Andrew Chrucky, Critique of Wilfrid Sellars' Materialism, 1990



      Wilfrid Sellars' main philosophical concern is to provide epistemological backing for a sophisticated version of a Materialistic ontology which will solve the mind-body problem. That Sellars is a Materialist may not be clear because he uses various phrases to label his own and his opponents' views. Following in his father's philosophical footsteps, Sellars calls himself by several labels: Critical Realist, Physical Realist, Evolutionary Naturalist, Scientific Realist, and Materialist.{1} These labels characterize epistemological and ontological views. In this chapter I will describe various epistemological and ontological positions and locate Sellars' position among them.



      From an epistemological point of view, Sellars wants to distinguish his position from Direct Realism,{2} on the one hand, and Idealism and Phenomenalism, on the other. As a rough approximation, the former holds that in perceptual knowledge we are not only directly confronted by physical objects but we also have a direct knowledge of physical objects; the latter, by contrast, holds that in perceptual knowledge we are directly confronted with sense data and we have knowledge about sense data. Now if we combine the idea that in perceptual knowledge we have a direct perception of sense data, and an indirect perception of physical objects, we get Indirect Realism. Since the traditional form of Indirect Realism is that of the Representative Realism of John Locke--a position which reputedly implies a copy theory of representation{3} -- Sellars wants to distance himself from this view. He calls himself a Critical Realist. This is the view that in perceptual knowledge we are directly confronted (non-cognitively) by sense data, but we have direct knowledge of physical objects.

      Now, is 'confrontation' as used above an epistemic concept? Those who believe that it is have been, for the most part, Foundationalists. In fact most philosophers have been Foundationalists in epistemology. Foundationalism is the view that knowledge is justified by basic empirical beliefs which are intuited or given, and which are independently warranted because of privileged access or otherwise. Sellars rejects a version of this view, calling it the Myth of the Given. But the belief in the given has not been limited to empiricists. The belief in the given has also been used by rationalists to argue for a 'mental eye' which is receptive of self-evident or innate truth. So Sellars' attack on the given has a broad target: it includes not only the views of empiricists, but also the views of many rationalists as well. Indeed, the attack is against most Western philosophy.




      Theories of the given are connected with theories of concept acquisition, and theories about the status of abstract entities and universals generally. Some Materialists, like James Cornman, for example, have tried to view the issue of abstract entities as neutral in regard to Materialism. He writes: "Materialists are not committed to denying the existence of entities such as classes, propositions, universals, and numbers . . ."{4} Not so for Sellars. I believe that Sellars would respond to Cornman by saying that to consider the issue of abstract entities to be neutral is tantamount to not knowing how to handle the topic, and, at best, it is a promissory note to discuss it at some other time. Sellars' own approach is nominalistic. As I understand his nominalism, it is the program of translating all sentences containing reference to abstract entities into sentences which do not.

      Sellars notes that historically the Realist position (about universals) was conjoined with a view of an immaterial 'mental eye' which received (or was 'impressed by') abstract entities. Such an argument is presented in Aristotle and Aquinas. It is also accepted, for example, by Mortimer Adler, a contemporary Aristotelian and Thomist, who agrees with the following: "They [Aristotle and Aquinas] posit the immateriality of the conceptual power in order to explain conceptual acts . . ."{5} For Adler this conceptual power in a person entails and is entailed by personal 'contra-causal' freedom of choice, and these together distinguish persons in kind from animals and machines (i.e., computers).{6} The consequences of taking this tack are monumental, as Adler correctly understands:


man's intellect (i.e., his power of conceptual thought) is the immaterial component in his constitution that makes him a person, requires his special creation, gives him the hope of immortality, and endows him with the freedom of choice.{7}

      Realism (about abstract entities), thus,--though not necessarily--may lead, as it does in Adler's case, to the positing of an immaterial principle in the person; and hence is a step toward dualism or idealism. Sellars' nominalism (which I will call Semantical Materialism) is intended as a part of an alternative theory of conceptual thought. And if his alternative theory is successful, then it undermines any argument that would posit the existence of an immaterial principle to explain the human capacity for conceptual thought.

      In chapter 5 I will examine in detail Sellars' views on concepts. And though he has much to say on the problem of freedom, I will not examine his views about this, except to repeat that he defends a version of the Compatibilist position.{8}



      The term Realism is also used to label the ontological view which affirms the independent existence of physical and scientific objects; hence Sellars uses the labels Physical Realism and Scientific Realism to contrast his view with various forms of Idealism and Phenomenalism, which deny an independent existence to physical and scientific objects. Idealism holds that only mind or minds exist independently; while Phenomenalism (Neutral Monism) holds that only phenomena (sensa) exist independently.

      As so far characterized, Realism is compatible with Dualism, the view that minds and bodies exist independently of each other; and it is compatible with Epiphenomenalism, the view that the mental is dependent on the physical, but not vice versa.



      This clarification of terms brings me to the terms Naturalism and Materialism. Elaborating on the idea of Physical Realism, Sellars points out that it "claims to continue and revitalize the great naturalistic and materialistic perspectives of the past."{9} The two terms Naturalism and Materialism are not equivalent. Materialism, at best, is a type of Naturalism. The reason for this is that Naturalism is characterized by a precept not to believe in the existence of things not natural (and the natural is usually identified with the objects of scientific study). Naturalism, as it is usually conceived, is a faith or guiding principle holding that those things which exist in the universe are accessible only through nature, which means through the sciences. Naturalism does not have a specific ontology, and is compatible with various philosophies of science. For example, both Positivists and Materialists are Naturalists, and there have also been Panpsychists who were Naturalists. It is clear, however, that for Sellars Naturalism should exclude the existence of anything supernatural, e.g., the existence of God, souls, or any other "spiritual" beings, understood to be invisible disembodied persons or quasi-persons (possessing intentionality).{10}



      Sellars' ontological commitment is expressed by the term Materialism. But because there is controversy about how to define a 'physical' (or 'material') property, and the characterization of Sellars' brand of Materialism relies on this, the definition of 'physical' will be discussed in due course.

      On the one hand, Sellars clearly thinks of himself as a Materialist in the old, classical tradition in holding that what primarily exists is matter.{11} But he vacillates between calling this position Materialism or Naturalism. This vacillation is due to reasons similar to his father's:


Now my father . . . has been both willing and not willing to classify himself as a materialist. Rightly so, for any such blanket term covers a spectrum of views ranging from the sophisticated to the absurd.{12}

      R.W. Sellars was reluctant to use the term Materialism because it suggested a Mechanistic Materialism tied to the view that matter is passive, inert--responding to, but not initializing pushes and pulls. The model for classical Materialism was the science of mechanics. But if a fundamental science, such as physics, is to provide the model of matter, then the characteristics of matter should include those recognized by contemporary physics. Thus matter is described as having a location in space and time, as possessing mass, and as having four fundamental forces: gravitational forces, electromagnetic forces, and weak and strong nuclear forces. R.W. Sellars' view of matter was based in intent on a dynamic view of matter--a matter, not as inert, but as containing various known and discoverable forces. Although, at first, he preferred the label Naturalist, he was not satisfied with its ontological neutrality, and switched back to using the label Materialist.

      Indeed the term 'materialism' is ambiguous and vague. I would define Materialism disjunctively as the view that only matter exists (Extreme Materialism), or as the view that only matter exists primarily (Moderate Materialism). Epiphenomenalism{13} is then a type of Moderate Materialism. Wilfrid Sellars is an Extreme Materialist.

      However, under Extreme Materialism we can subsume the following species: Reductive Materialism, Eliminative Materialism, and Holistic (Emergentist) Materialism. These can be distinguished in the context of how they respond to the sensation-brain identity thesis (formulated by U. T. Place).{14} Reductive Materialism accepts the identity thesis; Eliminative Materialism (e.g., R. Rorty, P. Feyerabend, and P. Churchland) dispenses with sensations as theoretical fictions; while Holistic Materialism either introduces a new distinction in physical entities and identifies sensation with the new postulated physical entities (Sellars' version), or introduces properties of wholes not possessed by the parts--a view accepted by, among others, James Cornman, but rejected by Sellars.

      Most contemporary Materialists--at least the prominent ones like D.M. Armstrong, V.W.O. Quine, R. Rorty, G. Ryle, and J.J. Smart--have tried to exclude the existence of "mental" entities from Materialism either by identifying them with neurological states or by eliminating them, i.e., by defending versions of Reductive or Eliminative Materialism. Not so Sellars: he has been a staunch defender of the existence of ineliminable and irreducible sensa; and for this reason philosophers like J. Cornman and W. Robinson refuse to characterize Sellars as a Materialist, and would probably classify him as a Dualist. In my terminology and Sellars', however, dualism is not necessarily incompatible with Materialism, depending, of course, on the nature of the dualism. Sellars' position is not a dualism which contrasts mental substances with physical substances; rather it is a dualism which distinguishes between physical states which are studied by (contemporary) physics and another set of physical states--i.e., mental states--which (as yet) are not. It is a form of Emergent (or Holistic) Materialism (i.e., Evolutionary Naturalism).


According to evolutionary naturalism, physical systems having certain complex structures are endowed with properties not found in less complex systems, properties which, in the case of living organisms, for example, do not require the postulation of controlling psychoids or entelechies.{15}

      The term 'evolutionary' is intended to capture the belief that in nature new, emergent laws and properties occur. This view contrasts with one that holds that the laws and properties studied by contemporary physics are exhaustive of nature and that the so called "emergent" laws (e.g., chemical laws) and properties (e.g., colors) are actually either identical or reducible to physical laws and properties. And if they are not, the contrasting view holds, then these so-called emergent properties (events) are, in the words of Herbert Feigl, nomological danglers; or, worse, nomological violators.{16} So there is a prevalent use of the term 'materialism' which excludes the belief in emergent laws and properties. An explicit expression of this view is found, for example, in the writings of J.J.C. Smart:


my definition will in some respects be narrower than those of some who have called themselves 'materialists'. I wish to lay down that it is incompatible with materialism that there should be any irreducibly 'emergent' laws or properties, say in biology or psychology.{17}

      And so, because the current fashion is to use the term 'materialism' synonymously with Reductive or Eliminative Materialism, Sellars writes, "I am so surprised by some of the views of the new, new Materialists, that until the dust settles, I prefer the term 'Naturalism'."{18} He writes "new, new Materialists" because his father had at times referred to himself as a New Materialist.{19}

      But perhaps Sellars' current reluctance to use the term 'materialism' stems from his polemic with James Cornman, who wrote, "I wish to examine Sellars' reasons for thinking that materialism is mistaken."{20} To which Sellars replied, "nowhere have I expressed the opinion that materialism is mistaken."{21} Sellars went on to locate the reason Cornman refused to call him a materialist: "Cornman clearly thinks that a position is materialistic if the only objects it countenances are 'physical'," and immediately adds,


I would not object to this criterion if a clear sense were given to the term 'physical.' Without it, controversies over who is and who is not a materialist may or may not be full of sound and fury, but they certainly signify nothing.{22}



      Sellars is an Extreme Materialist, countenancing only physical objects, physical events, and physical properties. But he makes a distinction between physical(1) and physical(2) properties. This distinction was, as far as I can tell, first introduced by Herbert Feigl{23} and then used by Sellars and Meehl in their co-authored paper "Concept of Emergence." [This is wrong. Feigl borrowed the distinction from Sellars and Meehl. (A.C. 1996)] This is how they formulated the distinction:


an event or entity is physical(1) if it belongs in the space-time network.
an event or entity is physical(2) if it is definable in terms of theoretical primitives adequate to describe completely the actual states though not necessarily the potentialities of the universe before the appearance of life.{24}

      Elsewhere, in his Carus Lectures, Sellars gives a more qualified formulation for the term "physical(2)":


Roughly, those features of objects are physical(2), which are, in principle, definable in terms of attributes exemplified in the world before the appearance of sentient organisms, i.e., attributes necessary and sufficient to describe and explain the behavior of 'merely material' things.{25}

      Sellars wants to distinguish a broad and a narrow sense of the term "physical": "physical(1)" and "physical(2)" respectively. His broad sense (physical(1)) was, I take it, initially meant to serve in distinguishing all those views which claimed that only those entities or events exist which are in space and time. But such a minimal definition of a physical trait would make the Epiphenomenalism of someone like Thomas Hobbes{26} a Materialistic view. So, in the Carus Lectures, where Sellars is explicitly interested in contrasting his view with Epiphenomenalism, he restricts the definition of the term "physical" to exclude epiphenomena. To do this he adds a third trait that a physical(1) entity or event must have: "Physical(1), features . . . are any which belong in the causal order."{27} And immediately adds that this would be in conformity with "a basic metaphysical intuition: to be is to make a difference."{28} He compares this to Plato's requirement: "We set up as a satisfactory sort of definition of being, the presence of the power to act or be acted upon in even the slightest degree."{29}

      But in view of the debate about causality in quantum theory, the use of the word 'cause' here may be misleading. Sellars wrote: "My remarks on quantum mechanics were designed to make it clear that I was not ruling out the possibility of irreducibly statistical laws."{30} If Sellars is granting the possibility of statistical laws at the basis of physical science, then surely he is using the term 'causality' in this context (and possibly others) in a narrow way; so that statistical laws are not causal. When he thinks about a narrow use of 'cause', he writes:


philosophers have often muddied the waters by extending the application of the terms 'cause' and 'causal' in such wise that any law of nature (at least any nonstatistical law of nature) is a 'causal' law.{31}

      However in some other contexts (as when characterizing Materialism) Sellars uses the term 'cause' in a wider sense, a sense which subsumes even statistical laws; as is clear in the following passage: "I understand the term 'causation' in terms of explanation, the variety of explanations we can give." {32} This wide sense of 'cause' seems to be equivalent to the one given to the term 'determinism' by Mario Bunge in Causality and Modern Science, which he defined to include two principles:


1. Genetic Principle: Nothing comes from nothing or goes to nothing. {33}
2. Principle of lawfulness: Nothing lawless occurs.{34}

      Sellars' considered definition, then, is that a physical entity or event must be (i) spatial, (ii) temporal, and (iii) have causal properties (in a wide sense). And Sellars requires the physical(1)/physical(2) distinction because he finds it impossible to identify or reduce sensa to currently available physical entities, events, or properties. Hence, sensa are physical(1), but not physical(2).



      Of the various labels, the one Sellars has most often identified himself with, and which others use to describe him, is Scientific Realist (Scientific Materialist would be just as appropriate). An examination of this concept will serve to highlight Sellars' position. And because Cornman has taken great effort to distinguish types of Scientific Realisms, it will be worth while to follow his distinctions.

      Cornman points out that Sellars' Scientific Realism is based on the scientia mensura doctrine, which reads, "in the dimension of describing and explaining the world, science is the measure of all things, of what is, that it is, and of what is not that it is not."{35} But all this implies is some type of Naturalism. What we need to add to Naturalism is an Ontological Realism about pure theoretical entities and events of science to get Scientific Realism.

      According to Cornman, we can distinguish three forms of Scientific Realisms, which he goes on to differentiate as: Minimal, Moderate, and Extreme. Minimal Scientific Realism holds that some pure theoretical terms of science are successful referring terms. This is to be contrasted with Scientific Instrumentalism which holds that all pure theoretical terms of science are merely instruments or calculating devices, which are in principle eliminable by such techniques as Ramsey or Craig transcriptions.{36} This is the position of Idealists, Direct Realists, and Phenomenalists. Sellars rejects all these and their instrumentalist strategies. The current debates between Bas van Fraassen's Constructive Empiricism and Sellars's Scientific Realism strikes me as a variation.{37} However, according to Ernest Nagel, the Realist-Instrumentalist debate is purely terminological.{38}

      Moderate Scientific Realism holds that physical objects are comprehensively and successfully described and explained by pure theoretical terms. Extreme Scientific Realism extends this claim to include persons.

      Having distinguished these three types of Scientific Realists, Cornman claims that Sellars falls within the rubric of at least a Moderate Scientific Realist, but not into that of an Extreme Scientific Realist because of his views on persons. To accommodate Sellars' seemingly hybrid view, Cornman introduces a new division into his classification; calling it Sellarsian Scientific Realism. For contrast, Cornman cites Quine as an example of an Extreme Scientific Realist--attributing to him the views that as concerns persons:


(1) Only human behavior needs explanation.
(2) Such explanation is achieved by positing, at most, neuro-physiological states of brains.

      By contrast, Cornman claims that Sellars believes that:


(1') Human behavior and perceptual propositional attitudes need explanation.
(2') Perceptual propositional attitudes are explained, in part, at the scientific level by postulating sensa which are not physical(2).

      In his reply to Cornman, Sellars{39} corrects him by denying an implication of (1'), namely, that human behavior and perceptual propositional attitudes are on the same explanandum level. He points out that the existence of perceptual propositional attitudes and sensa, unlike human behavior, are theoretical notions introduced to explain particular types of human linguistic behavior; thus agreeing with Quine's (1) that only human behavior (including human linguistic behavior) needs explanation. He goes on to add that his position differs from Quine's in requiring the intermediary position of postulated perceptual propositional attitudes and sense impressions (which ultimately are recategorized as sensa) to explain human linguistic behavior. And, although at one stage of his explanation Sellars identifies both perceptual propositional attitudes and sense impressions with physical states of the brain, because of the ambiguous use of 'physical', he disagrees with Quine that these physical(1) states are physical(2); and ultimately he rejects the view that sense impressions (transposed to sensa) are even "states": they are rather species of "absolute processes" or "events".

      In light of this clarification, the proper classification of Sellars' realism would be a variety of Extreme Scientific Realism with a commentary added about the physical(1)/physical(2) distinction.



      The physical(1)/physical(2) distinction introduces a difficulty concerning the status of Sellars' alleged Materialism. Cornman (like Smart) would define Materialism as the claim, roughly, that only physical(2) things exist.{40} And because of Sellars' espousal of irreducible physical(1) properties, Cornman concludes that Sellars is not a Materialist!

      Sellars replies that the term Materialism is, as such broad philosophical terms tend to be, vague. There is, he adds, enough historical warrant to call a position which includes the claim that all entities are either physical(1) or physical(2)--as Sellars claims there are--a Materialistic one.

      In view of this exchange between Cornman and Sellars, it is clear that Sellars thinks of himself as a Materialist(1), but not as a Materialist(2)--to correspond to the physical(1)/physical(2) distinction.

      Following the lead of Sellars, Emergent Materialism can be characterized as a Materialism which recognizes, in addition to the physical properties (of inanimate "matter") studied by physics, a hierarchy of "emergent" properties {41} to correspond to the sentient aspects of living organisms. Sellars' variety of Emergent Materialism, when applied to the study of human beings, can then be seen as requiring the emergence of physical entities, sensa, to correspond to human sense impressions ("raw feels"). In characterizing Materialism as early as 1952, Sellars was already conscious of the irreducibility of sense qualities:


Materialism must be interpreted as the claim that mentalistic concepts can be defined in terms of physicalistic concepts together with concepts of sense qualities, neither of these latter two types being definable in terms of the other.{42}

      The debate is between those who espouse a Reductive Materialism or an Eliminative Materialism and those who espouse, as Sellars calls it, Emergent, Holistic, Nonreductive Materialism, of which Sellars' Materialism(1) is a variety. The issue is whether some mental entities are reducible or eliminable, or neither.

      Let me briefly consider three seemingly plausible arguments against Emergent Materialism. The first goes something like this: According to Materialism, only physical events exist, and all and only physical events are determined by laws of physics. A non-physical event would be either a nomological dangler or a nomological violator. So, unless sense impressions are identical with physical events, they are nomological danglers or nomological violators. And so, either not all events are physical or some physical events are not determined. And both alternatives are incompatible with Materialism. This seems to me to be the strongest objection to Sellars' position. His best response is to deny that something which is physical(1) is necessarily a nomological dangler or violator; and anticipate (hope?) that a new law or theory will make room for sensa.

      A second related line of reasoning is that of Donald Davidson {43} who sees the mind-body relation as 'anomalous'. I am not sure I have understood his meaning. But let me explain how I interpret him. He classifies ontological views according to whether they accept of reject psychophysical identity and psychophysical laws. A psychophysical law is to be identified with a "correspondence rule" or "bridge law," or, as Davidson puts it: "a biconditional, '(x)(x is true-in-L if and only if x is P)' where 'P' is replaced by 'physical predicate' (a predicate of L)."{44} To deny this law, would mean holding that there is some x, a mental entity or event, which is true-in-L but is not P. To appreciate Davidson's problem, let us reformulate the biconditional as the conjunction of the following two conditionals:


(A) '(x)(if x is P then x is true-in-L)'
(B) '(x)(if x is true-in-L then x is P)'

      (A) is true for Davidson. This is just a way of formulating the "supervenience" of the mental on the physical, which Davidson accepts. The culprit here is (B). Since Davidson accepts the identity theory, causal interaction, and supervenience of the mental on the physical, but denies the predictability of some mental events, the best way I have of making sense of his claim is to ascribe to him a belief in the Token-Token Identity Thesis and a rejection of the Type-Type Identity Thesis. This would amount to the belief that mental types may have a number of correlative physical tokens. And there will either be a definite or an indefinite number of tokens for each type. If it is a definite number, but more than one, then either the biconditional has to be made more complex, or conditional statements used. Suppose there are two physical states that correspond to a psychological state, then the biconditional must have the disjunctive form: '(x)(x is true-in-L if and only if either x is F or x is N)'; or the relation must be expressed with a series of conditionals as, for example:


(x)(if x is P then x is true-in-L) & (x)(if x is F then x is true-in-L) &
~(x)(if x is true-in-L then x is P) & ~(x)(if x is true-in-L then x is F)

      If the number of tokens for each psychological type is indefinite, then the relation cannot be expressed in a definite biconditional. And perhaps it is this indefinite case that Davidson has in mind in rejecting the psychophysical law.

      Let me just cite one problem for type-type identity theories. I have in mind Edwin Land's experiments in color vision.{45} It was a received opinion that the human eye could see colors only as a result of stimulation by a combination of the wavelengths of three colors (red, blue, and green), but Land has demonstrated that this is not the case. There are more ways to produce full spectrum color vision including using only a combination of white and monochrome light. This is a peculiarity of color vision because a color photograph will not register such a color spectrum. There is thus a many-one relation between wavelengths and the seeing of colors.

      Given the above understanding of a psychophysical law, the following four positions are possible for Davidson:

Psychophysical IdentityPsychophysical Laws
Nomological MonismAcceptAccept
Nomological DualismRejectAccept
Anomalous DualismRejectReject
Anomalous MonismAcceptReject

      Davidson claims that Materialism is a type of Nomological Monism; while Parallelism, Interactionism, and Epiphenomenalism are types of Nomological Dualisms; Cartesianism is a type of Anomalous Dualism. Davidson identifies himself with Anomalous Monism.{46}

      On the basis of the type/token distinction, token-token identity is accepted by all mind-body identity theorists--this is what makes them Monists. Apparently, for Davidson, the type-type identity would make the monism nomological; otherwise anomalous. And it seems that Davidson is identifying Materialism with a type-type identity thesis. However, I find this use of 'materialism' odd. At best Davidson should have distinguished between a Nomological Materialism and an Anomalous Materialism. Calling a position a Monism does not distinguish it from Materialism, Idealism, or Neutral Monism; and clearly Davidson is distinguishing Materialistic Monisms from Dualisms. By Davidson's classification, I would call Sellars an Anomalous Dualist, since he rejects the Psychophysical Identity thesis. This is to say that Sellars would reject the token-token identity thesis. His position is that mental phenomena are species of physical(1) phenomena, whereas the Identity Thesis wants them to be physical(2) phenomena.

      The third problem for Materialism is this. Davidson, but not Sellars, is assuming that the Identity Thesis (in some form) is acceptable. But the Identity Thesis has been criticized for violating Leibniz's Law: (x)(y)[(x = y) iff (P)(Px iff Py)]. Suppose that x is a mental event and y is a physical event. If they are identical, then all properties of one are also properties of the other. Specifically, the identity thesis must satisfy three criteria: (1) identity of time, (2) identity of location, and (3) invariant correlative occurrence. Criterion (3) could be looked upon as a limiting case of mutual causality. It is pointed out by the critics of the identity thesis that it makes sense to locate physical properties in space, but they deny that it makes sense to locate mental properties in space on the ground that to do so is a category mistake. Therefore, the Identity Thesis is incoherent.

      Although this criticism is ineffective against Sellars who rejects the Identity Thesis, a problem of location is introduced for Sellarsian Materialism(1) in the article "Sellarsian Materialism" by William Robinson. His charge is that the concept of a sensum, which is described by Sellars as being physical(1), has no criteria for being located in space (which is a necessary condition for being physical in any sense) and is, therefore, not even physical(1). If Robinson is right, then Sellars, despite his own wishes, cannot be a Materialist(1).

      Robinson's charge makes sense only relative to his interpretation of Sellars' notion of "postulation" of sensa. On Robinson's reading, Sellars is claiming that secondary properties such as colors have 'successor predicates' in the Scientific Image. Shifting from the Manifest Image to the Scientific Image, the concept of the subject of these properties is changed first from a concept of a physical object to the concept of a person; then the concept of a person is replaced by a concept of a group of molecules; and finally this concept is replaced by a concept of a group of absolute processes. To mark these transitions of successor concepts of properties, Robinson uses a priming technique; such that we start with, e.g., red (object) and move to red'(sense impression), then to a red'' (group of molecules), and finally to red''' (sensa). Robinson apparently concedes that sense impressions can be located, but if the successor sensa are to be introduced as causes of these, then such a causal explanation does not of itself provide a criterion of location for sensa and is, in fact, compatible with sensa as being immaterial causes.{47}

      The details of Robinson's argument do not have to be reviewed, since his argument rests on a misunderstanding of Sellars. Robinson cites the relevant passage in Sellars' Carus Lectures for a correct understanding of Sellars, but, unfortunately, misunderstands it.{48}

      The way to understand Sellars' position in introducing sensa is clearly understood by Jay Rosenberg in his "The Place of Color in the Scheme of Things: A Roadmap to Sellars' Carus Lectures."{49} What Robinson took to be a postulational theory about sensa, Rosenberg sees correctly to be an "interpretive" theory:


For Sellars now proposes to distinguish sharply and explicitly between two kinds of explanatory theories. (III-36/47)[in the Carus Lectures] Let us call them "postulational" and "interpretive." The leading feature of a postulational explanatory theory is that it introduces new domains of entities. The leading feature of an interpretive explanatory theory, in contrast, is that it does not. An interpretive theory instead supplies a family of new forms of concepts by means of which what we may refer to in a category-neutral (transcendental) vocabulary as actual items or entities are re-represented in terms of new categorial structure.{50}

      As will be made clear in the chapter on perception, Sellars wants to single out an invariant property (e.g., red) in perception from the variant subject. The invariant property is given over to interpretive theory; while the variant subject is given over to postulational theory.

      Robinson's mistake was to conflate these distinctions under postulational theory. But once the distinctions are made, it is simply a mistake to view sensa as the causes of sense impressions, as does Robinson. Sensa are sense impressions which have been recategorized as absolute processes. So, if Robinson finds no problem in giving sense impressions a location in space, then he should find no problem in giving sensa a location either!

      Up to now, I have characterized Sellars' epistemology and ontology in general terms. To understand his position--as a philosophical position should be understood--it must be embedded in the context of concrete problems. The epistemological problem is to determine whether empirical knowledge has foundations in basic empirical beliefs. The ontological problem is the mind-body problem, which Sellars subdivides into the intentionality-body problem and the sensorium-body problem. These problems are viewed by Sellars from the perspective of a clash between two conceptual frameworks: the Manifest and the Scientific Images. I will, therefore, turn to a description and criticism of these two conceptual frameworks.

[Go to Chapter 2]