A lecture series taught in 2020. The course website is here.
Does ‘representation’ have a nature or a history?
Does it belong with “electrons” and “sulfur” or with “freedom” and “love”?
Brandom, being a Hegelian, thinks philosophy belongs in the latter category and proceeds to give a historical overview of ‘representation’.
Representation a modern concept.
Premodern theories understood relation between appearance and reality in terms of resemblance (sharing properties)
The rise of science made this untenable:
Copernicus: reality behind stationary Earth and revolving Sun is a revolving Earth and stationary Sun.
Galileo: effective strategies of understanding time as line lengths and acceleration as triangles … not easily understood in terms of shared properties.
Descartes invents representation with relationships between algebra (representation) and geometry (represented/reality).
Spinoza talks better about how Descartes used representation than Descartes himself. Spinoza saw Descartes’ philosophy as being understood in terms of his innovations of relating algebra to geometry.
“The order and connection of things is the same as the order and connection of ideas”
The properties of what is representing or represented are irrelevant, only that there is a mapping between the relations in each world.
If any things are known representationally, then some things must be known nonrepresentationally (to avoid infinite regress)
Representationalism gives rise to choice between epistemological skepticism vs foundationalism.
A Kantian insight: the real problem is semantic skepticism (can we really know something by representing it correctly?)
Kant says Descartes was right to think in terms of representation but that he didn’t distinguish two different kinds: picture like images/sensations vs sentence-like thoughts.
He saw both as different ends of a spectrum, while empiricists tried to reduce thoughts to pictures and rationalists vice-versa.
Spinoza’s interpretation of Descartes gives another view: within a representational picture, empiricists are atomists whereas rationalists are holists
Brandom’s interpretation using orders of explanation: empiricists treat representation as a primitive and infer reason-relations in terms of it. Rationalists treat reason relations as primitive, explain representational content in terms of inferential relations.
Rationalist Leibniz would have us understand the content of the map as the inferences that someone who treats it as a map could make about terrain facts (e.g. a river) from map-facts (wavy-blue line).
Sellars identifies both camps as descriptivists (to be conceptually contentful is just to describe / represent how things are).
Empiricists start with narrow postulate about what representing is and exclude a lot of genuinely contentful thought due to not meeting this standard (e.g. ethics, modality)
Rationalists take all our cognitively contentful expressions as therefore being part of the actual world, resulting in ontological extravagance (postulating objective values/universals/propositions/laws)
Sellars saw the Tractatus as teaching us how to get beyond this ideology with the case of logical vocabulary
Representation is a wider concept than description - Brandom thinks that Sellars’ anti-descriptivism is a form of anti-representationalism.
E.g. proper names represent without being descriptions, in Naming and Necessity.
Representation is a holistic conception, so rationalists were right about that (think: categories, relationships over properties)
Representation/description involve subjunctively robust relations between representings and representeds.
Considering the inferences of map facts to terrain facts, we also must accept that if the terrain were different, the map fact would be different.
Related to Fodor’s account of representation in terms of “one-way counterfactual dependencies of ‘horses’ on horses”.
Representation has a normative dimension
To treat representation as concerning what inferences we can make is a normative order.
Hegel appreciated this: to count as representing something is to be responsible to the represented thing (what is represented provides the normative standard for correctness of representing). What is represented has an authority over what is representing.
Declarativism: a relatively defensible representationalist position
Dual to “descriptivism” which too narrowly construes representation as description (which is too constricted a notion of representation).
This too broadly understands what all declarative sentences do in terms of fact-stating / truth-aptness (‘representation’ becomes too expansive). Expressivism is one way of negating this (by declaring “X is good”, one is commending rather than fact stating).
Intuition: the question of truth can be raised for whatever is expressed by declarative sentences.
Geach’s 1960 embedding argument: “If X is good, then …” a hallmark of fact-stating (if we had close the door or praise god, it wouldn’t make grammatical sense)
We can embed moral statements that 1st wave expressivists said were not fact-stating.
QUESTION: are there any grammatically declarative sentences which cannot be embedded? Would a counterexample have to be a non-truth-apt, declarative sentence?
Declarative sentences that may not fit into the fact-stating mold of “the frog is on the log”:
Logical (e.g. negative/conditional facts), modal (e.g. necessity), probabilistic, semantic (what expressions mean or represent), intentional (possibly about non-existent objects like golden mountains / round squares), normative, abstract / mathematical.
Are these all types of facts? Do they represent features of the world?
Rorty characterizes pragmatism as fundamentally anti-representationalist (Cheryl Misak strongly disagrees and considers Rorty to be a false heir of the tradition)
Representationalism is an ideology - that the meaning of thoughts/talk should be principally understood in terms of representational relations the thinkings/sayings stand in to what they (purport to) represent.
It’s a crippling ideology that must be rejected wholesale, no hope of redemption.
It’s synonymous with modern philosophy, so that must be jettisoned too.
Two sides of Wittgenstein:
Tractatus = representationalism (but providing the model for moving beyond it w/r/t logical vocabulary)
Logical tradition from Frege/Russell, operative paradigm of formal calculi for artificial symbolic languages.
Possible world semantics best distillation of its representational approach to meaning
Investigations = anti-representationalism.
Anthropological tradition focuses on natural languages, in tradition of Dewey. Rorty claims Heidegger also in this tradition, which both sides (pragmatists and Heidegger allies) don’t like. Focus is not on meaning but on use.
Arguments:
(Philosophy and the Mirror of Nature) - representational in semantics leads to an unproductive oscillation in epistemology between skepticism and foundationalism
Pragmatism about norms
antiauthoritarianism argument (completing the emancipatory project of the Enlightenment)
Simon Blackburn applies local expressivism in order to get a distinctive flavor of antirepresentationalism.
“Fact stating discourse” can be regarded as crucially important without dismissing all other kinds of discourse as defective/reducible.
Expressivism is a broad family of views claiming some areas of discourse are ‘in the business’ of giving expression to sentiments/commitments/non-cognitive or non-representational mental states or attitudes.
Huw Price synthesizes these two strands of anti-representationalism.
Agrees with Rorty that representation should never be used to do substantial explanatory work. They are global antirepresentationalists, which is a radical position currently.
He unites original German global expressivism (beginning with Herder) with second wave local expressivism of Blackburn and Gibbard.
rejects local expressivism, which requires us to distinguish vocabularies which should be given representational analyses or not - however he argues this cannot be done without embracing declarativism which the local expressivists wanted to avoid.
He reads into later Wittgenstein to unite global expressivism with Rorty
involves distinction between traditional object naturalism (how can we reduce facts in terms of natural science truth-makers) and the pragmatist’s subject naturalism (only seeks that reduction for the discursive practices consisting of use of language).
Brandom argues this global antirepresentationalism goes too far - prioritizing use over meaning (i.e. semantics answers to pragmatics) does not rule out representational/descriptivist accounts of vocabularies in general.
Disagrees with Pryce’s argument that local expressivism is not possible.
Neokantianism vs socialized/historicized/naturalized alternative.
Struggle between two camps in the 19th century, which is then re-enacted in the 20th century
Neokantianism (Marburg / Freiburg to CI Lewis and Carnap). Philosophy has sovereign authority.
Idea originates with Plato, but Descartes/Kant/representationalism are modern versions.
Neohegelianism (Hegel socialized philosophy, Marx naturalized it). Philosophy brought ‘down to Earth’.
Bertrand Russell and Husserl found things for philosophy to be apodictic about. Basis of Rorty’s ‘astonishing’ claim that analytic philosophy is just a phase of neokantianism
Russell and those downstream don’t think of themselves this way.
But they share the idea that philosophy of language is “first philosophy” and that linguistically-inflected philosophy of mind could advance our notions of epistemology and general theory of representation.
Shared emphasis on understanding language semantically, distinct from understanding knowledge epistemologically.
Husserl subject to Sellars’ critique of the Myth of the Given, Russell subject to Quine’s critique of Myth of the Museum. (Carnap subject to both).
“Uncritical semantics is the myth of a museum in which the exhibits are meanings and the words are labels. To switch languages is to change the labels”. Against the museum myth, Quine argues here for the indeterminacy of “meaning” and translation.
These are both (semantic) holist arguments.
Rorty regards the pragmatist/holist Quine+Sellars rebuttals as bringing us back to Dewey’s socialized/historicized/naturalized philosophy, with the 20th century being a pointless, big detour (despite increase in technical advances in logic/language).
Two kinds of privileged representations: sense impressions and representings of meaning (meanings thought representationally instead of functionally)
Rorty sees common genus to analytic meaning-articulating claims and the sensory given: privileged representations
common origin: role as infinite regress stoppers for the two types of representations, sensuous intuitions and inference-licensing concepts, that Kant distinguished (which each have these relations of privilege/authority).
Agrippan trilema of alternatives to skepticism
justification always in reference to other claims which require justification (this can be circular or infinite regress) OR there are unjustified justifiers / foundations.
If we have foundations, we need two kinds of regress stoppers:
premises that have authority
inferential transitions that have authority
These are not the same as premises, c.f. Tortoise and the Hare
That the “privileged representations” are authoritative and immediate, a kind of atomism follows from their privilege.
In order to play their epistemically-privileged roles, the regress stoppers must be semantically-privileged insofar as not depending on any collateral epistemic commitments.
Quine and Sellars’ critiques render such semantic privilege as practically unintelligible. A requirement that cannot be fulfilled.
Holist/pragmatic arguments of Sellars (against sense-givenness) and Quine (against meaning-givenness) that connect epistemology and semantics.
Sellars in EPM
Locke got confused by conflating the causation of a belief from its justification.
Sense data may be causally prerequisite to knowledge, but it cannot justify belief (it’s not conceptually contentful, in the sense of standing in reason relations of implication/justification).
To stand in reason relations requires lots of other infrastructure, such that atomic sense data cannot do this on its own (this is a semantic holist argument)
How do we acquire knowledge in the semantic holist POV?
How to get into the game of giving and asking for reasons? “The light dawns slowly over the whole” We need to get good enough at making the ‘right’ moves (as judged by ‘competent speakers’ before counting as a ‘competent speaker’.
Quine in TDE
Target: analytic truths, e.g. “cats are mammals” supposedly not depending on any other commitments but rather immediately from the meanings of the words.
What is the practical difference between these truths and very general facts, such as “there have been black dogs”.
Duhem-Quine thesis: “it is impossible to test a scientific hypothesis in isolation, because an empirical test of the hypothesis requires one or more background assumptions”.
What inferences we are allowed to make depends on the whole of collateral beliefs we have.
The unit of meaning must be the web-of-belief rather than the concept or the sentence.
Fodor argues against this, considers mixing epistemology and semantics to be a big philosophical mistake initiated by Quine.
Is Rorty attacking a strawman? Empiricism / Epistemic foundationalism is not popular now:
“Default and challenge” structure a way of avoiding Agrippan trilemma.
Bayesianism sees all justification as comparative (never have to justify one’s prior commitments).
though the problem of semantics is still open if it is to not be representationalist
Naturalism + representationalism remains unchallenged.
picking an ontologically privileged base vocabulary
So Rorty and Pryce need to do one or both of:
Argue independently for pragmatism, then use that to attack representationalism without passing through foundationalism
Argue independently against representationalism and propose pragmatism as the best alternative approach
(From the discussion of truth as “what is best in the way of belief” as opposed to correspondence with reality):
How the combination of:
declarativism
blurring out all distinctions of kind of claimable
expressivism
a local expressivism about what one is doing in attributing truth—namely not describing the claimable, but endorsing it—
underwriting a global antirepresentationalism because of special properties of the vocabulary of truth
together underwrite a Jamesean understanding of truth-talk.
This role for an expressivist move in a pragmatist argument forges an important link between the first and the second halves of this course.
(From the discussion of representation):
What is really at stake in the battle between a representational model of the content of expressions and a pragmatist model is the best order of explanation (a way of thinking about conceptual priority) between representational relations and reason relations (of implication and incompatibility).
Davidson teaches us that and how taking reason relations as primary (the pragmatists says, because giving and assessing reasons, implicitly and practically appealing to justificatory reason relations, specifiable in a deontic normative vocabulary of “commitment” and “entitlement”’) holistically determines representational relations in top-down explanatory stories.
Representationalists are committed to atomistic objective usually causal relations (specifiable in an alethic modal vocabulary) determine reason relations.
Rorty takes for granted a distinction of Dewey:
| Platonism | pragmatism |
|---|---|
| Principles | practices |
| Theoria | phronesis |
| Knowing that | knowing how |
| End the | Continuing the conversation |
Platonists look for a principle or rule, something explicit or that could be made explicit, behind every implicit propriety of practice.
Pragmatists argue that explicit principles or theories float on a vast sea of implicit practical skills.
For example, a cobbler can make good shoes. The Platonist looks for what form is behind his mastery, what principle / mental representation makes it possible that the cobbler does that? The pragmatist treats the skill as prior to the principle.
Rorty sees representationalism as the distinctly modern form of Platonism as described above.
Kant is the avatar of this form of representationalism: representations and rules are two sides of the same coin. He puts principle over practice (opposed by Dewey, who was followed by Heidegger and Wittgenstein in this respect).
This is Rorty’s way of talking about language games. Or paradigms (in the Kuhnian sense - Kuhn was writing just down the hall from Rorty). Vocabularies are what we deploy in discursive practice.
He thinks is needed as a successor notion to the idea of languages and theories, which was rightfully taken apart by Quine’s Two Dogmas of Empiricism.
Examples, how big are they? - on the one hand, the vocabulary of 16th century theology - also, the vocabulary of modernity - where that’s presumably an autonomous discursive practice
Philosophers think that they are debating about what is true in a given vocabulary when really their question is better framed as trying to pick the right vocabulary (important to realize this, since it changes how we compare arguments).
This point appears to be accepting the Carnappian distinction of theory and
Rorty makes fun of representational realists as being committed to the idea that there is a thing such as “nature’s own language” / “nature’s own vocabulary”
He sees Lewis as committed to this as the language which determines what the ‘natural properties’ are.
Rorty doesn’t draw a line between speaking in a vocabulary and changing your vocabulary (he thinks almost all speech acts change it).
Duhem point: if we acknowledge that your meanings at least partially determine what inferences are good. What follows a given sentence depends on what auxiliary hypotheses you’re allowed to use as collateral premises for
We can see what else you’re committed to has an effect on that sentence, given the meaning of that sentence is characterized by its inferential relations.
In mature sciences a lot of work is taken to allow for discourse to proceed as if the vocabulary were fixed
The discursive equivalent of “clean rooms”, maintained through heroic social disciplinary measures
This is for mature sciencies: if you think of the history of temperature, every single time a new way of measuring temperature was discovered, the concept changes
But it would be a serious mistake to take this extreme, artificial case to be the paradigm on the basis of which we understand the use of language in general. (Here we might think of Heidegger on the effort it takes to precipitate Vorhandenheit out of Zuhandenheit.
Rorty says: “On the pragmatist account, a criterion (what follows from the axioms, what the needle points to, what the statute says) is a criterion because some particular social practice needs to block the road of inquiry, halt the regress of interpretations, in order to get something done. So rigorous argumentation-the practice which is made-possible by agreement on criteria, on stopping-places - is no more generally desirable than blocking the road of inquiry is generally desirable. It is something which it is convenient to have if you can get it. if the purposes you are engaged in fulfilling can be specified pretty clearly in advance (e.g., finding out how an enzyme functions, preventing violence in the streets, proving theorems), then you can get it. If they are not (as in the search for a just society, the resolution of a moral dilemma, the choice of a symbol of ultimate concern, the quest for a”postmodernist” sensibility), then you probably cannot, and you should not try for it. The philosopher will not want to beg the question between these various descriptions in advance.”
The generation of new vocabularies.
This is the essence of discursive practice, is to be committed to a view of conversation as something to be continued.
But one can do other things with vocabularies than use them to describe. So “redescription,” though evocative, might be replaced by “recharacterization,” or “reconceptualization.”
“Pragmatists follow Plato in striving for an escape from conversation to something atemporal which lies in the background of all possible conversations”
Conversation is about redescribing our vocabulary as much as using it. It is the process that produces redescriptions.
Quantifying over all possible vocabularies is a temptation and something you would only attempt to do if you are trying to end all conversation. This is a fundamental mistake. - This is something the early Wittgenstein did in the Tractatus - He then learned not to do this. It’s related to his later view that language is a motley. - Dummett interpretation of Wittgenstein: - Wittgenstein would think that if there were any use for a philosophical notion of meaning, the point of having a notion of meaning would be to codify proprieties of use. - But there is no limit of things one can use language for, so we cannot systematically find all meanings of all expressions. - Wittgentstein tool analogy: - you would think that you could describe the different ways of using things in terms of what you do with them in the way you could tools, so that you could think of hammer and nails, screw and screwdriver, glue and glue brush, all his ways of attaching things to one another. And that would be a sort of common function that they could perform - Early W. thought ‘Yes, representation is like that. That’s what language is for’ - What about a wrench What about the pencil that the carpenter uses, or the level that the carpenter uses, or the tool belt or tool chest? Or the set of plans that they’re using? - All these are functioning differently, and there isn’t going to be a systematic way of saying all the different kinds of tools that you could have - Classic Wittgensteinian anecdotes turn on the malleability of language. (tooth example) - The metaphyiscal puzzle comes from having a STATIC, totalizing pictures of language rather than accepting it as a motley that evolves whenever you use it. - We carefully design mature natural science and math to not have this happen, but this should not be a model for how our language works. - This is why Wittgenstein is a semantic nihilist, he doesn’t think there are actually meanings. The plasticisty of language makes this impossible.
Rorty has another term that is part of the constellation that starts with “vocabulary” and includes “redescription” and “conversation.” It is “coping.” It is his generic term for what we do with vocabularies, generally. It is in terms of success at coping that we are able sometimes to assess and compare vocabularies as better and worse. In that regard, it plays a role analogous to the notion of accuracy of representation, that Rorty wants to persuade us to discard as specifying the dimension along which arbitrary vocabularies can be assessed as better or worse.
It is crucial to this notion of coping that standards for it are rigorously internal to the vocabularies being assessed.
Different kinds of facts are identified individuated by the vocabularies we use to state them. (e.g. Physical Facts, normative facts, nautical facts,…)
Rorty concludes if it doesn’t make sense to quantify over all possible vocabularies, then it doesn’t make sense to quantify over all possible facts (a new vocabulary is going to make it possible to state new facts)
Rorty rejects the distinction of objective and subjective facts.
(Stopped taking notes at 1:11, pg 6 of Brandom’s presentation notes)
Is there a bifurcation in ordinary empirical description vocabulary vs vocabularies where there isn’t a good correspondence between parts of the sentence and parts of the world? -E.g. “Jupiter has moons” vs “The universe is infinite” and “Love is the only law”
Representationalists can either 1.) postulate objects represented by the latter claims (e.g. love) or 2.) consider such sentences that don’t fit the representational mould as defective/inferior.
These latter claims definitely have a meaning though, as observable by the reason relations they stand in.
Brandom claims that the representationalism-vs.-antirepresentationalism issue is distinct from the realism vs antirealism one, because the latter issue arises only for representationalists.
Is “coping” talk just an evolutionary biology / memetic thing?
This ‘reductionist’ interpretation is not true for Rorty:
Language is not a tool, as Dewey would have it, though it’s a nice metaphor for some purposes, it can be stretched too far:
Tool requires a common purpose that you can compare different tools for (e.g. nails / glue / screws all are tools for sticking wood blocks together)
But we cannot formulate the goal of language without already having language (we don’t have access to “nature’s true language” to do this, either).
Thus the meaning coping must be within to some vocabulary.
Brandom summarizing Wittgenstein: ‘Meaning’ is a theoretical object-kind, postulated to codify proprieties of practice
it is no truer that: - “atoms are what they are because we use ‘atom’ as we do” - than that “we use ‘atom’ as we do because atoms are as they are.” - Both of these claims, the antirepresentationlist says, are entirely empty. - Both are pseudo-explanations. - It is particularly important that the antirepresentationalist insist that the latter claim is a pseudo-explanation.
The most general lesson of the discussion of Davidson is that the overall collision is between
the intrinsically holistic demands of reason-relations, understood in interpretivist terms by Davidson, and (also) in social-practical terms by Rorty,
Interpretism: to say someone is a believer is to invoke the possibility of interpreting their beliefs and actions together (in a way that maps onto our own beliefs)
While having a conversation with someone is how you learn what they mean by X, the fact you can have a conversation is what it means for them to mean something by X.
To say that someone believes something is to claim we can have a conversation with them.
and the claims representational relations determined independently of social practices of giving and assessing reasons, for instance (and paradigmatically) by objective, causal relations describable in alethic modal terms
Davidson is happy to impute extensions (referents) as an intermediate stage of interpretation. But he insists that the process be top-down, starting from reason-relations to assignments of referents.
Davidson’s big contribution:
flipping Tarski’s theory of truth:
Tarski: if you take meanings fixed, I can give you a recursive theory of which statements are truth.
Flipped: You can take ‘meanings’ to be the truth conditions (of all of the language). The starting point is the reason relations, and from those we derive the meanings.
We can argue about which order of explanation is better. Bottom up vs top down.
Bad theories
E.g. witches, phlogiston
We often find ourselves saying “Hard to say whether they’re talking about real things but are wrong about most of them or not talking about real things”
To what degree do ‘witch’ and ‘phlogiston’ refer?
It’s a matter of degree and a pragmatic decision.
So any theory that has as a consequence that there is a precise line between reference or non-reference is wrong.