INTRODUCTION

      This manuscript is a critical commentary on the philosophy of Wilfrid Sellars. By a critical commentary I have in mind the sort of task which Sellars himself employs when dealing with the history of philosophy. Sellars in his examination of philosophies and philosophers sets out to discover the truth, that is, to do philosophy. The focusing on the views of a particular philosopher is meant to provide a counterpoint to one's own views. The examined philosopher provides an occasion for developing one's own philosophy, and this is especially rewarding if the examined philosophy has verisimilitude, as does that of Wilfrid Sellars. The conclusions I reach are very close to Sellars' own -- so close, in fact, that I am not certain whether what I am offering as correction are of things I am only misinterpreting.

      What is philosophy? Sellars put it succinctly as the attempt to understand how things, in the broadest sense, hang together, in the broadest sense. A little fleshing out of this leads us to the characterization of philosophy as the attempt to systematize a categorial scheme. According to Sellars we have two categorial schemes which need to be reconciled: one is the inherited product of common sense, which Sellars calls the Manifest Image; the other is the scheme of microscience, which in its ideal limit is called the Scientific Image.

      Sellars is speculating that theoretical science -- which for him is the measure of what exists -- is heading towards an ontology of absolute processes; and he is further speculating that this ontology must include new types of emergent entities, called 'sensa', which are to be the counterparts of human sensations. All things in the universe are to be understood as constituted by an interplay of these and other absolute processes. Such a view is a variety of what is called Emergent Materialism: a position defended by, among others, Sellars' own father, Roy Wood Sellars, and C.D. Broad.{1} Dialectical Materialism seems to be another variety.

      But because of ambiguity in the term 'materialism', and because current usage is to view materialism either as a psychophysical identity thesis, or as a reduction thesis, or as an eliminative thesis -- none of which characterize Sellars' variety of materialism -- there is a need to clarify the sense in which Sellars is a materialist. Specifically, Sellars' materialism requires a distinction between two kinds of physical (or material) properties. But precisely because of the presence of this distinction, two commentators, James Cornman and William Robinson, have argued that Sellars is not a materialist, but a dualist. I explain Sellars' distinction between the two kinds of physical properties and defend his version of materialism from such criticisms.

      The justification for Sellars' ontological claims requires grounding in epistemology, whose history can be characterized by 'the quest for certainty' (to borrow a title of John Dewey's book).{2} Historically such certainties were claimed for both a priori and a posteriori judgments. However, since Sellars is primarily concerned with the ontology of physical objects, persons, and their properties, his major concern is with the availability of a posteriori, empirical claims as foundational for empirical knowledge. Such empirical foundational certainties were attributed to elementary sense experiences -- variously called 'sense impressions', 'sensory intuitions', 'sensations', or, characteristically in the present century, 'sense data', 'sensa', and, currently, 'qualia'. Knowledge of sense data, it was claimed, is available through an analysis of perception. Sellars denies this claim. To think otherwise, for him, is to embrace a myth: the myth of the given in knowledge. Thus, the conclusion can be drawn that for Sellars empirical knowledge has no certain, privileged empirical foundations. I will agree that these foundations are not certain, but I will argue that they are indeed privileged.

      As to ontology, in order to defend Emergent Materialism, Sellars has to overcome two ontological difficulties. He has to provide a place for physical objects and persons in the scheme of the theoretical entities of science. The latter problem, which is the mind-body problem, requires an account of thoughts, sensations, and free will in a scientific framework. To make his case, Sellars provides a behavioristic account of language learning and of how we acquire introspective knowledge of our mental acts and states, which are conveniently grouped into thoughts and sensations. Introspective knowledge is rendered as a theoretical extension of a publicly observable base. As concerns thoughts, Sellars sees no problem with them for materialism: thoughts are, on his view, functional states of some unspecified medium or other. Talk of thoughts is, thus, ontologically neutral; and because of this, the medium of thoughts could be -- and is for Sellars -- the neurophysiological activity of a living brain.

      It is otherwise with sensations. Unlike many contemporary Materialists who either deny the existence of sensations as objects of introspection, or who try to reduce sensation talk to talk of behavioral dispositions, or who would eliminate sensation talk altogether, Sellars acknowledges unreduced introspective knowledge of sensations -- a knowledge which, however, is not available through an analysis of perceptions. Consequently, this requires giving them a twofold theoretical status: first, as adverbially understood sense impressions in the Manifest Image (roughly, the common sense framework); and then, as 'sensa' in the Scientific Image (roughly, an ideal scientific framework) consisting of absolute processes. As to free will -- a problem I will not discuss --- Sellars adopts, what is called, a compatibilist position. {3}

      In a rough way, this is how I see the broad features of Sellars' philosophy. Do I agree with his conclusions and arguments? I agree with the final ontological conclusion that the universe consists of physical (material) absolute processes, but I do not agree with some of the intermediate ontological and epistemological claims. One disagreement concerns the epistemological availability of phenomenal sense data. I believe that empirical knowledge has, contrary to Sellars, foundations in the sensing of sense data. But no major ontological difference with Sellars over the Realism/Idealism (Phenomenalism) controversy follows from this. For example, the availability of sense data is a necessary -- though not a sufficient -- condition for Phenomenalism, and, therefore, Phenomenalism can be rejected on other grounds. I agree with Sellars in accepting some form of Realism, and could call my version Theoretical Realism to emphasize the theoretical nature of physical objects; but I will opt for calling it Animal Realism to echo George Santayana's sentiments about 'animal faith'. But this as yet does not provide me with any radical difference with Sellars. He too gives the conceptual framework of common sense a "theoretical" status in his own way.

      However when the issue turns to the methodological and epistemological priority of the common sense versus the scientific conceptual framework, I start to balk. For Sellars the Manifest Image, which is his regimented rendition of a common sense framework, is in principle replaceable through conceptual retraining. I have two difficulties with this thesis. The first is that I don't know what Sellars has in mind by this retraining. As far as I know, he never fully explains this thesis. The second is that I am convinced that we have a pre-conceptual representation of the universe as composed of something like physical objects, and I am further convinced that this pre-conceptual representation is an unconscious and involuntary construction of animal representation. So even though we can recategorize the universe in terms of some categories, as, for example, shifting from an animistic view of the universe to a de-personalized one, we cannot recategorize it by de-physicalizing it. I can conceive of a de-physicalized universe, but I cannot imagine it. And I would immediately add that our common sense perceptual universe is limited by what Kant called the schemata of the productive imagination. But if there is an unalterable core of physicalism in common sense perception, this would account in part for our ability to understand the language, writings, and behavior of all sane human beings in all times and places -- despite Quine's thesis of the indeterminacy of radical translation {4} to the contrary. So the issue which separates me from Sellars, besides the epistemological availability of the awareness of sense data as foundational for empirical knowledge, is the belief in a pre-conceptual system of representations which restrains us from eliminating our view of physical objects.

      This dissertation is my effort to state this dialectical confrontation with Sellars. My first task will be to exposit the nature of Sellars' Materialism, and to consider some difficulties with it. After this, in the next chapter, I examine his view of common sense, which he calls the Manifest Image. My objection here is that the Manifest Image should not be strictly identified with a common sense view. My reason is that common sense recognizes the existence of absolute processes (or events); the Manifest Image does not. I require the recognition of absolute processes at the common sense level to provide a model for sense data. Sellars acknowledges something like this rather late in his writings when he tries to reconcile his particular adverbial theory of sensing with classical sense data theories. But even here he thinks he is revising the Manifest Image rather than working in it.

      The next issue I grapple with is the mind-body problems. I say 'problems' because Sellars subdivides the mind-body problem into an intentionality-body and a sensorium-body problem.{5}

      The intentionality-body problem requires a discussion of language, thought, and semantics; and their grounding in a behavioristic methodology. My dissatisfaction with Sellars' approach is that he is not consistently clear in making a distinction between non-conceptual and conceptual thoughts. This has resulted in a misunderstanding of Sellars by his critics, and perhaps has even introduced some unclarity for Sellars himself. The remedy here is to focus on the distinction between what Sellars of late has called a 'representational system', which he credits to animals; and a conceptual representation system, which only language users possess. Once this distinction is pushed to the foreground, two consequences follow -- a good one and a bad one. The good one is that it becomes possible to see a compatibility between Sellars' view on thoughts with those of many of his opponents, including C. I. Lewis, E. Hall, R. Chisholm, R. Firth, R. Clark, A. Marras, Alan Goldman, and T. Russman. The bad one is that the recognition of the two representational systems creates a tension in Sellars' philosophy, or, I should rather say, an incoherence. What I have in mind is that the existence of a pre-conceptual representational system begs to be put in the service of the construction of an 'animal world' composed of something like physical objects. But in that case Sellars' claim that the Manifest Image is replaceable, I believe, must be abandoned. And if that thesis is abandoned, then some presupposition of that thesis too must be abandoned. The most vulnerable presupposition, I suspect, is Sellars' view on the so-called 'bridge laws' or 'correspondence rules' in theoretical science, which he understands as candidate re-definitions. The alternative interpretation which has merit is to view them as causal laws.

      This brings me to the problem about the foundations of empirical knowledge and Sellars' rejection of The Given as a myth. I examine Laurence Bonjour's argument against foundationalism. I note that he fails to present the correct logical form of a plausible foundationalism. The correct form is neither a case of deduction or simple induction, but a form of what Charles S. Peirce called 'abduction'.

      Bonjour also fails to elaborate on the possibility of giving an a priori justification of empirical knowledge. Such an a priori justification seems to be implicit in both Chisholm and Sellars himself. I distinguish Pseudo Foundationalism from Genuine Foundationalism, and agree with Sellars that Pseudo Foundationalism is a myth. As I am using these phrases, Pseudo Foundationalism relies on a physicalistic language for foundations; while Genuine Foundationalism relies on phenomenal reports. Sellars argues that phenomenal reports as foundational for physicalistic language are unavailable. I disagree, and argue for the availability of phenomenal reports and hence of Genuine Foundationalism.

      To get at the root of my disagreement with Sellars, I try to disambiguate the notion of the Given in Sellars. And to pinpoint the problem, I examine and reject William Robinson's argument that The Given is not a myth. Robinson's analysis of Sellars' position seems to be flawed by some misunderstandings. Despite this, Robinson's position ultimately relies on the claim that there is such a phenomenon as a counting use of concepts which does not presuppose a classificatory use of concepts. I argue that he is mistaken about this.

      I accept the given as understood by C. I. Lewis and defended by R. Firth. It turns out that the Given is, as far as I can tell, simply the awareness of a world through a non-conceptual representational system, or what Lewis called 'sense meaning'.

      Next I take up the sensorium-body problem in the context of an account of perception. The word 'account' is here doing double service for two separate problems. One is to give an explanation of perceptual reports; the other to give an analysis. I agree with Sellars about explanations, but not with his analyses of perceptual reports. His analyses are not reductive enough for me. I side here with the position of Firth in their debate about this issue. Furthermore, I disagree with Sellars about the, so-called, sense-datum inference. Sellars believes with Chisholm and Cornman that it is a fallacy. I side here with Frank Jackson who accepts the validity of this inference.

      In addition, Sellars has an argument for the conceptual unavailability of a model for sense data. His argument seems to rely on a distinction between essential and accidental features of a model. I reject his argument on the ground that what he takes to be essential connections are in fact accidental.

      I also have a disagreement with Sellars about the intentionality of sense data. He says that they are nonintentional. But in saying this he conflates the intentionality of conceptual representations with the intentionality of non-conceptual representations. By contrast, I side with Romane Clark and Everett Hall in claiming that sense data are constituents of non-conceptual representations.

      I also find Sellars', so-called, 'adverbial' approach to be either unacceptable, or acceptable if it can be made equivalent to a version of the sense data approach, as Sellars tries to do. The reconciliation of the sense data and the (adverbial) sensing approach relies on the use of a new model of subjectless sentences, as illustrated by the form 'It is thundering'. I applaud this strategy. But immediately the question can be raised: From which conceptual framework does this model come from? It cannot be the Manifest Image, since that Image is based on a substance ontology. So the implication is that we have access to another conceptual framework, which includes an ontology of processes. I call this conceptual framework the Common Sense framework. It includes the Manifest Image as a part, but it is not identical to it.

      Having made my case for sense data and foundational knowledge, I defend Animal Realism. As I am using this phrase, it is a position consisting of two theses. The first thesis is that realism about the external world is correct. The second thesis is that our representation of the world is a product of two innate faculties: one is a faculty of receptivity and the other is faculty of schematization. These faculties are evolutionary products which are necessary for the possibility of survival of all animals which must learn to cope with their environment.

      I assume that animals are capable of making unconscious, involuntary inferences. The psychologist, R. Gregory, talks about perception as hypothesis formation. I agree with this.

      Does Sellars disagree? No matter how Sellars answers, the tension between the conceptual and the non-conceptual representational systems again must be faced. If there is a non-conceptual representational system, then it is the medium in which perceptual hypothesis are constructed. But if so, it would seem that conceptual representations cannot replace these pre-conceptual constructions; but can, at best, only supplement them.

      My emphasis on the non-conceptual and the conceptual representational systems is meant to reveal the major tension in Sellars' system. In his polemics against the Given, Sellars seems to be rejecting both conceptual and non-conceptual representations as given. But in his latest writings he is willing to grant to animals non-conceptual representations -- some of which are learned and some of which may be innate. Of these learned animal representational systems, it seems that their nature is the product of invariable animal capacities and impressions. The suggestion is that though the perceptions of animals are constructed, the construction is more or less invariable. In that sense, we could say that despite the need for construction, the product of construction is relatively invariable; hence, given. If this is a faithful reconstruction of Sellars' position, then there is a tension between two sorts of givens in Sellars' system, i.e., a tension between the claim that there is no given whatsoever and the claim that animal representations are given. Perhaps the tension can be alleviated by attributing to Sellars the claim that there is no conceptual given and the claim that there is a pre-conceptual given.

      If that is Sellars' position (which it very well may be), then there is a question about the extent of conceptual relativity if an invariant preconceptual representational system is indeed given.

      Pressing Sellars on this issue has repercussions on Sellars' global problem of reconciling the Manifest and the Scientific Images. If animal representation is a limiting factor in the construction of the Manifest Image, then there may be more irreplaceable aspects of the Image than Sellars is willing to grant. For one, Sellars believes that the perception of colors as characterizing the external world is a case of mis-perception. The implication is that a new generation of children could be retrained to conceptualize perceptions properly. But on the contrary, if mis-perception is a matter of evolutionary innate animal representation, then the alteration of such perception would not be a matter of reconceptualization or training, but a matter of genetic engineering or "rewiring" of the brain.

      The upshot of my disagreement with Sellars about the replacement of the Manifest Image is that I am committed to a more conservative view of science as built on unalterable categorial foundations, or, in Sellars' terminology, I am committed to a form of the Given.