I'm going to talk about Wilfred Sellers and I'm going to try to explain, well, there's a, in a way, there is a continuity between this, my current work on Sellers, which I hope will result in a book in about a year or so, and the work, the book I published a decade ago. My book opens with an account of Sellers' distinction between the manifest and the scientific images of man in the world. And this provided me with the basic framework I wanted in order to formulate the problem of nihilism.
After writing the book and reading more of Sellars' work, I came to realize that of all the philosophers I discussed in the book, he was the one I had, I think, whose work I had least understood or whose work I had skimmed over most superficially. And I realized that understanding the proper nature of his philosophical system forced me to reformulate the questions that I had tried to raise in that book. So this is why I've spent the past several years trying to understand Sellers' work. So I hope this is a kind of a justification for what follows.
So the paper's got three parts. First I'm going to give an account of Sellers' transcendental naturalism. Then I'm going to discuss Robert Brandom's critique of Sellers' naturalism. Brandom is a philosopher who is deeply influenced by Sellers. His whole inferential semantics was sketchily formulated in Sellers' own work. It's given a systematic articulation in Brandom's work. But Brandom rejects Sellers' commitment to naturalism and to scientific naturalism. And finally, I'll close with an account of the... Defend Sellers' naturalism against some of Brandom's criticisms, or at least try to.
And try to emphasize the importance of the distinction between practices and processes. processes, between discursive practices and material processes, which I think underlies this whole kind of disagreement between Brandom and Sellers. Sellers is a transcendental philosopher, and the transcendental perspective alters the conditions in terms of which the issue of realism is framed. Transcendental philosophy can be contrasted with dogmatic rationalism on the one hand and skeptical empiricism on the other. The dogmatic rationalist assumes that the mind enjoys a priori cognitive access to mind independent reality and believes that reason can deduce the fundamental features of that
reality. The skeptical empiricist for her part insists that all knowledge is rooted in but limited by sensory experience and denies that reason can deliver a priori knowledge of mind independent reality. The transcendental philosopher rejects both dogmatism and skepticism. She rejects the rationalist assumption that the mind enjoys a priori cognitive access to reality, just as she rejects the empiricist claim that all knowledge derives from sensory experience. Both dogmatism and empiricist skepticism remain beholden to what Sellers calls the framework of givenness. Now, Sellers is famous for his attack on the myth of the given. But this attack is misunderstood.
It's often reduced to a critique of a kind of naive positivistic empiricism. And this is not the target of Sellers' critique. What he calls the whole framework of givenness applies both to certain kinds of rationalist metaphysics and to, so I would argue, certain kind of strains of Husserlian and post-Husserlian phenomenology. So if he rejects the framework of givenness it's because dogmatism takes the correlation between thinking and being as given and skepticism takes the intelligibility of sensory experience as given. At the heart
of the framework of givenness is the assumption common to empiricism and rationalism that mental states are self-intimating. And to reject this framework is to refuse the assumption that the mental is self-intimating. This means that minds do not necessarily know themselves. There is a fundamental difference between thinking and knowing what one is thinking of. By the same token there's a fundamental difference between sensing and knowing what is sense. The awareness of something is not the awareness of something as something. And this difference between thinking and thought or sensing and sense falls from the rejection of the framework of givenness. I think it is the transcendental difference for Sellers.
He recasted the transcendental difference as obtaining between thinking and thought or sensing and sense. The core feature of the framework of givenness is the premise that the fusion of thinking and thought or of sensing and sense is guaranteed either by intellectual intuition in the case of rationalism or sensory intuition in the case of empiricism. But this intuitive identity, whether intellectual or sensory, immediately generates metaphysical divisions such as between mind and nature, experience and theory, immediacy and mediation, The rejection of this fusion raises the spectre of representation. Since thinking stands to thought and sensing stands to sensed,
as representing stands to represented. And notwithstanding all the critiques of representation in 20th century clauses, specifically those of Heidegger and Deleuze, I think Sellers' account of representation circumvents both those critiques. This is not a Cartesian conception of representation. It's a radicalized Kantian account. The transcendental question then becomes, what is the nature of the connection between thinking and thought or sensing and sense? More generally, how are we to understand the connection between representing and represented? The problem of representation arises from the rejection of metaphysical fusion, whether of thinking and thought or sensing and sense. But it is not the problem of the division of substance, the one and the many, or of the connection between divided substances, whether thinking or extended, animate or inanimate.
The problem of representation is not the problem of how a thinking substance can somehow reach out into extended substance and retrieve a representation of an object or some features or characteristics of extended substances. It is the problem, the problem of representation is the problem of the transcendental torsion between the represented inside and the representing outside. It's a problem of transcendental exteriority. This torsion is the focus of what Sellers calls transcendental logic, which precisely suspends the metaphysical reification of the difference between mind and nature. There's a quote from Sellers' 1966 paper,
Some Remarks on Kant's The Year of Experience. The task of transcendental logic is to explicate the concept of a mind that gains knowledge of the world of which it is a part. The acquisition of knowledge by such a mind involves its being acted on or affected by the objects it knows. Rejecting the framework of givenness means relinquishing the assumption of a resemblance between the categories of thought, the structure of representing, and the order of nature. Our understanding of nature is discursive, not intuitive. This is Seller's Cantinism. In other words, because it's discursive, it represents nature. But this representation of nature,
minimally defined as a spatiotemporal continuum, and I know that there's a whole problem about how successfully one can avoid smuggling in metaphysical assumptions into any kind of characterization of nature. But nevertheless, let's try and stick to a kind of a minimal definition of the spatiotemporal continuum. It's conditioned by the way nature appears to us both spatially and temporally. And since what is empirical is coextensive with spatiotemporal appearance, nature is cognizable to the extent that it is empirically representable. But, says Sellers, not every system of empirical representables constitutes nature,
but only that system of empirical representables, the representings of which would be true. Transcendental logic then aims to uncover both what it is for something to be an empirical representable and what it is for something to be a true representing. Understanding the former involves grasping the way in which thinking interacts with sensing, while understanding the latter requires grasping the nature of the connection between truths pertaining to what is conceptually represented and truths pertaining to non-conceptual representings. We need this distinction between truth at the level of the represented and truth at the level of representing.
No single account of truth can, you know, the mistake is to kind of focus on one at the expense of the other. So, this distinction between representing and represented is not the difference between two separate things, but the formal distinction between, let's say, act and content. And this yields the distinction between objective reality, reality imminent to the represented, and what can be called reality in itself, i.e. non-represented reality. But non-represented does not mean non-representable. The distinction between objective or represented reality and non-represented reality or reality in itself is epistemological and not ontological.
So this distinction between represented and non-represented or between what is for us and in itself is contained within the eminent distinction between representing and represented. And the relation of analytical dependence between represented and representing renders the objective reality of the represented content conditional upon the formal reality of the representing act. Now here Sellers is basically drawing on what he thinks is the scholastic distinction which underwrites Kant's distinction between appearance and in itself. In other words, the objective reality is reality as the correlate of some kind of mental act, whereas the formal reality is substantial reality.
It's independent. It's the reality that substantial form has. Since every represented implies a representing, the objective reality of a represented entails the formal reality of the representing through which it is represented. Now, this argument establishes only that if there are represented, then there must be representings in themselves. Not that there actually are such representings in themselves, or that we can know either them or non-representings in themselves, because they're also non-representing in themselves. Transcendental skepticism insists on the possibility of a complete disconnection between represented
contents and the in itself, whether construed in terms of representings or non-representings. And Sellers now, I think, has two arguments, I'm summarizing, but I think he has two basic arguments against transcendental skepticism. First, we can question the truth of some of our beliefs, or we can question the kind of the adequacy of some representatives by holding others to be true, but we cannot question the truth of all our beliefs at once. What cannot be represented is the falsity of every represented. This is Sellers' in other words, I think, most powerful argument against a kind of radical skepticism is that it has to appeal to a kind of a version of the myth
of the given where you can somehow imagine the possibility of appearances, you can imagine the intelligibility of appearances that have no objective correlates. For sellers, this supposedly appeal to a fundamental stratum of appearance uncontaminated by any taking of appearances to be thus and so. An account of seeming uncontaminated by any claim that something seems or a representative that is not of something or of something as something. And I think the Szilagyi's claim is that no such stratum of appearance is even conceivable
because the appearing is always the appearing of something. The second strand in his argument against transcendental skepticism is the argument from what Sellers calls double affection. Okay, and this is the one I'm going to focus on. And Sellers' proposal is that the transcendental exteriority of things in themselves as grounds of appearances has an analog in the empirical exteriority of physical objects as causes of sensations. and here's a quote from from Sellers where he proposes to where he sets out his strategy which is the analogical reclamation of the in itself Sellers writes in Science and Metaphysics from 1967
even if we attribute to Kant the view that things in themselves are analogous in structure to the world of appearance the analogy would for him be one which could only be cached by God much as according to traditional theology only he can cache the analogies in terms of which we attempt to conceive him. God would have a non-analogical grasp of things in themselves by virtue of the fact that his intuitive representations are not passive but are the very volitions by which they are created. It's precisely God's intellectual intuition creates or generates the objects which it intuits which is why the distinction between particular and universal doesn't exist for him. He can grasp everything in its infinite complexity.
So, but Sellers continues, the thesis I wish to defend but not ascribe to Kant is that although the world we conceptually represent in experience exists only in actual and obtainable representings of it, we can say from a transcendental point of view, not only that existence in in itself accounts for this obtainability by virtue of having a certain analogy with the world we represent, but also that in principle we, rather than God alone, can provide the cash. For as I see it, the use of analogy in theoretical science, unlike that in theology, generates new determinate concepts. It does not merely indirectly specify certain unknown attributes by an analogy of proportion.
One might put this by saying that the conceptual structures of theoretical science give us new ways of schematizing categories. So, we supply the conceptual currency for constructing the analogy between the determinations of appearances and the determinations of the in itself. And this construction is the analogical schematization through which non-representings in themselves can account for the obtainability of conceptual representings. This obtainability, I think, is the key term for sellers. So how much, how might such schematization proceed? Things in themselves are not the causes of appearances in the sense in which electrostatic discharges are the causes of lightning.
And this is not because the category of causation cannot be applied to things in themselves. It can, provided we bear in mind the distinction between pure and schematized categories. The pure or unschematized category of causation is the logical relation of ground and consequence. And as such, it can be applied to the relation between appearances and things in themselves, so long as we are clear that this is a strictly conceptual rather than a cognitive determination. Thus, we can think things in themselves as the grounds of appearances provided that this grounding relation is understood in terms of a modified analogy with the way in which appearances cause other appearances.
And the relevant modification is that whereas a schematized category of causality always involves a consequence relation between temporal events. The grounding relation between things in themselves and appearances involves a consequence relation that operates at the level of transcendental reflection. So, the postulate of the thing in itself is a transcendental and hence purely conceptual analog of the causal relation between material objects and sensations. This is a quote from Kant now. Even if we cannot cognize these same objects, appearances as things in themselves, we must at least be able to think them as things in themselves. For otherwise they would follow the absurd proposition that there is an appearance without anything that appears.
This is the point of Kant's insistence that the distinction between appearances and things in themselves or between phenomena and noumena is precisely not a metaphysical distinction. It's not a two-world metaphysics. He's not dividing the world into sensible objects and intelligible objects. Things in themselves are the same things as appearances, but considered independently of the way in which they affect us. Now, Sellers' break with Kant is over the constraint that sensibility exercises over understanding, and hence the relation that thinking bears to knowing. For Kant, conceptual cognition is constrained by sensible intuition.
Concepts without intuitions are empty. They have no cognitive content. We can think but cannot know things in themselves. But Sellers parts company with Kant is that he thinks that Kant conflates the manifold of intuition with the manifold of sense. By distinguishing them, we can conceive of the grounding relation relation between appearances and things in themselves as analogous to the relation between conceptual representants and non-conceptual representants. Now, so the distinction is going to be, so Sellers is going to say that, Sellers wants to distinguish between intuition and sensation. And he wants to say that there is that space and time are conceptually intuited, okay,
but so that the understanding, the synthesis of apprehension is already at work in sensibility and it forms the kind of the manifold of sense into this such is, into particular objects of which can then be conceptually characterized in other ways. That's conceptual intuition whereas there's a non-conceptual representing which operates at the level of sensation. So the distinction is between conceptual representing, which is intuitive, which can be intuitive, and non-conceptual representing, which is not intuitive. So the manifold of sense comprises complexes of representations
that are also representations of complexes. But the representing of complexity does not simply mirror the complexity of the representing. For example, a representing of a green square adjoining a red square need not be understood as comprising a representing of a green square, a representing of a red square, and a representing of the relation of a joining. Just as from the expression a cat and a mat, we can form the complex common noun cat on a mat, which occurs in the sentence, this is a cat on a mat. So this is now a particular object, a cat on a mat. So from any expressions, a kind 1 and a kind 2, we can form the complex common noun, a kind 1 related to a kind 2,
or conjoined with a kind 2. The expression a kind 1 conjoined with a kind 2 is an instance of the same general form as that exemplified by the expression cat on a mat. But since this also has the same general form as the expression cat, then so does a k1 or a k2. So this is not a propositional expression of the form xry. In other words, this is not a relation between objects. A representing of a green square need not comprise anything green or square. Its structure does not mirror the structure of what it represents. And this feature can be noted by characterizing it as an of a green square representing.
Thus the representing of a green square adjoining a red square can be expressed as an of a green square adjoining a red square representing or as a green square adjoining a red square representing. In other words this is Selva's adverbial account of sensing where sensation doesn't involve relationship between sensing and sensed, where the sense is some kind of object, but rather it's a non-relational representing which can be adverbially characterized through these kinds of locutions. And the philosophical point being made here is that the representing of complexity is correlated with a complexity of representing via what Sellers calls counterpart
relations between representing and represented. But there is no resemblance between the complexity of the representing and the representing of complexity. So this is Sellers' hypothesis. All the possible ways in which conceptual representings of empirical properties and relations can resemble and differ correspond to ways in which what he calls their immediate non-conceptual occasions, i.e. non-conceptual representings of sense, can resemble and differ. So we explain how non-conceptual representings constrain conceptual representings by postulating a correspondence relation between two distinct systems of resemblances and differences, one conceptual, the other non-conceptual. And Sellers calls the latter sense impressions.
Now, this correspondence does not require a resemblance relation between items in the two systems. The correspondence is what Sellers calls a second-order isomorphism between non-resembling systems of resemblances and differences. Nor, of course, does it consist of a relation of resemblance between representation and represented. The conceptual system does not represent the non-conceptual system. The correspondent consists in the structural equivalence between properties of relations among two distinct sets of relations. So, this then provides the framework for analogical theorizing. Just as the characteristics of non-conceptual representings cannot be construed as resembling what is conceptually represented through them,
so the characteristics of sense impressions cannot be identified with those of their typical causes. These characteristics must be understood as new theoretical predicates introduced by analogy with those of certain observables. Now, although modeled upon the properties of represented physical observables, the properties of sense impressions are not physical properties or states. So once we reject the claim that the representation of something is a representation of it as something that it is, and see that all representing of something as something is conceptual, the way is open to viewing the distinction between the observable and the unobservable, which is, let's say, the theoretical, as methodological and not ontological.
And postulating theoretical entities to explain conceptually characterized appearances establishes the sought-for analogy at the level of transcendental reflection between observables and postulates on one hand and appearances and things in themselves on the other. Now this, Sellers' arguments, which is the elaboration of Sellers' argument from Double Affection, allows us to accept Kant's rejection of dogmatic rationalism and skeptical empiricism while avoiding his idealism. Here's another quote from Sellers. Indeed, it is only if Kant distinguishes the radically non-conceptual character of sense
from the conceptual character of the synthesis of apprehension and intuition, and accordingly, the receptivity of sense from the guidedness of intuition, that he can avoid the dialectic, which leads from Hegel's phenomenology to 19th century idealism. Okay. Okay. So, Okay. I'll move now to Selvius scientific realism.
Okay, so this is a condition for reconciling transcendental logic understood as an account of how non-conceptual reality constrains the structure of its own conceptual representation with naturalism understood as the claim that representings are a part of nature, minimally defined as a spatial temporal continuum. So Sellers writes,
This realism is an explanatory hypothesis about the obtainability of conceptual representings. It requires a metaphysical postulate, where metaphysical means identifying the proper functioning of categories, understood as conceptual summa genera, and when necessary, recategorizing phenomena to explain the obtainability of conceptual representings. And the postulate of sense impressions is metaphysical in this sense. They're not micro-physical entities, but re-categorizations of the sensible qualities of macro-physical objects. Now, I'll skip over this distinction between two senses of physical to get to Celsius Naturalism.
So Celsius Naturalism is not Aristotelian, where nature is a repository of substantial forms. It's not simply methodological, in a sense where methodological naturalism is... where the claim that the bounds of the natural are delimited by the explanatory resources of current science, but nor is it reductive, where every natural phenomenon must be decomposable into its microphysical components. Now, here we get to Brandom's critique of Selger's naturalism. Brandom's claim is that despite his Kantianism, Selger's scientific realism turns out to be a brand of reductive naturalism, contradicted by the anti-reductive consequences of Sellers' modal expressivism. What is this? This is the thesis that, I won't read out the quote here,
but it's simply the claim that alethic modal concepts like necessity and possibility express lawful relations between empirical descriptive concepts, such as house, cat, opera, electron. And what they express are the structural framework features conditioning the deployment of empirical concepts. So for random, following sellers, the deployment of any empirical descriptive concept presupposes knowing how to deploy these aletheic modal concepts. Modal vocabulary allows us to say what we are doing when we use descriptive vocabulary. It explains why describing something as A entails describing it as B. Modal assertions provide inference tickets from one empirical characterization to another.
So to say dogs are necessarily mammals is to say that one is justified in moving from the claim, this is a dog to this is a mammal. And this is why there's a kind of an intimate connection between describing and explaining. Describing and explaining are distinguishable but inseparable, says sellers. So that describing and explaining things always goes hand in hand. To explain means to understand the modal properties or the modal characteristics of any empirical entity. It's part of the condition of being able to properly classify or properly describe an entity that we understand its modal features, what it can and cannot do.
So, here's what Branham calls the modal Kant-Seller's thesis. Modal vocabulary renders explicit features that are implicit in the use of ordinary empirical descriptive vocabulary. Every empirical descriptive concept has modal consequences. The necessary conditions for its correct application are rendered explicit using subjunctive conditions. For example, the conditions for the correct application of the empirical descriptive concept organism are rendered explicit in subjunctive conditionals such as if this organism were to ingest a toxic substance, it would be harmed. But these inferences are non-monotonic. They are not robust under arbitrary addition of auxiliary premises. In other words, if the organism has been given an antidote to the poison,
it will not be harmed. Okay, so now we come to... Have I still got five minutes left? Yes. Okay. So what I'll do is I'll simply explain Brandom's critique of Sellers' naturalism, or why he thinks, and then maybe kind of elaborate what is misguided about this critique in the discussion. So Brandom's argument is as follows. Since descriptive properties, empirical descriptive properties, are modally involved, they locate objects in the space of implication. The identification of an object A with another object B requires identifying all of A's empirical descriptive properties with those of B. But this includes modal properties, because all of A's modal properties, i.e. the subjunctive conditionals that are true of it, must be identical with all of B's modal properties,
all the subjunctive conditionals that are true of B. Now, among empirical descriptive properties are sortle predicates, lion, house, person, molecule, sortle being any kind of noun which can be counted. You can count lions, houses, persons, molecules, etc. Sortles determine criteria of identity and individuation. Different sortles can have the same criteria of identity and individuation, e.g. the sortles kitten and cat differ in criteria of application but not criteria of identity. All kittens are cats, but all cats have been kittens, so the kitten then becomes a phase sortle of cats. However, there are different sortles, such as passenger and person, which have distinct criteria of application and identity.
The number of passengers counted by airlines is not the same as the number of persons taking flights. The same person making a round-trip flight counts as two different passengers for that airline. If one tallies up passengers in terms of flights, there will be more passengers than persons, even if every passenger in every flight is a person. Identifying person A with passenger one is a strongly cross-sortle identity because it requires identifying all of A's modal properties with all of B's. But if person A had not taken a flight, he would not be passenger one, but still be person A. So, Brandham concludes, person A and passenger one differ in at least one modal property. So, since manifest object and scientific objects fall under different descriptive sortles with different criteria of identity and individuation,
identifying them would require what Brandom calls strongly cross-sortle identities. But such sortles differ in the non-monotonic subjective conditionals that govern their application, and therefore the identification must fail. Strongly cross-sortle identity claims, says Brandom, are never true. Now, according to Brandom, Sellers' claim is that our everyday descriptions of things, cats, persons, passengers, is explained by postulating non-conceptual complexes of representations. According to Brandom, this explanation requires identifying the properties of conceptually represented contents with those of non-conceptual representing. Since such identification is impossible, the requirement of explanation is in this instance
misguided. It is a symptom of what Brandom calls object naturalism. Object naturalism seeks to locate the truth makers of claims in a target discourse in the world as specified by a favor naturalistic vocabulary, whether physics or some other science. So object naturalism is a naturalism about objects and properties that our vocabulary allows us to think and talk about. By way of contrast, subject naturalism is a pragmatic naturalism that simply wants to capture to give an account of the discursive practices of using the target vocabulary as meaningful in the way in which it is meaningful. And this according to Brandon
requires seeking what he calls a pragmatic meta vocabulary for every target for every kind of every vocabulary containing empirical descriptive claims okay so no okay I'll just simply summarize the third and final section why I think Brandon two things Brandom thinks that the ability to deploy empirical descriptive terms and therefore to master the relevant subjunctive conditionals involves what he calls systems of pragmatic inference. Pragmatic inference is knowing how to use, knowing what to do to deploy an empirical concept appropriately.
Pragmatic inference is precisely what is captured in what he calls a pragmatic meta vocabulary. The pragmatic meta vocabulary specifies what one is doing, says what one is doing, when employing an empirical descriptive vocabulary. But according to Brandom, this is precisely, it's this, the characterization of what one is doing, doing is precisely what cannot be explained in terms of any kind of discourse that is entirely, you know, that would objectify what is going on in this doing. So in other words, what Brandom is saying is that pragmatic
meta vocabulary explicates a descriptive vocabulary as meaningful in the way in which it is meaningful and thereby it immunizes and this insistence that a vocabulary can only be explained in its own terms where you know so any kind of saying can only be explained you know by rendering explicit what one is doing when what it when is saying something this means that pragmatic inference the practice of pragmatic inference or the competence involved in pragmatic inference is immunized against any kind of objectivation or any kind of scientific explanation or any kind of theoretical explanation that simply doesn't explicate the relationship between the competence at the level of description and the competence at the level of categorization.
Okay, so my final claim is this, is that... Sellers' naturalism is actually not about trying to characterize discursive competences in terms of the object categories proper to the scientific image. It's a claim that in order to understand that our discursive competences are tied to the world through our sensible capacities. So that all conceptual representing is guided by non-conceptual representing. And the claim is that this relationship, the constraint that the non-conceptual exercises of the conceptual is precisely what philosophy must forge resources that would allow it to be empirically explainable.
explainable. So ultimately the claim is that Sellers' insistence, he misconstrues Sellers as a kind of as a reductive naturalism and I think he misunderstands the most important motivation for Sellers' introduction of a metaphysics of pure processes which is to forge a category, a conceptual category, that can explain how non-conceptual representings in themselves, together with conceptual representings, are both, are, you know, ingredients in a natural process. And what he calls, you know, since processes are concrete universals, they're a new species of individuals.
They are not objects, nor are they kind of, can they be explained in terms of systems of relations amongst objects. So, by trying to immunize discursive practice from objectification, what Brandom does is he resorts to a kind of a familiar kind of hermeneutic transcendentalism, where discursive practices, the way in which we understand ourselves and the world, is in a way incorrigible in principle because it cannot be subjected to any external explanatory constraint. Okay, that's it.