete entities just like the actual world. Others, like Timothy Williamson, have argued that, since S5 is the most plausible logic of modality, the contents of those worlds are identical with the contents of our world.Footnote 3 The consequence of this is that all the possible objects must be located in our actual world. They are actual but not concrete. My possible pet lion exists after all, I just can't pet it.
If Brandom is right, a lot of these debates arise because we have been approaching things from the wrong direction. Brandom takes modality to be fundamental, not because we live in a universe of essences but because it is a necessary condition for any content to be determinate that it stand in relations of incompatibility to other contents and incompatibility is a modal notion. The modality which permeates the world in the form of modal facts is an expression of the necessary conditions for the possibility of grasping any determinate content. To say ‘cats are necessarily mammals’ is not to describe any kind of super-fact about the world that is made true by metaphysically interesting objects (e.g., possible cats, non-concrete cats, cat-essence). It is simply how we express that it's incompatible for something to be both a cat and not a mammal. Even if Quine is thinking about an actual bald man in the door, in order to grasp this content he must understand that the baldness of the man precludes him having a full head of hair. Being bald and having a full head of hair are incompatible and we make sense of this incompatibility by appealing to the modal idea that it's not possible to be both bald and hirsute.
A central idea of the book is that, assuming that our experience has content, this content is only determinate insomuch as it excludes other contents. ‘Circular’ doesn't mean anything if it is compatible with everything, at the very least, it must be incompatible with something (‘square’, for example). Likewise, you only classify something as a ‘cat’ if that classification is incompatible with others (e.g., ‘fish’). On this model, the content of a universal like ‘red’ can be understood as the contrary of everything that it is incompatible with (’the negation of the negation’). Since we identify and individuate particular objects according to their properties, the identity of a particular is determined by the negation of the negations of these properties. Universals are explanatorily prior to particulars.
The idea that determinateness is achieved by precluding alternatives isn’t claimed to be new. Hegel attributes it to Spinoza (Omnes determinato est negato), Brandom connects it to Shannon's theory of informationFootnote 4, and we might trace it to Dignaga's theory of apoha.Footnote 5 The particular form it takes, that content is determined by material incompatibility relations, is the key thesis upon which the book stands and much of the text works to tease out its significance. As noted above, one consequence is that, if having contentful representations of objects requires grasping relations of incompatibility between contents, and incompatibility is a modal notion, then we shouldn't analyse modality in terms of pre-existing objects.
A second big idea is that this relation of material incompatibility shows itself to us in two different forms (pp. 53–54). Objectively, we understand it as alethic modality (it is not possible for Kermit to be both a frog and a mammal). Subjectively we experience it as deontic modality (you should not believe that Kermit is both a frog and a mammal). To stand in these relations is to be ‘conceptually structured’ (p. 57). The term ‘conceptual’ is not taken to denote an ontological category but instead provides a functional classification. The idealism advocated in the text is not one according to which the world is composed of concepts or Berkeleian ideas but one according to which reality is conceptually structured and must be understood to be so. The virtue of this is that there is no ‘gulf of intelligibility’ between mind and world (p. 51). When we grasp a fact, we aren't forcing it into a conceptual structure that it doesn't already possess. The sceptic who worries about such a gulf has failed to understand the nature of semantic content (p. 95).
1. Space, Time and Anaphora
At times, Brandom seems to accept that incompatibility depends upon the prior existence of a system of spatiotemporal relations: ‘The experiences we label “red” and “green,” and those we label “rectangular” and “triangular,” are experienced as incompatible, as ruling each other out (as simultaneously located), while those labelled “red” and “triangular” and “green” and “rectangular” are experienced as different, but compatible’ (p. 140).
The notion of things being ‘simultaneously located’ is doing serious work here. If a system of spatiotemporal relations is needed to make sense of the experience of incompatibility, then it would seem that we should think of such a system as non-conceptual on pain of regress. At the least, incompatibility needs a medium and if to be conceptual is to stand in relations of incompatibility, it would seem that we need that medium to be non-conceptual.
Anyone who embraced such a medium might reasonably think that ‘[s]patio-temporal position provides the fundamental ground of distinction between one particular item and another of the same general type, hence the fundamental ground of identity of particular items’.Footnote 6 If the extensionalist can help themselves to non-conceptual spatiotemporal structure, then they should be able to help themselves to a notion of ‘mere difference’ as well. Objects would be ‘merely different’ on account of their different spatiotemporal locations. It's not implausible to assume that the idea of ‘mere difference’ grounded in spatiotemporal difference is what is in the back of the extensionalist's mind when they write about domains of ‘merely different’ objects.Footnote 7 The merely different objects could be abstractions from the objects of our spatio-temporally structured experience. While angels, arhats, and artificial intelligences might not have spatiotemporally structured sensory perception, humans do, and we don't experience conceptualised properties like ‘redness’, without those properties having a location in our visual field.
2. The Anaphora Argument
Brandom acknowledges that ‘incompatibilities among features require units of account’ (p. 150) but doesn't seek these units of account in the spatiotemporal structure of sensory perception. Instead, he looks downstream to the structure of anaphoric chains to provide identity and difference. The key claim here will be familiar to readers of Making it Explicit (or, it seems, The Phenomenology of Spirit); deixis presupposes anaphora.Footnote 8
The argument is as follows. Different tokenings of indexicals pick up on different contents. Every time you say ‘now’, you pick up on something new and something which can't be repeated by saying ‘now’ again. To be cognitively significant, the contents of these indexical tokenings must somehow be made repeatedly available. That is, if we are to use the contents as premises in inferences, antecedents in conditionals, objects of negation, if they are to play any cognitive function for us, there must be some mechanism for ‘picking up’ the content of the original demonstrative tokening. Without the possibility of this ‘picking up’, a ‘now’ would be like an ‘ouch’; a gesture but ‘not a move in a language game’ (p. 128). Such a mechanism would give us the means to hold on to the content of an experience while no longer having that experience. Hence the slogan, ‘deixis presupposes anaphora’. ‘Demonstratives can sensibly be used only when there are anaphoric pronouns available to pick them up and use them, and so give their epistemic authority some significance for the rest of thought’ (p. 127). While ‘now’ and ‘that’ can't be used repeatedly to track the same content, ‘then’ and ‘it’ can be repeated in chains.
Rather than looking to spatiotemporal location to serve as the locus of incompatibility, Brandom finds this locus in anaphoric chains. Brandom is effectively inverting the idea that objects presuppose space and time by arguing instead that, if we have a grasp of demonstratives, then because content must stand in incompatibility relations, we have a grasp of objecthood.
One consequence of having these ‘units of account’ implicated in any demonstrative activity is that the distinction between object and properties is thereby implicit even in feature-placing language. Reminder: ‘feature-placing language’ is Strawson's term for statements that locate universals without containing any reference to particulars.Footnote 9 Standard examples are ‘now it is raining’, ‘there is water here’, and ‘there is gold here’. According to Strawson, feature-placing language is contentful and indexical but makes no reference to objects. If Brandom is right, it cannot be indexical without presupposing anaphoric chains and once we have these chains, we have something that can serve as loci to repel incompatible commitments and combine compatible ones. In doing this, these chains perform the logic function of objects. I can't say of that cat that it lays eggs but I can say that it is purring. The anaphoric chain, rather than a spatiotemporal location, is the site at which we combine properties and acknowledge incompatibility. To be clear, Brandom isn't saying that objects simply are anaphoric chains. The book develops the idea that what a representing represents exercises a kind of authority over the correctness of that representing. What it is about is what it is responsible to (p. 300).Footnote 10 But it seems that anaphoric chains are needed for incompatibility.
So is this successful? Can Brandom accomplish with anaphoric chains the task for which others have invoked space and time? What is the ‘holding on to’ that connects a demonstrative and anaphoric pronoun? This is less clear.
Brandom lies firmly within a tradition that blends linguistic and phenomenological terminology. At times, the book discusses the anaphoric ‘relation between a pronoun and its antecedent’ (p. 129), at other times, it is concerned with relations between utterances (p. 128), and at other times, it's concerned with sensory episodes: ‘The authority of any immediate sensory episode depends on its being situated in a larger relational structure containing elements that are not immediate in the same sense’ (p. 128). Experience is always the linguistic expression of experience and this makes it a harder to follow what exactly is being claimed.
When Kantians invoke space, they seem to mean something with the well-studied properties of Euclidean geometry. Objects can be identified according to their spatiotemporal location (and objects with different locations and histories are distinct). It doesn't look like Brandom is trying to replace geometry with syntax or claiming that syntactic structure is a necessary condition for the possible appearance of objects of experience. This is good because, if the relational structure that underlies our experience of objects was just syntactic structure, then we would simply have replaced the spatial form of outer intuition with an inner form of linguistic intuition, i.e. the syntax of our languages. We certainly might be able to individuate and re-identify objects within this structure, but it would look like we are saddled with an appeal to another non-conceptual structure to make sense of incompatibility. Similarly, if Brandom were proposing some kind of file-change semantics in order to make sense of how anaphoric pronouns can retoken demonstratives, he would be presupposing a set of (mental) objects and a structure in order to explain this.
The text’s discussion of the relation between demonstratives and anaphora is rather dark but doesn’t appeal to any syntactic notions (e.g., c-command, government, binding). In fact, it takes the relationship between a demonstrative and an anaphoric re-tokening to be a negation of a negation! It is ‘the mediation that articulates their immediacy’ (p. 510), ‘the negation of the negation of immediate unrepeatable being’ (p. 127). When an anaphoric pronoun is used, it exists in the present as a negation of the past demonstrative act but it doesn't negate the demonstrative as a new demonstrative tokening would (e.g., saying ‘now’ and then saying ‘now’ again a minute later). ‘The past, which is the truth of the future, the only reality it has, is a negation of the present. But this negation is in turn negated’ (pp. 126–27). The claim appears to be that the pronoun affirms that the demonstrative from which it draws its content and authority is past. The result is that, in the present, we simultaneously affirm that the demonstrative moment has passed (a negation) while negating this negation to co-opt the content of this demonstrative tokening. To use a pronoun that is anaphorically co-indexed to a demonstrative is to use an expression that performs two moves, it affirms that the context which fixes the content of the demonstrative has passed while simultaneously making that content available for further predication. Brandom calls this act recollection. By affirming that the content of the demonstrative is not present, we capture that content as a repeatable unit.
If we can make sense of the relationship between demonstratives and anaphoric pronouns without appealing to a structure that is fundamentally different to the conceptual structure produced by relations of incompatibility, then it may be possible to make sense of incompatibility in a non-circular fashion. If, and it's a big if, this account of the relationship between demonstratives and anaphoric pronouns (whether in language or thought) can be made to work, then Brandom may have an account of ‘units of account’ for our ascriptions of incompatibility which does not rely on spatiotemporal structure and which can in turn underlie our grasp of objects.
The arguments discussed here last several pages in a book that is over 800 pages long. I have not touched up the book's rich and complex philosophy of action, the social account of normativity and the role that desire plays within it, the discussions of faith and trust, forgiveness and Vernunft (and nor have I speculated whether this really is an accurate reading of Hegel). At the most, I hope to have highlighted to the reader some of the complex, interesting and ingenious material that this text has to offer.Footnote 11
