Correspondence between Wilfrid Sellars and J. J. C. Smart

Edited in hypertext by Andrew Chrucky with the assistance of Willem deVries.
Editor's Note: This correspondence is one of the many correspondences which Sellars made available to his graduate students at the University of Pittburgh. It was preserved by Willem deVries from the time he studied with Sellars. Insertions by the editor are within slashes, e.g., /red/.

Due to the limitation of current hypertext, the following conventions have been used. All Greek characters (except phi) are rendered by their names; thus, epsilon, which serves as a sign of membership is rendered as 'epsilon.'

Logical connectors and quantifiers are all expressed by bold characters as follows:

--> = material implication
<--> = material equivalence
& =conjunction
~ = negation
v =disjunction
(x) = universal quantifier
(Ex) =particular ('existential') quantifier

Contents

c. 1961: Sellars to Smart
Feb. 27, 1964: Smart to Sellars
March 9, 1964: Sellars to Smart
March 14, 1964: Smart to Sellars
March 23, 1964: Sellars to Smart
(c. 1961)

Professor J.J.C. Smart
The University
Adelaide 5, Australia

Dear Smart:

I have carried your letter around with me in my briefcase with practical intent, but the pressure of the end of term and of program making for the APA have kept me from doing any more than re-reading it a number of times and, correspondingly, mentally rewriting my private reply. At last I am relatively free, so here goes,

I agree with you that "the manifest image is not all wrong." But, as I see it, there is a right way and a wrong way to put this. I prefer to say that whereas the objects of the manifest image don't 'really' exist [for the force of really see the "Language of Theories"]. What does 'really' exist has statistical properties which are, in a structure-theoretical sense, isomorphic with properties of manifest objects. We are not, I take it, very far apart on this point.

You write, "I think I can give an account of color which squares with both the scientific image and the manifest image." It depends on what you mean by 'an account of color' and 'squaring with.' Clearly there is an isomorphism in the above sense between

Manifest [what colored objects do to perceivers in various circumstances and the discriminations these impingements make possible]

and

Scientific [what colored objects do to perceivers in various circumstances and the discriminations these impingements make possible]

But in what sense is there one account and not two accounts which are isomorphic plus the higher order comment that they are so?

I don't see at all why we can't speak of a physico-chemical theory of biological phenomena. Electronics is a species of engineering, and does, indeed, differ from electro-magnetic theory. [We could use "electronic theory" to refer to the theories made use of by electronic engineering, but this would be a derivative way of referring to the theories]. I do not think, however, that the application of physical and chemical theory to living things need be constructed as an engineering or practical application (as 'bionics'). The term 'application' can conceal a difference between two radically different ways in which a theory is 'applied'. [See my account of the reduction of biology to physics in the second part of "Image." /"Philosophy and the Scientific Image of Man"/]

In my account of perception, impressions of red triangles [red and triangular impressions in the analogue-senses of these adjectives--and, more basically, states of being redly and triangularly (I mean to write red-triangle-ly) impressed] in the manifest image are states of the perceiver. They are not, however, conceived in the manifest image as states of the nervous system, though they are 'identified' with the latter in the sense of "identity" mobilized when we speak of 'identifying' chemistry with a branch of theoretical physics. [The fact that scientific theory seeps into common sense, so that people come to believe that sensation and thinking "go on in the brain" may conceal, but should not blind us to, the levels of conceptual structure involved. To understand what is meant by "in the brain" we must sort out and relate these levels.] Furthermore, impressions of red triangles in the manifest image are not particulars--though, of course, they are derivative logical subjects--and are not to be confused with the 'color particulars' to which I refer at the end of EPM in the course of sketching the logical space in which the colors of the manifest image might find a final resting place. Impressions are what I called theoretical-plus entities in the manifest image. The 'color particulars' in question are intended to be the analogues in an ideal scientific image of colors of physical objects in the manifest image, and to be 'identifiable' in the language of theories sense with the latter.

Here it is essential to note that the colored and shaped particulars of this ideal scientific framework are to be 'identified' not with aspects of manifest objects but with aspects of the visual impressions we have on looking at these objects. {I will be developing the point shortly that the micro-counterpart of visual impressions include more than the counterpart of the latter qua (in the analogue sense) colored and shaped.} As I see it the early mistake which the classical "sense-datum inference" involves is the suggestion that sense-data belong to the manifest image and mediate perception epistemically rather than being elements in an 'external' theoretical explanation of perception.

It is therefore important, as I see it, not to be misled by the fact that the scientific image includes objects (systems of micro-entitites) which, by virtue of certain statistical properties, are isomorphic with manifest objects and by the fact that certain statistical properties of these items are isomorphic with the color similarities and differences of manifest objects into supposing that it is the colors of manifest objects which are to be 'identified' in the L of T sense with statistical properties of the scientific counterparts of manifest objects. For while it is a conceptual truth in the manifest image that e.g. red objects in standard conditions produce red visual impressions (in the analogue sense of 'red') in standard chaps, the manifest image by no means precludes the idea that it is by virtue of some fact other than but correlated with the redness of these red objects (e.g. their surface texture promisorry-note-ishly conceived) that they produce these red impressions. It is interesting to note in this connection that in general the secondary qualities (I shall use this term in its usual sense and not refer to Locke's powers) play no role in the explanatory mechanisms of the manifest image, though they are that with respect to which these mechanisms are identified. [This combination of correlation and inefficacy, indeed, was the reason why they were called 'secondary'.] The manifest image not only demand[s] that color itself be a factor in the mechanism, but contains no hint as to how it could be. Thus, "that which accounts for the causations of impressions of red in the perceivers of red objects in standard conditions" can be 'identified' (L of T) with a statistical property of the scientific counterpart of red objects in the manifest image without 'identifying' this statistical property with the redness itself. This interpretation of the fitting of the scientific to the manifest image is supported by the fact that the statistical property to be correlated with the color of the manifest objects is selected because of its connection with the reflection and absorption of light rays as stimuli for optical processes. Compare also the case of sounds. mC /middle-C/ sharp sounds produce mC sharp auditory impressions in standard chaps in standard conditions, but the mechanism involves not the mC sharpishness of the sound, but its status as vibration moving through the air. Manifest sounds are, e.g. mC sharpish goings-on. Although their micro-theoretical counterparts have statistical properties, which can be 'identified' with micro-vibrations, nothing "out there" is to be identified with the mC sharpishness.

In short, the inefficacy of the secondary qualities reflects the fact that in the scientific image they turn up in chaps theoretically construed.

You write, "I still think that if my views about colors is correct, we can account for the existence of the concept of a pink ice cube and of the human eye, so that the percipient just couldn't be expected to react to the micro-level inhomogeneties of the ice cube. But this is an ignoratio elenchi inhomogeneties of the ice cube," but the visual cortex and the having-a-pink-cubical-sensum. The problem is, "is it legitimate to identify having-a-pink-cubical sensum--and I assume that you realize that by speaking of the mC /?/ cube as colorhomogenous I do not mean that the only color involved is pink; a dappled ice cube would be homogeneous in my sense; homogeneity means, roughly, continuous-in-terms-of-color -- with the system of microparticles in such and such a state of statistical state of aggitation? You lay great stress on the idea, sound as far as it goes, that in the context of impression talk, the adjectives 'pink' and 'cubical' (and therefore, presumably 'continuous in terms of color') occur not in the form 'red and cubical x' but in the form 'P has red and cubical x' where P is the percipient. But, as I see it, when you spell out the logic of such contexts it turns out that they mean not simply

P is in a state normally caused by pink and cubical objects and which enables such and such discriminations and is bound up in speaking animals with such and such verbalizations,

but, without neglecting the force of the above descriptive characterization (description in Russell's sense) of the state,

P is sensing (homogeneous)-pink-cube-ly

or, making the appropriate grammatical shifts involved in using the noun "impression,"

P has a (homogeneous) pink, cubical impression

where the italics serve to remind us that the expressions are theoretical derivatives from the physical object use of the terms 'red' and 'cubical' and are intended to highlight the fact that the role of the theory is to explain such facts as that human perceivers seem to see (homogeneous) pink cubes (a) when there is a transparaent cube there but it isn't pink; (b) when there is nothing there (hallucination). It is, therefore, incorrect on your part to assume that the fact that

P has a pink and cubical sensum

puts 'pink' and 'cubical' (and therefore, presumably, 'homogeneous' as a categorizing epithet) in the embrace of "P has . . ." means that an unembraced (though analogical) role of these terms is not involved. Hobbes seems to have made the same mistake (in the material mode) in his treatment of phantasms.

What exactly do you find in the cortex (constructed as a system of particles) which correspond to the homogeneity of the manifest pink ice cube as this homogeneity appears in "P has a (homogeneous) pink and cubical sensum"? After all, putting it crudely, whatever else manifest objects are, they are continuous color solids. Their continuity (in the manifest image) is pure occurrent continuity, not a matter of every re[li]gion being sooner or later occupied within a small interval of time by a color spot. Granted that there are really no such objects; granted that they are a conceptual projection of visual impressions which are caused by statistically stable clouds of micro-entities, -- is there no foundation in these visual impressions for the color continuity of the manifest objects? [The cortex-cum-eye may respond in a statistically continuous way to the statistically continuous cloud; but this throws no light on the relation of the state of the cortex to the visual impression]. Granted that the esse of the manifest objects is concipi)? Descartes could say that nothing is red and leave it at that, because for him there is no sense in which an impression of red is red. That there are impressions of red was for him an ultimate fact for which the creative activity of God was responsible--more accurately, that there is such a thing as impression of red is an ultimate fact not to be analysed in terms of anything more elementary; 'impressions of red' is, so to speak, a primitive in his system. Nowhere in his scheme of things is anything both real and red (save, of course, in the power sense). As I see it, on the other hand, although in the genetically primary (manifest) sense of 'red', nothing is both real and red, in the genetically derivative but ontologically primary sense of 'red' something is both real and red, namely sensa as particulars of the scientific image (ideal).

As far as I can see, your view is Cartesian without a theory of what it means to say of an impression that it is of red (unless you propose the description of the impression in terms of its standard causes and effects as an analysis of this meaning--but see my remarks on this type of move in /Section #61/ paragraph (2) p. 323 of Empiricism and the Philosophy of Mind). My suspicion is that instead of doing justice to the theoretical role of 'having a pink sensum' in the manifest image, you 'identified' it too quickly with 'statistically stable systems of excited particles' of one level of scientific theory, and that by doing so you threw away the guiding thread of the problem. In short, it seems to me, that either you must deny that color as we conceive it is essentially homogeneous (roughly, continuous or assert that although our concept of color is of this character, it is radically illusory in that we cannot construe color thus conceived as a projection in the manifest image of a genuine feature of reality. Indeed, I am really puzzled about your view as to the real counterpart of color. Is it spatially (or quasi-spatially) dis-continuous (a very sophisticated notion)? You need not, of course, think of this counterpart as a quality. (Perhaps this is one source of trouble. Not all one-place predicates stand for qualities. Why couldn't there be particulars which are reds, 'red', in this theoretical use, being a common noun? Why couldn't there be particulars which are portions of red as table tops are portions of marble?)

On my view nothing has color in the scientific image, not because there is no color there, but because in this image, color exists as expanses of color (in the theoretical analogue of 'color') which, as space-time worms are 'entwined' with the space-time worms which are the ultimate counterparts of the particles which make up, at one level of scientific theory, the visual cortex.

It is these 'entwined' structures and not the color worms by themselves which are the ultimate counterparts of visual impressions and with which the latter are to be 'identified'. [Compare p. 2 above, the paragraph beginning "Here it is essential to note . . ."] You will remember that I stress on a number of occasions that one must cut beneath the particular image to find the micro-structures with which visual impression can be 'identified.' As I see it, to accept the particulate image as ultimate is to be on the road either to dualism (epiphenomenalism, to begin with) or to a rejection of the scientific image or to the radically tough-minded expedient of finding no resting place for the sensable features of experience. It would, of course, be absurd to deny that the extension to the cortex of the particulate physical theory developed in connection with inorganic substances, i.e., the intepretation of the cortex as a system of micro-particles) throws a fantastic amount of light on its behavior. It must also be admitted that the hypothesis of the "identifiability" (L of T) of living cortices with the system of microparticles cannot at present be disconfirmed by appeal to the results of the experimental techniques of physics today. Consequently, of course, if all we know about brains was what we learn through the experimental techniques of physics and the theoretical interpretation of their results, there would be no basis (other than, perhaps, conceptual paradoxes which are thought by some to be involved in the particular image) for supposing that the later is a mere way station in the development of science.

A remark or two about the "abstract entities," in "Grammar and Existence." I distinguish between "quantifying over classes" as in

(EK) x epsilon K

and "quantifying over classes" as in

(EK-kind) x is a member of K-kind

(I intend the distinction to obtain in the case of extensional classes for which

(x) (x epsilon K1 <--> x epsilon K2) --> K1 = K2

as well as for kinds in the ordinary sense for which this is not true.) This distinction parallels that between 'quantifying over properties' as in

(Ef) x is f

and 'quantifying over properties' as in

(Ef-ness) x f-ness

and similarly, between 'quantifying over propositions' as in

(Ep) p or is it raining

and 'quantifying over propositions' as in

(Epthat-p) that-p is inconsistent with that-q

I argue that in each pair quantification of the first kind doesn't mention abstract entities, whereas quantification of the second kind does. On the other hand, I don't dish abstract entities, I construe them as, in the first instance, linguistic roles. Their objectivity is not the objectivity Plato and the Platonist give them, it is the intersubjectivity which roles in games have, e.g., the bishop role in chess. Thus "triangularity" is the name of the role played in our language by 'triangular' and in German by 'dreieckig.' The German name of this role is "Dreieckigkeit." To develop this point requires a theory of the status of roles. This is the problem on which I am working these days. It requires a theory of rules and of rule-governed behavior. I have made some earlier tries, but a lot remains to be done.

........................

Many thanks for the ethics monograph. I thoroughly resonate to it. I am convinced that something very much like it must be true.

The Time essay /"Time and the World Order"/ is now out. I have asked the University of Minnesota press to rush a copy of the volume which contains it (and Adolf's definitive essay) in case one hasn't already been sent to you.

I hope that the micro-particles in your pleasure center are buzzing in a way which implies that your nervous system is in goal states with respect to a reasonable number of acquired and unacquired drives.

Sincerely yours,

Wilfrid

Contents
February 27, 1964

Department of Philosophy
Yale University
New Haven, Connecticut

Dear Wilfrid,

Can I ask you a question about your paper on the identity theory that you sent me recently? (I may have asked you a similar question before.) It is to find out how much I'd be committed to if I were to agree that there are physical(1) things that aren't physical(2). (Let us say that a thing is physical(2) if it has only physical(2) predicates true of it.)

Let us suppose for the sake of illustration

(1) persons are physical(2)

(2) Nation sentences can't be translated into person sentences but nations are nothing 'over and above' persons. [of course you might not agree that (2) is self-consistent.]

Do (1) and (2) allow nations to be physical(2)? Or are they only physical(1)?

If (2) by itself prevents nations from being physical(2), then your notion of emergence seems metaphysically innocuous and I could probably accept it!

Thanks again for sending the paper. ......

Best wishes,

Jack Smart

Contents
March 9, 1964

Professor J.J.C. Smart
Department of Philosophy
Yale University
New Haven, Connecticut

Dear Jack,

This monster of a letter is designed to lead up to simple and straight forward answers to your questions. If you become bored by the preliminaries, you have my permission at any time to peek at the concluding paragraphs.

To multiply or not to multiply individuals (or particulars or 'logical subjects') seems to be at the center of your concern about emergence, and rightly so. But the answer to the question "How many logical subjects is an X?" is a function of the framework in which X is being considered{1}. I shall assume, in what follows, that the logical subjects with which we are concerend are the basic ones of the relevant framework, e.g. persons and molecules, and not such obviously derivative items as the death of Socrates and the collision of two molecules. The following numbered paragraphs contain the gist of what I would want to say on the subject of emergence.

1. In the manifest image or Strawsonian framework (i.e. ordinary discourse) a person is one logical subject which satisfies both physical(2) and physical(1)-but-not-physical(2) predicates.

2. In the intermediate scientific framework of neurophysiology, a person appears as a plurality of logical subjects which are severally physical(2) things in your sense of this phrase (according to which "a thing is physical(2) if it has only physical(2) predicates true of it") But though these logical subjects are severally physical(2) and stand in certain physical(2) relations which given them the physical(2) structure of a living brain, certain statements are true of them as a group which are not translatable into physical statements about their physical constituents. It is at this point that the classica1 epiphenomenalist introduces a queer set of physical(1)-but-not-physical(2) particulars ("epiphenomena"), and interprets the surplus predication about the physical(2) entities as a matter of their being accompanied by these particulars. But this move is not necessary, though it has a rationale which I shall explore in a moment. No new logical subjects need be introduced, for one can simply say that certain physical(1)-but-not-physical(2) predicates apply jointly to the physical(2) things; for example, one can say that a group of nerves jointly "feel a pain" or "sense-bluely". Notice that

Senses-a-red-triangle (n(1), n(2), . . . n(n))

would be the many-logical-subjects-cum-verb analogue of the relational form
'R (X(1), S(2), . . . X(n))'.{2}

I would, of course, interpret these predicates as analogical derivatives of their Strawsonian counterparts. The crucial point is that the use of these predicates involves no new things or particulars over and above the nerves. It is in this precise sense that "persons are nothing over and above physical(2) things."

3. The same two types of move can be made after the "identification" of the nerves of macro-neurophysiology with bio-physical systems of micro-entities.

4. Yet, as I see it, neither type of picture is ultimately satisfactory.

(a) The epiphenomenalistic introduction of new particulars is not satisfactory because it makes unintelligible why certain systems of physical(2) entities should be accompanied by clouds of "epiphenomena" which reflect changing physical(2) patterns without playing any role in the explanation of the behavior of these neuro-physiological systems. Surely the regulative ideal of science is a system in which "everything makes a difference."

(b) The interpretation of emergence in terms of the irreducibly physical(2) group activity of physical(2) entities is equally unsatisfactory for a reason which is more difficult to formulate, but which goes somewhat as follows: Consider the group activity predicate.

(alpha) 'Senses-a-red-triangle'

which applies, ceteris paribus, to groups of nerves or groups of micro-physical particles. This predicate, (a), is an analogical derivative from the verbally similar predicate

(beta) 'Senses-a-red-triangle'

which applies to Strawsonian persons. Now the latter predicate, (b), is itself derivative from the physical object predicate

(gamma) 'is red and triangular on the facing surface'

Notice that predicates like (alpha) and (beta), although they contain a reference to a system of spatially related items, do so only indirectly, for spatial items are brought in only in the sense the sign designs which stand for them in English are built into the sign designs which are used in English to form such predicates as (alpha) and (beta), in order to bring before the mind the physical models from which the conceptual structure of these predicates is analogically derivative, and by doing so make this conceptual structure intuitive. [Cf. Science, Perception and Reality, (46-49, 92-95, 190-193.)] As I see it, an ultimately satisfactory scientific account of what it is for Strawsonian persons and systems of scientific objects to satisfy such physical(1)-but-not-physical(2) predicates as (alpha) and (beta) must be in terms which make a direct reference to spatial entities, i.e., entities which satisfy the postulates of a pure geometry. One of the primary attractions of the additional particulars or "sense-data" approach to emergence is that it satisfies this requirement. Philosophers who make this move conceive of a person's visual sense-data of the moment as a system of spatially related particulars. (It is usually thought that these particulars cannot be given a location in physical space, but the reasons offered involve serious confusions.)

5. The only way I can see in which the dilemma posed by the unsatisfactory character of these two alternatives can in principle be resolved, is by a finer grained theoretical framework in which the physical(2) entities of the coarser framework appear as complex systems of entities of the new framework. Some of the entities of the new framework satisfy "sense-quality" predicates.{3} Let me call them sensa. [Cf. SPR, pp. 102ff., 197ff.] The key features of this new framework are that

(a) Sensa exist only in systems whlch are the fine-grained counterparts of living brains.
(b) Sensa are physical(1)-but-not-physical(2) in that they are conceptually irreducible to non-sensa.[How about that!]
(c) The framework is such that physical(1)-but-not-physical(2) predicates are as essential as physical(2) predicates to the process laws governing the systems which are the counterparts of living brains. (Thus at this level epiphenomenalism is false.)
(d) But the "difference made" by sensa is of such an order of magnitude that at the coarser level a scientific account of living brains in the physical(2) concepts of that level would be an approximation to the truth in roughly the same sense in which macro-thermodynamics approximates to micro-thermodynamics. In other words, epiphenomenalism would be "strictly false" but, at certain levels of scientific investigation, "practically true."

Against this background, let me try to answer your questions. Since you may not have a copy of your letter, I quote the relevant passages:

"Let us suppose for the sake of illustration

(1) Persons are physical(2).

(2) Nation sentences can't be translated into person sentences, but nations are nothing 'over and above' persons."

At this point you comment parenthetically, "Of course, you might not agree that (2) is self-consistent."

Answer: I agree that (2) is self-consistent. See end of paragraph (2) above.

You then ask: "Do (1) and (2) allow nations to be physical(2)? Or are they only physical(1)?"

Answer: They do not allow nations to be physical(2).

You continue: "If (2) by itself prevents nations from being physical(2), then your notion of emergence seems metaphysically innocuous, and I would probably accept it!"

To this I can only say that I hope you will accept it. This does not mean, of course, that I hope that you will also accept the larger framework in terms of which I seek to place emergence, thus understood, in the scientific image. The latter may sound like metaphysics in a rather less innocuous sense. Perhaps you will allow me to hope that some day I may hope this as well -- after you have helped me to sharpen the issues.

Cordially,

/s/ Wilfrid Sellars

WS/esr

Contents
March 14, 1964

Department of Philosophy
Yale University
New Haven, Connecticut

Dear Wilfrid,

Thank you very such for the long and meaty letter--I'm afraid it must have cost you a lot of work. It does look as though your concept of emergence is all right from my point of view--that it does not have any of the scientific unplausibility of the old 'emergent evolution' and vitalist doctrines. Your account of emergence seems to be (as you explain) closely connected with your concept of analogical predication. I would even agree with (b) near the bottom of p. 4 that sensa are conceptually irreducible to non-sensa, but I would wish to say that in a sense there are no sensa--the notion of sensum is an abstraction from 'having a sensum' and 'having a yellow sensum' is reducible. But there are difficulties in my view on account of the fact that 'yellow' (in its primary sense, referring to lemons, not sensa) can't be defined in terms of lemons. A chap might know 'yellow' but not 'lemon'. So even 'having a yellow sensum' is not reducible in the sense of strictly translatable into physicalist terms. I try to get over this by talking rather vaguely in terms of 'the sort of thing we mean is . . .' and about ostensive definition. Your doctrine of analogical predication enables you to avoid this sort of difficulty which comes up for me. (For you don't have to reconcile the manifest image into the scientific one as I do.) So I like your approach, which I think would give the same ontological conclusions as mine. What I feel is somewhat this: I would like to defend my own account if I can, but if I can't I would find yours a very acceptable second line of defense! I think I'll try to have a session at my seminar here on your views on analogical predication and their relevance to the mind-body problem.

With best wishes

Yours,

/s/ Jack

Contents
March 23, 1964

Dear Jack,

Thanks for your note. We are, I think, approaching one another--if only asymptotically.

Perhaps the most useful move I can make at this time is to comment briefly on our respective uses of the term "sensum." They are by no means the same. Yours, as I see it, is bound up with a level of discourse at which we speak either of a person or of his scientific counterpart--a system of physical entities--as

sensing a red triangle

If the term "sensum" is used for what is sensed, then I agree with you completely that "in a sense there are no sensa." Compare my definition of a "second approach" to sense contents (which is the term I use in this connection) on pages 68-71 of SPR, which "second approach" is the first adumbration of the view I subsequently develop. According to this approach, one cannot go from

Jones senses a red triangle

to

(Ex) x is a red triangle.

This theme is picked up on pages 92-95.

On the other hand, the items I called 'sensa' are not introduced in terms of the context

S senses . . .

but rather as scientific objects in a framework designed to sketch a 'micro-micro-theory' of facts of this form. At this finer grained level, the counterpart of

Jones senses a red triangle

is a Space-Time worm (Jones) consisting of particulars some of which are physical(2) particulars of that level, while others are physical(1)-but-not-physical(2) and form an actually existing red triangle in an ontologically primary but methodologically derivative sense of this expression. A red triangle in this ground floor sense is the sort of thing I call a sensum. Sensa, then, are constituents of systems which are the finer grained counterparts of the visual cortex construed as a system of nerves or micro-physical particles. (This, as I see it, is the sort of thing Russell was trying to say, but didn't quite bring off.)

Thus I agree with you that as you use the term, there "really are" no sensa, but only cases of "having a sensum." But I still insist that although at the intermediate scientific level "having a sensum" is a predicate of systems of physical(2) objects, it is neither a physical(2) predicate of this level, nor is it reducible to physical(2) predicates of an adequate finer grained account of sentient organisms.

I hope this letter moves things along a bit.

Cordially,

/s/ Wilfrid Sellars

WS/esr

Contents
Notes

{1} The problem of analysing what it is to 'consider the same thing in different frameworks' is the general problem of which the identity of theoretical and observational constructs is a special case. [Back]

{2} This means that we must be careful in our definition of a physical(2) thing, for if 'senses-a-red-triangle' is a physical(1)-but-not-physical(2) predicate of a group of nerves, thus

Senses-a-red-triangle (n(1), n(2), . . . n(n))

then, of course, we can introduce a derivative physical(1)-but-not-physical(2) predicate which applies to any member of the group,

R(a,b)

to

ø(a)

by using the definition

ø(x)=df R(x,b)

[Back]

{3} These predicates would be still mere derivative in the dimension of concept formation from the sense-quality predicates of the manifest image which apply to physical things. They would, however, have a priority in the dimension of explanation by virtue of which they would describe "what really is." That by methodologically derivative predicates we grasp the ontologically prior, is a modern version of Aristotle's distinction between "priority in the order of knowing" and "priority in the order of being" [Back]

[Contents][Return to Main Page]